This page contains discussions that have been archived from Village pump (technical). Please do not edit the contents of this page. If you wish to revive any of these discussions, either start a new thread or use the talk page associated with that topic.
Updates seem to be at random intervals. The latest was on 22:48, 27 September 2018, i.e. over 6 days ago. Previous updates have appeared at intervals anywhere between 2 days and 12 days.
New redlinked categories appear at a rate of 50–150 per day, which mounts up fast. it would be much easier to keep it cleaned up if updates were at regular and shorter intervals, ideally every 24 hours or 48 hours.
I don't think it's feasible to make these faster. Generating these special pages takes days because on a wiki like English Wikipedia, the queries have to scan billions of database rows to give you the results (e.g. all page links on all pages), and we generate ~30 reports across ~700 projects. I am not a DBA expert, but I expect that the limiting factor here is the speed of the disks from which we read the data. (Also, if we had such faster servers, we'd probably put them to work on things where it matters more to be fast, like page views and edits, rather than generating reports.) Matma Rextalk18:07, 4 October 2018 (UTC)
I agree with Matma Rex, that mostly these reports aren't considered a super important optimization target and that there's probably not a whole lot that can be done which would be deemed worth the effort. Some potential approaches to improve things (These would probably require someone to come forward and work on these things):
run the script in parallel for each slate of wikis. Wikis are divided into 8 slates (s1 through s8.). Generally speaking, the different slates are independent of each other, so probably we could run the script in parallel [At least I think we could, would of course require sign off from the DBAs]. Which wiki is in which slate can be found at https://noc.wikimedia.org/conf/db-eqiad.php.txt . In particular note, enwiki is in its own slate, so I imagine that enwiki could be updated independently of the other 700 wikis
Improve query efficiency for these things. Probably not possible, but nobody really looks at these closely very often, so maybe there's an optimization that could be easily applied
Throw more hardware at the problem - always a possibility, but probably one that won't be done unless someone has a good justificiation why it is needed.
Probably exploring running these independently for different slates would be the best approach in terms of easiness vs results. Bawolff (talk) 16:42, 7 October 2018 (UTC)
FWIW I have a Quarry query that crudely approximates to SWC, which takes 5-10 minutes. Is part of the problem that SWC currently runs off the live database? If so, there's probably no need for it to work that way, it could run off the mirror on the toolserver. In that case, perhaps it would be better off as a WP:DBR - in fact my Quarry query is an adaptation of a Quarry query to emulate Wikipedia:Database reports/Red-linked categories with incoming links which is a good test of how well Quarry is running - it's had long periods when it ran within about 10 minutes, but has had significant spells when it tripped the 30 minute timeout. Maybe @BernsteinBot could run it on a daily basis? At the moment SWC has been running quite well at 3-day intervals, except just in the last couple of weeks (because BHG is back?!) and with a 5000-record cut-off. I think the SWC would be happy to trade a 2000-record cut-off for a daily run, if that helps with server load - ~2000 has been the recent high, it takes a couple of weeks of nobody looking at it to get to that stage.Le Deluge (talk) 14:29, 9 October 2018 (UTC)
For reference, I pasted the precise query Special:WantedCategories uses at https://quarry.wmflabs.org/query/30262 . The Special page runs on the "vslow" replicas (Also used for dumps and some statistics). Which is pretty close to being the live DB but not the same DB as the one that actually serves web requests. The main problem is that basically all the special pages (with some exceptions) for all wikis just get done in alphabetical order. If there ends up being some slowness, it could take a long time to get through the full list. For reference, the last run of Special:WantedCategories took 15m 51.97s on english wikipedia (compared to 3m 13.12s pre-DC switch). I think commons here is a major part of the problem. It appears in the last run, updateSpecialPages.php took 52 hours to just do commons. I think this may be related to the data center switch over (Which among other things, switches which database server these queries run against). Based on logs, when eqiad was primary datacenter, the script only took 3 hours to do commons. I have filed phab:T206592 about this. Bawolff (talk) 03:44, 10 October 2018 (UTC)
A considerable number of special reports should just be ditched. They make no sense whatsoever, and seem to have been largely built either because people were bored or because someone just asked. Consider special:mimesearch vs special:newfiles, the former is 100% unnecessary. Most of its functionality can be accessed through special:newfiles, and all of its functionality can be accessed through the search engine. There is no point to having ancient pages, long pages, and short pages. The former is downright useless, and it can be easily retrieved with the search engine. The latter two could have been combined to a single Special:allpages with sort options. Uncategorized (templates|pages|et al.) could someday be easily obtainable by something like "Special:Search/-incategory:", and for simple pages can be detected using something like "-insource:/\[\[category/". Similarly, some Deadend pages can be detected using "insource".
"Wanted" reports tend to be the worst of the lot, considering that even "wanted categories" contain joke nonsense useless to readers. This is due to an architectural blunder of treating most pages equally, when in fact they aren't. As an example, a lot of sandboxes and discussions deliberately contain lots of broken markup that inevitably ends up in these reports.
It would be instead more sensible to add tools to limit the ability of new users adding new categories. Considering that in many cases people create new categories out of ignorance or laziness of not wanting to search for an adequate existing category. 197.235.166.54 (talk) 13:41, 10 October 2018 (UTC)
Consider that MediaWiki is not built just for English Wikipedia, nor for Wikimedia properties in English, nor for Wikimedia properties, alone. On English Wikipedia these pages may have some utility of lower value. Not so elsewhere. --Izno (talk) 15:28, 10 October 2018 (UTC)
Some kind of edit filter or notice warning editors on talk pages when not signing?
Would it be technically be feasible if some kind of feature, whether mandatory or optional (i.e. opt-in), could be done when an editor forgets to sign an edit of theirs on talk pages and other kinds of discussion page? It sounds like it could be a useful feature and could help prevent unnecessary excessive edits just to add a signature. Narutolovehinata5tccsdnew10:38, 5 October 2018 (UTC)
We have such abuse filter on the Finnish Wikipedia (here). It gives a warning when you try to save edits without a signature on talk pages or in Wikipedia namespace. Stryn (talk) 11:31, 5 October 2018 (UTC)
Thanks for pointing to this filter. Do you know how well it handles copy edits to already signed text? isaacl (talk) 18:36, 8 October 2018 (UTC)
A short reminder: to make sure we can handle an event where our primary data centre is unavailable – or destroyed – we're doing a server test. This means that at some point after 14:00 UTC (it's only for part of the process we need to lock the database for editing) it won't be possible to edit the wikis today, 10 October. It could last up to an hour, but will probably be much shorter than that. /Johan (WMF) (talk) 10:53, 10 October 2018 (UTC)
It's a new record: just four minutes and 41 seconds. The "switch back" is usually faster than the "switch to", but this is really close to the theoretical minimum.
Also, they found a few problems (worst one: the two sites did not have identical copies of Wikidata), so in addition to learning how to do this process even Faster, Higher, Stronger than before, they're going to be able to prevent some problems that might otherwise have occurred if there had been a real-world disaster.
Thank you all for your patience, and extra thanks to those of you who helped spread the word and answer other editors' questions about it.
I have read nothing about when (or even if) this will be done again in the future. However, they've done it approximately once a year for the last three years, so I'd expect it to happen next year, too. Whatamidoing (WMF) (talk) 19:30, 11 October 2018 (UTC)
CAT:CSD transcludes ((Adminbacklog)) with a 50-item threshold, i.e. it automatically detects the number of contents and displays as a backlog only if there are 50+ items. Can anyone imagine why it's displaying now? There are only twelve pages and sixteen files in the category; there's no way this should be showing up. I've purged it repeatedly (first time was over an hour ago; second was maybe half an hour or more ago) using the (recount) link in the template, so the item counts shouldn't be old information. Nyttend (talk) 04:22, 12 October 2018 (UTC)
This is a screencap of bicycle brake. The thumb image has a vertical line of black and gray dots. It doesn't look this way on the actual commons:File:Bicycle brakes - animated.gif image page. Seems like it's at the boundary of the transparent background? I tried changing the |upright=1.35 parameter, and any other value does not have this artifact. I tried various purging of the file page and the enwiki article page, with no change. Another user says they do not see it. Any ideas what's happening? DMacks (talk) 12:30, 11 October 2018 (UTC)
220px in that preference for me also. The direct link from right-click is [2], where I also see the dots and it does animate (in both cases). I'm using Firefox 62.0.3 on OS X. DMacks (talk) 13:20, 11 October 2018 (UTC)
Okay, now I don't (neither direct-view of that URL nor in the aricle). Glad I kept a screenshot so neither y'all nor I think I'm crazy. DMacks (talk) 13:21, 11 October 2018 (UTC)
Deleted log events will still appear in the logs, but parts of their content will be inaccessible to the public. Other administrators will still be able to access the hidden content and to undelete it, unless additional restrictions are set.
When changing the visibility of a log entry, there are three items that may be redacted individually or together: delete action and target; delete edit summary; delete performer's username/IP address. In theory, I could redact a block log entry on behalf of somebody else, this would then create a further entry in the deletion log, showing that I had changed the visibility of another log entry; I could then redact my own username on that new entry, which would then create a further entry ...
There's always a trail for those who need to know, and those who abuse the tools can be traced.
PDF file cited as source downloading onto computer
I'm wondering why clicking on the url for the source cited in Business network#cite_note-:0-2 is downloading onto my computer instead of opening in a web browser. I've clicked on PDF files cited as sources before and they usually open up in the browser. If it's my computer then no biggie; if, however, it's the formatting of the citation or something about the website, then maybe the citation syntax can be tweaked to stop the automatic download. There's another PDF file cited in the same article as Business network#cite_note-SS_1995-6 which does open in the web browser, and doesn't download. -- Marchjuly (talk) 06:44, 5 October 2018 (UTC)
OK. It seemed a bit weird that it was doing it for one and not the other, but I'll sort it out since it's on my end. Thanks for taking a look. -- Marchjuly (talk) 07:19, 5 October 2018 (UTC)
I followed the instructions in the link you provided, but my settings were set to not download PDFs. Kinda weird. Can a website set things up to override a user's individual settings? -- Marchjuly (talk) 07:24, 5 October 2018 (UTC)
@Marchjuly: a quick look at that site indicates it is not really serving "a file" but a data stream of the file - and browsers wouldn't normally be configured to deal with that type of data in-browser. — xaosfluxTalk20:23, 5 October 2018 (UTC)
@Xaosflux: Thanks for taking a look. Ah...that’s sounds a bit diabolical, but I expect it’s fairly common. The question then is whether such a thing is OK for Wikipedia’s purposes; it seems like it would be better to cite a site which actually hosts the desired content than downloads it to your computer when clicked. Is there a way to tweak the citation template syntax which tells whatever/whoever needs to know that clicking on the link will download it onto your computer? — Marchjuly (talk) 21:38, 5 October 2018 (UTC)
I would think that if you can point to one page "up" where a link to the PDF is obvious, generally in such cases where that page is describing what the user will see in the document, then that's an ideal solution. But the "stream" link is not bad if that's only way to provide the PDF. --Masem (t) 21:44, 5 October 2018 (UTC)
@Masem: It took going back a couple of pages to ec.europa.eu/docsroom/documents/5563/attachments/1/translations to find a link which wouldn't download or which wasn't listed as an error, but I'm not sure that helps anything since it still seems to require that the source be downloaded if you want to verify it. Is there a way to mark the citation, so that readers are alerted to the fact that clicking on the link may cause the file to be downloaded to their computers? -- Marchjuly (talk) 07:39, 12 October 2018 (UTC)
That's really annoying. I tried to find a better "landing" page (something aking to [3] where there's information about the report and a clear link to download). Best I can tell there's no way to stop the download without severe CSS/JS intervention on the client side. (you can force a PDF to download through simple HTML5, but not vice versa). And best as I can tell from the cite templates there's no way to throw an indicator to a user. I definitely don't want someone to insert a ref thta ends up downloading a 10 gig document without warning, for example. --Masem (t) 14:17, 12 October 2018 (UTC)
Yes, as other examples consider these three pages of the same document: sheet 1 of 3; sheet 2 of 3; sheet 3 of 3. Those links take you to the file description pages, but if you click the image or the "Original file" link, it downloads instead of displaying. The smallest of these three is 1.28 MB. --Redrose64 🌹 (talk) 15:11, 12 October 2018 (UTC)
@Masem, Xaosflux, and Redrose64: Appreciate all of the input and info. FWIW, the citation isn't formatted using a template (sorry for not noticing this before); so, I think a "warning" about the link can just be added manually. However, perhaps the best thing to do in a case such as this would be to reformat the citation in a way that deactivates the url; for example, ((Cite web)) requires an active url to work properly, but there are some citation templates which don't. So, maybe treat this as a type of WP:SAYWHERE source, providing as much information about it as possible. If a citation template is used, then using |via= to treat the landing page as WP:CONVENIENCE and |others= or |quote= to either (1) quote the specific information being referenced, (2) add a "warning" to the reader that the file will download on to their computer if clicked, or (3) some combination of those two things. Maybe the person who added the citation or someone else who has been contributing to the article would be willing to do that, so I'll add ((Please see)) for this discussion to the article's talk page. -- Marchjuly (talk) 23:00, 12 October 2018 (UTC)
Do make sure that |format=PDF is used. They know that clicking the link in that case will bring them a file - whether that is viewed in the browser or being downloaded to view outside the browser, its still being downloaded. --Masem (t) 23:11, 12 October 2018 (UTC)
Thanks for the reminder. That's a good point since there'll be no PDF icon displayed after the url if it's deactivated. -- Marchjuly (talk) 23:28, 12 October 2018 (UTC)
Except for the cases of a link to another point in the same page, and a "link" that actually fires some JavaScript, clicking any link - PDF or otherwise, external link or internal - will bring back at least one file. For example, clicking this link - User:Masem - will bring back the file https://en.wikipedia.org/wiki/User:Masem plus several others (stylesheets, images, JavaScript etc.)
Line break leaves lone quotation mark at end of line
Can we do something to make edits like this one unnecessary? It seems likely that we would in general want to suppress a line break after an opening quotation mark; is there some reason that's not easy? Or is this more of a browser bug? Dicklyon (talk) 06:05, 13 October 2018 (UTC)
Line breaks without nowrap are decided by the browser. My Opera and Chrome versions will break between "( but not Firefox, IE and Edge. In theory our software could be changed to automatically add nowrap in certain places where breaks are probably unwanted but some browsers are known to make them anyway. I don't support actually bloating our html with extra code like that. Example without nowrap if people want to test their browser by changing window width:
dont use nowraps around that. That’s insane. The world wide web is no typesetting software. It is supposed to look different for different people. It’s supposed to have browsers make decisions for you so that overtime it can improve doing so. By adding these nowraps you reduce the legibility of the wikicode, for a minor rendering issue that will confuse absolutey no-one. It also messes with the wordbreaking algorithm of browsers (for instance when i have a stylesheet to hyphenate words before linebreaking, then this nowrap would mess with that). If the parent element is narrower than the word, then it will either overflow or be forced down the page to where there is space (especially problematic on mobile. Nowrap is evil and should be avoided to where it is absolutely necessary. The templates description doesn't say ‘use sparingly’ for nothing. —TheDJ (talk • contribs) 10:28, 13 October 2018 (UTC)
Changing ((nowrap|"(musician)")) to ((nowrap|"())musician)" would avoid interference with browser hyphenation of words but further reduce the legibility of the wikicode and still bloat the html so I don't support it either. If there are browsers breaking between )" then should we also write ((nowrap|"())musician((nowrap|)"))? No thanks. PrimeHunter (talk) 11:12, 13 October 2018 (UTC)
Wanted Templates not updating
In Malayalam wikipedia, in the Special page Wanted Templates, what could be the reason for a certain range of templates to appear again and again even after purging happens a number of times?Adithyak1997 (talk) 05:27, 13 October 2018 (UTC)
Please give a link and example when you report an issue. ml:Special:WantedTemplates is generated periodically and says in English: "The following data is cached, and was last updated 09:46, 10 October 2018." You didn't say what or when you purged, and some things require a null edit and not a purge to update link tables. Nothing done after 09:46, 10 October 2018 will affect the list until next time the list is generated by the software. PrimeHunter (talk) 09:43, 13 October 2018 (UTC)
I don't know Malayalam but there appears to be an issue with automatic conversion between two similar characters. The third link at ml:Special:WantedTemplates says ml:ഫലകം:തടയപ്പെട്ടവര് where "ഫലകം" means "Template". The link is red in the list but works. However, it automatically changes the last character and goes to the page ml:ഫലകം:തടയപ്പെട്ടവർ from 2007. The last characters ര് and ർ look different to me, and my browsers find function says they are different. They also produce url's with different ends but both url's automatically redirect to the same third url. Maybe the software generating ml:Special:WantedTemplates doesn't know about the automatic conversion and produces an entry if a non-canonical form of the name is used in a page. According to https://r12a.github.io/uniview/?charlist=%E0%B4%B0%E0%B5%8D%20%E0%B5%BC, ര് is "0D30 MALAYALAM LETTER RA" followed by "0D4D MALAYALAM SIGN VIRAMA", while ർ is "0D7C MALAYALAM LETTER CHILLU RR" and not followed by something else. PrimeHunter (talk) 10:58, 13 October 2018 (UTC)
Hi, am I the only one getting the following error on G-test?
Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "/mathoid/local/v1/":): {\textstyle \hat{\theta))
Please try to insert the wikitext of Special:Diff/863443818/863448552 to any page (preferably Wikipedia:Sandbox or an userpage sandbox). This should fail when all the following conditions are true:
Despite all this, it works for @PrimeHunter (who's an administrator) at Wikipedia:Articles for deletion/Anti-Jewish violence in Eastern Galicia involving soldiers of the Blue Army: Special:Diff/863619579?!
The most mysterious part about this is that it's not Special:AbuseFilter related, as it seems. There are no logged events at Special:AbuseLog. There are no special editnotices. Attempting to publish the changes repeatedly does nothing. I didn't get a CAPTCHA for the external link.
While trying to write this VP/T post, I got the red "Incorrect or missing CAPTCHA" box at top but CAPTCHA security check at bottom. The issue started around ~22:30 UTC; was there a MediaWiki update behind this? (I've never seen the CAPTCHA box be at the bottom before.)
Please, I'm very curious for someone else to attempt to reproduce this as an unregistered contributor. For the curious, this is on Firefox 62.0.2 on GNU/Linux. I feel like to have encountered the weirdest bug of my years here. 84.250.17.211 (talk) 23:58, 11 October 2018 (UTC)
I have so far reproduced the missing CAPTCHA issue by reducing the issue to these two lines:
<div class="infobox" style="width:33%">AfDs for this article:<ul class="listify">((Special:Prefixindex/Wikipedia:Articles for deletion/Anti-Jewish violence in Eastern Galicia involving soldiers of the Blue Army))</ul></div>
:(((Find sources AFD|Anti-Jewish violence in Eastern Galicia involving soldiers of the Blue Army)))
In combination, the CAPTCHA goes missing. Perhaps I should blame the nominator here? But I'm still curious. (I got the incorrect or missing CAPTCHA while writing this, with CAPTCHA below the edit summary again and red box at top.) 84.250.17.211 (talk) 00:32, 12 October 2018 (UTC)
This is the smallest test case I could make for CAPTCHAs to go missing:
To be clear: This issue is experienced with Template:Afd2 (in example), with a caveat that one must be signed in to nominate an article for deletion. I must be one weird person to handle AfDs and cleanup while refusing to have/be signed into an account. 😏 84.250.17.211 (talk) 00:58, 12 October 2018 (UTC)
I can reproduce this. It happens if you try to save an edit when these three are all satisfied:
You are unregistered or not autoconfirmed at the wiki. It also happens at other wikis.
The edit transcludes a special page, either directly in the source or via a template like ((Afd2)). It can both be an existing special page like ((Special:Prefixindex)) and a non-existing page like ((Special:NoSuchPage))
The edit adds an external link which would normally have caused a CAPTCHA.
The expected behaviour is getting an edit page with a CAPTCHA. The actual behaviour is getting an edit page with no CAPTCHA. There is no apparent way to complete the edit. If users want to experiment without logging out then try a wiki where you are not autoconfirmed. The default requirement in Wikimedia wikis is four days and zero edits but some wikis require edits with wgAutoConfirmCount in https://noc.wikimedia.org/conf/highlight.php?file=InitialiseSettings.php, e.g. 20 edits at no:. Special:CentralAuth can show where your account is already registered. PrimeHunter (talk) 10:05, 12 October 2018 (UTC)
It looks like you found a bug in the CAPTCHA code, please report the bug so it can get looked in to. This is not limited to the English Wikipedia, so it will need to be looked at by developers. — xaosfluxTalk22:44, 12 October 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Problems
Some pages, edits and users disappeared for a short while after the server switch. Missing content and users was fixed within a day. Some preferences and other things might take a few more days to fix. [4]
Wikis are updated with new and updated translations from translatewiki.net again. This will happen once a week. The developers are working on fixing the problem so we can have translation updates more often again. [5]
Changes later this week
When you create an abuse filter that prevents edits you can now write a specific error message for it. Before this all abuse filters that prevented edits had the same error message. [6]
The new version of MediaWiki will be on test wikis and MediaWiki.org from 16 October. It will be on non-Wikipedia wikis and some Wikipedias from 17 October. It will be on all wikis from 18 October (calendar).
Meetings
There will be no more meetings with the Editing team. This is because not enough Wikimedians were interested. To tell developers which bugs you think are the most important you can use Phabricator as normal. [7]
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 17 October at 15:00 (UTC). See how to join.
Hi. I'm wondering why do I keep getting "&veaction=editsource" in the URL address when I try to edit a page. I'm guessing this is something that has to do with Visual Editor/wysiwyg (which is something I'm not interested in using, at the moment) and I made sure to de-select the option "Temporarily disable the visual editor while it is in beta" in my Editing Preference. Any help or guidance here will be greatly appreciated. Thank you. --Ciphers (talk) 04:04, 16 October 2018 (UTC)
Pedantic detail: The 2017 wikitext editor (aka "new wikitext mode" or "VisualEditor's wikitext mode) is part of VisualEditor (the extension). It is not, however, the visual editor (aka "VisualEditor's visual mode"), which is the thing (currently, the only thing) you use for editing that looks like a word processor.
(said with love:) A refund, as in you want us to take back the money that the WMF paid you to attend those meetings? We volunteers will have to have a meeting about that. – Jonesey95 (talk) 04:51, 17 October 2018 (UTC)
Question about a block
I helped a reader create an account. They wanted to edit but they were unable to edit as an unregistered editor because of a range block. they now have an account, but even after logging him they get a message that they are not permitted to edit because of a block.
I must be missing something — I thought if one had a registered account, the good edit while logged in even if their IP happen to be part of a range block.
If anyone reading this happens to be an OTRS agent, they provided a screenshot of the message.
What I did (or at least think I did!) was to locally disable the global block and then install an anon-only en block. When this problem arose with schools, the advice that we gave that seemed to work, was to make a few edits on a clear IP, to establish the account, before going back to the blocked IP. I don't have the technical knowledge to speculate why this worked but it is worth a try. Just Chilling (talk) 00:43, 16 October 2018 (UTC)
I had a similar thing happen a few weeks ago. I was using a VPN (proXPN, Toronto server IIRC) and was logged in, but got a message that the IP address was blocked when I tried to edit a page. (I could try to find the IP address of that VPN server if needed, but it has been dropping my connection frequently in the time since then, so I haven't been using it.) Just Chilling, the advice to make a few edits from elsewhere doesn't seem like it would have been helpful for me—I've been editing with this account once in a while for many years. – PointyOintment ❬‽ · ✍❭20:37, 17 October 2018 (UTC)
@Doc Taxon: there is a lot of oddity with precedence of 1px borders, a hack is to use style 'double', as a double border has higher precedence, but the same appearance like this:
@Enterprisey: I get "Use of "mw.toolbar" is deprecated" and "title=User:Apoc2400/refToolbarPlus.js&action=raw&ctype=text/javascript:1 You installed the userscript User:Apoc2400/refToolbarPlus.js
Hi, Over the past hour or 2 I've been repeatedly logged out and ticking the "keep me logged in for 365 days" doesn't seem to achieve anything,
Never had the problem before and as far as I know everything my end is sound so didn't know if there was some sort of bug here,
I'm about to go to bed so if anyone replies I'll respond in the next 8-9 hours,
Many thanks, –Davey2010Talk03:33, 18 October 2018 (UTC)
I was having this issue briefly. It's usually related to a corrupted cookie. My issues cleared up after 5-10 minutes. Let us know when you're back from the dream lands if it is still happening. --Izno (talk) 03:52, 18 October 2018 (UTC)
Hi Izno, Oh okay I had no idea corrupted cookies could cause it - Learn something new everyday! :), All seems fine now so maybe it was a corrupted cookie, Many thanks for your help, Thanks, –Davey2010Talk12:42, 18 October 2018 (UTC)
@Davey2010: If you experience problems with your login, one thing to try is to deliberately log out. This action might not destroy a corrupt login cookie, but it will invalidate all login cookies bearing your login ID, on all machines, no matter how old (or how recent). Then when you next log in, you get served a fresh clean cookie, and any corrupt cookies that you may still have will not be recognised should they be sent back to the Wikimedia servers. --Redrose64 🌹 (talk) 22:45, 18 October 2018 (UTC)
Redrose64, no idea, but the actual numbers show a lot of 0 in the data. It seems that this graph uses quite a bit of easing and i suspect that those two combined are the cause of this. Not sure if that an error in the module or the graph lib.—TheDJ (talk • contribs) 20:20, 17 October 2018 (UTC)
Yeah, the problems might be from the fact that the article was created very recently (October 10) and the graph shows days before that. SemiHypercube✎20:39, 17 October 2018 (UTC)
@Redrose64: When I need to leave multiple notices like this, I find that leaving bare-bones messages at the "satellite" talk pages pointing to the "central" discussion location works better than leaving a full description of the issue at every location. (IOW, you probably should have just said something like: "Please see mw:Template talk:Graph:PageViews#Reverse curve where a strange problem I have found in the output of that template is being discussed.") - dcljr (talk) 00:14, 19 October 2018 (UTC)
File:Allauddinum Albhutha Vilakkum.jpg was uploaded to the English Wikipedia as a non-free image for use in a specific article with a fair use claim. Files can only be used if they are uploaded to the same wiki or to Wikimedia Commons. Commons does not accept non-free images so you would have to upload a copy to the Malayalam Wikipedia, assuming non-free images are allowed for the purpose by the wiki. Wikis have different rules and I don't know anything about Malayalam. PrimeHunter (talk) 11:08, 19 October 2018 (UTC)
The English Wikipedia has no rights which can be violated here. It is the rights of the copyright holders which must be respected. I don't know which procedures the Malayalam Wikipedia has for this but it may involve ml:Template:ScreenshotU. PrimeHunter (talk) 11:42, 19 October 2018 (UTC)
When I edit a template documentation page to add or change WP:TEMPLATEDATA, the edits do not materialise. I use the "Manage TemplateData" edit helper (button). After editing in theat screen, I press "Apply" then no edits are made to the /doc page edit box (and so saving would be a null-edit). Any ideas? -DePiep (talk) 16:38, 7 October 2018 (UTC)
Extra test: installed Firefox v62 on Windows10 desktop. Same process. Using safemode=1: success, without safemode=1: fail. -DePiep (talk) 16:01, 10 October 2018 (UTC)
Running with debug=true, webconsole message in Chrome: "Uncaught TypeError: Cannot read property 'accessKey' jquery.js[...]" (in Firefox: "jQuery.Deferred exception: element is undefined [...]"). Need more? -DePiep (talk) 16:17, 10 October 2018 (UTC)
I can confirm this: the error message appears in the debug mode after ((Currency/doc)) is opened for editing. It only appears after Manage TemplateData button appears on the screen. Also tried to edit the template data but without success. Ruslik_Zero09:29, 11 October 2018 (UTC)
I can't reproduce this (on my Mac). I'll file a bug report.
Since ?safemode=1 seems to resolve the problem, the problem is probably in a user script or gadget. This could be anything from WikEd to something that only the two of you are running, but it's probably not something that I'm running (which is pretty much the defaults plus NAVPOPS, and whatever you can see in my common.js/global.js files). Do you have any guesses about what scripts youtwo might have in common? Whatamidoing (WMF) (talk) 19:50, 11 October 2018 (UTC)
My scripts afaik: 'User:Js/ajaxPreview.js'; 'MediaWiki:Gadget-HotCat.js'; 'User:Cameltrader/Advisor.js'; WikEd, JS WikiBrowser = Wikipedia:AutoWikiBrowser/Script, also WP:AWB. Don't know if this is the complete list.
@DePiep: Can you get and copy-paste a backtrace from the browser console for this error (either browser)? There should be a little arrow/triangle next to the error message, clicking it will display the backtrace below. It's impossible to identify what is causing the error from just this message (I am guessing that the error is generated within the 'jquery.accessKeyLabel' module, but the real problem is probably in whatever code is using jquery.accessKeyLabel here). Matma Rextalk18:34, 12 October 2018 (UTC)
@DePiep: The stack trace could allow someone to deduce which browser extensions you are using or which gadgets you have enabled (in this case, the latter is kind of the point). It should not expose any other private information (although it technically could, it would take very weird code to do it). If you're concerned, you can email it to my WMF account bdziewonski@wikimedia.org instead of posting it publicly, or review it yourself (I'm looking for a function name or a file name that would reveal which script is causing your problem). Matma Rextalk01:55, 13 October 2018 (UTC)
I looked into this: the issue is caused by the wikEd gadget. If anyone still maintains wikEd, I documented some details of the problem and ideas on how to fix it in a comment at phab:T206795#4668857. I guess for now you'll need to turn it off before using the TemplateData editor. Matma Rextalk23:32, 15 October 2018 (UTC)
This is not related to this issue. The issue we're talking about in this section is purely about how the watchlist is displayed, not what's on it. --Roan Kattouw (WMF) (talk) 19:26, 19 October 2018 (UTC)
@Wumbolo: When you view Special:Watchlist, the page names that are displayed are the current page names (if you want the technical reason, just ask). This means that edits made prior to a page move are listed, but not under the name that they had at the time. If you go to Special:EditWatchlist, you should find that both page names are listed. --Redrose64 🌹 (talk) 22:04, 19 October 2018 (UTC)
The diff and hist links say curid=58713610 and title=%22Grievance_Studies%22_affair. The links go to the diff and history at Grievance Studies affair. Page information says 58713610 is the page ID of Grievance Studies affair[11] while the redirect has page ID 58824385 [12]. curid overrides title on the hist link so it goes to the history of Grievance Studies affair and not the redirect. If curid is manually removed [13] then the title is used and it shows the history of the redirect. Like Wumbolo, I have verified that Grievance Studies affair is not in Special:EditWatchlist. Only the redirect is there. The error happens both with and without "Hide the improved version of the Watchlist" at Special:Preferences#mw-prefsection-watchlist. I tested other recent page moves and they don't have similar errors when you only watch the old title. PrimeHunter (talk) 22:23, 19 October 2018 (UTC)
United States presidential election, 1896 and Draft:Jean Morrison are examples of other recent moves where the page has recent edits which show up when you only watch the redirect. The watchlist only shows edits made before the move. If you swap the watchlist entries to only watch the new title and not the redirect then edits made before the move disappear from the watchlist when they should have been shown, while edits made after the move are shown as expected. PrimeHunter (talk) 22:44, 19 October 2018 (UTC)
The TardisIndexFile template affixed in the "External links" section of every article on a Doctor Who episode is faulty and/or outdated. These links all lead to empty error pages. I would fix the template myself if I knew exactly which part of the code was the culprit.Bjones (talk) 22:31, 19 October 2018 (UTC)
@Bjones: looks like this just provides an external link to wikia, such as [14] - this should be able to be adjusted, but we would need to know where to point these links to on wikia. Do you know what an example wikia link for this would be? — xaosfluxTalk22:38, 19 October 2018 (UTC)
This is a problem specific to mobile and specific to Wikia's redirection on their site. It's not our fault. There's a thread somewhere recently on this page or its archives of an IP having an issue. Please contact the people at Wikia support. --Izno (talk) 22:54, 19 October 2018 (UTC)
I've updated the interwiki map at meta for 'wikia', this may take some time to propagate. Please ping me on meta if you need it further adjusted. — xaosfluxTalk14:53, 20 October 2018 (UTC)
@Xaosflux: It appears all Wikia links are fixed by inserting w: before c:, e.g. http://www.wikia.com/w:c:tardis:Davros. Three other Wikia prefixes need similar changes as requested at meta:Talk:Interwiki map#Wikia. I have done this in ((Wikia)) for now [15] since it sometimes takes months before updates to the interwiki map are copied to the software. Template:TardisIndexFile and other templates and pages using ((Wikia)) should now produce working links. Pages using interwiki prefixes to Wikia without calling ((Wikia)) are still broken until the software is updated. ((Wikia)) will still work after that since the fix only inserts w: if it isn't already there. PrimeHunter (talk) 14:13, 21 October 2018 (UTC)
Good to hear, we usually just let those go when they get around to them, but it can be requested to do-now, looks like a dev was available today. — xaosfluxTalk19:09, 21 October 2018 (UTC)
Greetings, WP 1.0 bot is having on-going issue for several months. It starts daily processing, then stalls out at various WikiProjects, not processing the rest. Lately it does not get past "B" in the alphabet. The issue reports are at Wikipedia talk:Version 1.0 Editorial Team/Index.
A second issue is bot not creating WP logs for many WP since October 8. For example here
and here.
Thanks AfroThundr - When I click on link above it gives a 404 error. Then I'm unable to signin, and it does not like my username or email address or pwd. :-( JoeHebda (talk) 02:00, 22 October 2018 (UTC)
Apparently Github is having database problems right now, so the issue I posted won't show up at the moment. Here's hoping they get this sorted out soon. — AfroThundr (u · t · c)03:39, 22 October 2018 (UTC)
3 questions about API usage and Pywikibot
I tried dusting off Muninnbot to fix an old bug and I see that it generates some deprecation warnings when querying revisions
snippet of command-line session from running Munninbot's source code
(...)
WARNING: API warning (revisions): Because "rvslots" was not specified, a legacy format has been used for the output. This format is deprecated, and in the future the new format will always be used.
WARNING:pywiki:API warning (revisions): Because "rvslots" was not specified, a legacy format has been used for the output. This format is deprecated, and in the future the new format will always be used.
WARNING: API warning (main): Subscribe to the mediawiki-api-announce mailing list at <https://lists.wikimedia.org/mailman/listinfo/mediawiki-api-announce> for notice of API deprecations and breaking changes. Use Special:ApiFeatureUsage to see usage of deprecated features by your application.
WARNING:pywiki:API warning (main): Subscribe to the mediawiki-api-announce mailing list at <https://lists.wikimedia.org/mailman/listinfo/mediawiki-api-announce> for notice of API deprecations and breaking changes. Use Special:ApiFeatureUsage to see usage of deprecated features by your application.
(...)
Hence three questions:
My understanding of mw:API:Revisions and mw:Manual:Slot is that I just have to tack rvslots=main in the API query and everything goes as before, at least for now. Is that correct?
How does Special:ApiFeatureUsage work? I just see an empty list if I fill in the bot's name (whether logged in or not from my account or my bot's account).
Whether or not that is what needs filling for the above, how does one go about seeing the user agent that Pywikibot sets? I have read mw:Manual:Pywikibot/User-agent but there is no user_agent_format in my config file and the default includes a git revision number apparently. I searched a few places but did not find anything (for instance, I do not see a method or attribute in the pywikibot.data.api.Request class).
Thanks for the pointer, but I am not sure I should be spamming the main mailing list for help/support (none of the above is even remotely a bug report or feature request). Furthermore, here, most of the questions are not PWB-related. TigraanClick here to contact me12:18, 22 October 2018 (UTC)
There's been one thread this entire month on the main mailing list. I doubt anyone would see it as spamming. Besides that, there is also the IRC channel. My feeling is that your question is too specialist for this forum, but someone did reply, so... --Izno (talk) 13:57, 22 October 2018 (UTC)
I only use PWB for login and user-agent, ((nobots)) and sleep time compliance. I format the API calls myself so I do not blame PWB for that one. The change to fix (1) is trivial to implement but I want to make sure I understood what is going on. (I am using Toolforge's installation, which I would expect to be up-to-date.) TigraanClick here to contact me12:18, 22 October 2018 (UTC)
Technical issue with characters on en.quote
I realize this isn't the right place, but I don't know where else to find a lot of people smart than I am. (See here). There appears to be some issue with adding certain characters on certain accounts that for some reason gets solved when the person registers an alt account. Suggestions? Is this a phab issue? GMGtalk15:04, 21 October 2018 (UTC)
I've noticed today that my watchlist is no longer differentiating between changes I've seen and changes I haven't. Each entry now just starts with the same period/full stop. Using Firefox on Mac, and no relevant preference options that I can see. Tried Chrome on Mac too, and the differentiation between seen and unseen changes also does not show. Anyone else seen this, and any idea what the problem is? Boing! said Zebedee (talk) 12:54, 19 October 2018 (UTC)
Same here, using Firefox 62.0.3 under Windows 7, 32-bit. The bullet points are now much smaller (are the size of periods) and undifferentiated. Dhtwiki (talk) 13:09, 19 October 2018 (UTC)
Geonotice: display notices on your watchlist about events in your region
Display watchlist notices
(This loads the base style for the watchlist. Please do not disable this option.)
The other three are disabled. If I enable the "Display pages on your watchlist that have changed since your last visit in bold" it marks the new changes in bold so I can at least see what's new, but it doesn't effect the bullet points - instead of solid and hollow? bullet points, they're still just periods. Boing! said Zebedee (talk) 14:19, 19 October 2018 (UTC)
I'm having this problem as well. Everything looks normal (bold and unbold) except that each line begins with a period. Natureium (talk) 14:31, 19 October 2018 (UTC)
Gah this is horrible. Makes maintaining Wikipedia very difficult. "Hide the improved version of the Watchlist" under Preferences -> Watchlist returns two colors "green" for "not reviewed" and "blue" for "review" but than you loss some of the other improvements such as "filters". Doc James (talk · contribs · email) 16:41, 19 October 2018 (UTC)
I noticed this yesterday. I also have noticed that since yesterday morning US Eastern Time, no new changes have shown up. I presume this is because work is being done? If so it would be nice to have some sort of bannered notice about what's going on, what progress is being made and when we can expect normal service to resume. Daniel Case (talk) 17:52, 19 October 2018 (UTC)
I've put in some temporary CSS that should fix this issue on English Wikipedia for now. I will now work on a proper fix, because we can only fix one wiki at a time this way, and the issue affects all ~900 WMF wikis. --Roan Kattouw (WMF) (talk) 18:12, 19 October 2018 (UTC)
I think you can take the notice down anyway. The symptoms appear to be limited to editors who are running a volunteer-maintained gadget written some years ago by User:Edokter. It is not widely used except among long-time editors at this wiki, and it is solved by not using the gadget. Whatamidoing (WMF) (talk) 18:19, 19 October 2018 (UTC)
It's a series of gadgets, grouped under the "Watchlist" heading at Special:Preferences#mw-prefsection-gadgets. If you are still having problems, even after reloading your watchlist, then you should be able to resolve them by ticking the box in that section that says "Display pages on your watchlist that have changed since your last visit in bold".
I believe there had been some talk a while ago about resetting the gadgets system so that new accounts would just have normal MediaWiki behavior, rather than using this system. The original purpose of the gadget you need to enable was to disable the gadget that disables normal MediaWiki software functions, which is unnecessarily confusing. Whatamidoing (WMF) (talk) 18:39, 19 October 2018 (UTC)
With all of the "Watchlist" options in the Gadgets prefs disabled, I still see the problem with both seen and unseen changes showing as periods with no distinction. Do you have a "non-gadget" set of prefs to make it work properly? Are you saying the bullet point changes version should not be used and we should rely on the bold/non-bold version permanently? Boing! said Zebedee (talk) 18:42, 19 October 2018 (UTC)
Oh, and the "Display pages on your watchlist that have changed since your last visit in bold" pref is in the gadgets section, so is that the gadget that you're suggesting we should not use or not? I'm really getting quite confused by what you are saying here. Boing! said Zebedee (talk) 18:44, 19 October 2018 (UTC)
I put my workaround in the "watchlist base" gadget, assuming that nobody would have it disabled (because it's enabled by default, and its description tells you not to disable it), but aparently 482 users have disabled it anyway. I'll move my workaround to MediaWiki:Common.css so it actually applies to everyone. --Roan Kattouw (WMF) (talk) 19:29, 19 October 2018 (UTC)
We have deployed a permanent fix for the issue (and therefore removed the temporary CSS fix). After refreshing your browser, the watchlist should be behaving as expected and back to normal now. Please let me know if any of you are still experiencing the original issue, and thank you for your patience as we worked on this. There is only one outstanding issue: when using highlighting, all highlighted rows have filled-in circles, whether they are seen or unseen. The fix for that is written and will be deployed on Monday. -- MMiller (WMF) (talk) 20:57, 19 October 2018 (UTC)
One final update: that last fix I mentioned above is now deployed. When using highlighting, highlighted rows again have empty and filled-in circles that agree with whether they are seen or unseen. Thank you! -- MMiller (WMF) (talk) 18:44, 22 October 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 23 October. It will be on non-Wikipedia wikis and some Wikipedias from 24 October. It will be on all wikis from 25 October (calendar).
Meetings
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 24 October at 15:00 (UTC). See how to join.
Future changes
The Community Wishlist Survey begins on 29 October. The survey decides what the Community Tech team will work on. You can post proposals from 29 October to 11 November. You can vote on proposals from 16 November to 30 November.
AVG can occasionally be over-sensitive. Wait a few days (or more) to see if AVG pushes an update before attempting to use it again. The Javascript on meta seems to be just fine, which is good. Have you tried pinging the creator? I'd start with that. --Izno (talk) 00:16, 23 October 2018 (UTC)
I have transferred part of a page to my sandbox at User:Paine Ellsworth/sandbox. Will someone explain why only the first section has an edit link? All the other sections are listed in the TOC; however, none of them have edit links. This began just a few days ago on my workpage, and I don't know what is broken or how to fix it. Paine Ellsworthput'r there19:37, 23 October 2018 (UTC)
How can I permanently hide the page curation rightside toolbar
When I open new pages from the Special:NewPagesFeed, I now get on those pages on the right side a (to me) annoying large page curation toolbar overlapping the text. With the uppermost button I can minimize it, and then with "X" I can hide it completely, but I have no interest in doing this on every page again and again. I have looked in preferences, and I can't find a way to get rid of this new thing I don't need or want (I have an unobtrusive Twinkle dropdown at the top of the page for the thngs I need, and I don't need things like "go to next page in the queue"). Fram (talk) 09:20, 19 October 2018 (UTC)
@Fram: once you close it it should stay closed unless you click on 'curate this article' on the side pane. If you want to hide it, you can put the following in Special:Mypage/common.css
/* Hide the page triage toolbar */#mwe-pt-toolbar{display:none!important;}
It stays closed on that article, but it reopens on every new article I open from the new pages. Thanks for the css code but I don't use or want special css, these kind of changes shouldn't be made in general without an opt-out function as they are rather intrusive and not wanted by everyone. I can't find this change announced here at all, but it's not always easy to find these things. Fram (talk) 12:13, 19 October 2018 (UTC)
Hello @Fram:, can you please check if ?showcurationtoolbar=1 is in the URL? If it's there, then the curation toolbar will always shows up maximized. If you minimize or close the toolbar and navigate to another page that's in the Special:NewPagesFeed queue and don't have ?showcurationtoolbar=1 in the URL, then you should see see the toolbar minimized (if you minimized before) or "Curate this article" in the tools section (if you closed it). KHarlan (WMF) (talk) 13:37, 19 October 2018 (UTC)
Hi @Fram: I'm not seeing any change that should be causing this, the page curation bar has been around for a long time, and it should only activate if you load it on purpose with "curate" command, and should go away once you minimize it, then close with the x. If you have multiple tabs, close all but one, refresh the page, then try minimize/x closing it, then restart your browser and see if it stays away. — xaosfluxTalk12:34, 19 October 2018 (UTC)
@Fram: OK so it looks like the dev team put in a url hack to include a force load via ?showcurationtoolbar=1 when following links from Special:NewPagesFeed. As this is the 'default' now, it would require some hacks to avoid it (such as the css snippet I placed above). I could make that css in to a simple 'gadget' that could be enabled? — xaosfluxTalk14:59, 20 October 2018 (UTC)
MediaWiki:Hide-curationtools.css contains the code above, so we could make an opt-in to "Disable page curation toolbar" for people that never want to use it? - Not sure how big this group will be. — xaosfluxTalk15:08, 20 October 2018 (UTC)
@Fram: there is now a Gadget in the bottom of the gadgets list (in the Test section) called "Disable the page curation toolbar" - enabling this will remove your ability to use the curation toolbar via css. — xaosfluxTalk01:57, 24 October 2018 (UTC)
I'm not sure where to really ask this, so I figured this might be a good place. Was wondering if someone who is familiar with the formatting of family trees could take a look at the one in the this article. It could just be my computer, but the tree is seems to be too wide to fit on the page. It's probably just in need of a tweak or something. -- Marchjuly (talk) 00:59, 18 October 2018 (UTC)
I've mostly been using a tablet for years, but once in a while for various reasons I end uo using my laptop. Lately when doing so I have noticed features seem to be disappearing, for example:
No field for edit summary
No toolbox in edit window
When creating this very thread, no field for section title, had to add manually
Various Twinkle functions not present or not fucntional
It's an older MacBookPro, the most recent OS it can run is 10.6.8. Is there anything I can do short of spending a thousand dollars on a new one to get that functionality back? Beeblebrox (talk) 21:51, 22 October 2018 (UTC)
@Beeblebrox: most of these are more of a function of your browser than your operating system. You may be using an old version of Safari - see if you can update, or try installing another browser such as Google Chrome. — xaosfluxTalk22:05, 22 October 2018 (UTC)
Last time I tried any of that I wasn’t able to switch or update, al th browsers I tries only supply the newest version and none of them are compatible with that OS... Beeblebrox (talk) 22:07, 22 October 2018 (UTC)
I think that's snow leopard, which puts it at about 8 years old, which sounds about right for an OS not able to run the latest browsers. This is why the "real" geeks are all linux guys: we know how often we have to upgrade and we don't want to go broke doing it. And therein lies a possible solution: replace your OS with linux, which is generally much more forgiving of older hardware, and will allow you to run the latest Chrome or Firefox. It'll be slow, but it's cheap (free) and it'll work. ᛗᛁᛟᛚᚾᛁᚱPantsTell me all about it.22:27, 22 October 2018 (UTC)
Interesting. I may just do that, as another (non-wiki) issue I have is that my music is “stranded” on it in an obsolete version of iTunes and I can’t figure out how to get it out. I ripped several hundred of my old CDs into it, and that’s soemthing I really don’t feel like repeating. Beeblebrox (talk) 22:58, 22 October 2018 (UTC)
@Beeblebrox: Before you start, make sure you backup that music (and all your other files), because installing Linux will almost always involve reformatting your hard disk. Also, any music you downloaded from the iTunes store prior to 2009 is probably wrapped in DRM, so you should check to ensure that the iTunes store still has those songs, then delete them and re-download them (to get DRM-free versions). If that doesn't work, use iTunes to burn a music CD (which will sometimes erase your files; don't worry too much if it does, you have the music CDs), and then you can rip the music CD after you've done so. You may need to strip some sort of music CD DRM when you rip it, but that should be easy enough to figure out and do on Linux. ᛗᛁᛟᛚᚾᛁᚱPantsTell me all about it.15:49, 23 October 2018 (UTC)
@Beeblebrox: I'm using an early 2008 Mac Pro and that can run OS up to 10.11.6, so yours must be a really old one (though I still have an old G3 that runs the Classic OS, for when I really feel nostalgic - I used to do software dev on that). As for your iTunes rips, they should exist somewhere as .jpg or .m4a files or something - ones you ripped shouldn't have DRM protection. I have some in "/Users/username/Music/iTunes/iTunes Media" (though that's on my current MacBook, so yours might be different, but they should be somewhere like that). Boing! said Zebedee (talk) 16:24, 23 October 2018 (UTC)
Wow, thanks for all the feedback. Guy’s solution looks promising actually. This poor old laptop isn’t even really a laptop anymore, the display went bad years ago so it’s hooked up to a monitor and a trackball and effectively a desktop, which means I have to keep a desk that I use for literally nothing else. Beeblebrox (talk) 19:14, 23 October 2018 (UTC)
I mean, I !voted in an RfA recently using Opera 10 on Windows 98 SE. The major things that I noticed was some fonts didn't work properly (I could install them), and the extended watchlist didn't collapse (Javascript support issue). Mostly, everything seemed fine. Then again, Opera 10 was released in 2009, so is relatively contemporaneous. While you won't be able to get the very latest Firefox [18], you probably can get a fairly new version. ∰Bellezzasolo✡Discuss17:59, 24 October 2018 (UTC)
Major changes to talk page archive assistant script
When using Archy McArchFace to archive talk page threads, selecting a level n header will now select all of the sub-headers that it contains. You no longer have to select those headers yourself. Happy archiving! →Σσς. (Sigma)06:34, 25 October 2018 (UTC)
initial sort order
If there is a sortable table is there a way to have it initially display in a different sort order than the natural order? (There are sortable table that are stored in reverse chronological order but someone tagged it for change saying it should be in chronological order. Rather than change the order it's written in can it just be coded to initially show in ascending order by the year field?) RJFJR (talk) 20:59, 24 October 2018 (UTC)
This is not possible today on Wikipedia. Someone could ostensibly build some template or Lua for it. If the table is simple enough, you can copy and paste it into e.g. Excel, resort it with the corrected default sorting, and then paste it back into the VisualEditor. (VE is really good for table manipulation, even if it isn't very performant with long tables.) --Izno (talk) 21:53, 24 October 2018 (UTC)
@RJFJR: Sorting is Javascript. Javascript only runs in smart browsers. There are many 'dumb' browsers. By using the solution you present, you would EXCLUDE some audiences from the presentation you want to give them. This is why it doesn't exist. We tend to only 'enhance' things with Javascript, not solely rely upon them. —TheDJ (talk • contribs) 05:26, 25 October 2018 (UTC)
Wikimedia images don't load for me, even when using a VPN/Proxy
I don't know what's wrong. I messed around with Sniproxy a bit, flushed the DNS (on Windows) and changed the IPv4 DNS to Google's DNS, but they were to no avail. I have asked many people and they've all said the images work for them so it's not my ISP or the Iranian FIrewall. What could be the issue? — Preceding unsigned comment added by 5.232.151.237 (talk) 06:22, 25 October 2018 (UTC)
In the page Portal:Contents/Overviews, the language link for Malayalam wikipedia is not redirecting correctly. When I click on the malayalam language link, it redirects to കവാടം:ഉള്ളടക്കം/അവലോകനങ്ങൾ/ആമുഖം. I actually needs to redirect it to കവാടം:ഉള്ളടക്കം/അവലോകനങ്ങൾ. I changed in wikidata and also tried purging all the three pages (malayalam and english wikipedia pages and wikidata page) but in vain.Adithyak1997 (talk) 10:28, 25 October 2018 (UTC)
You added an override to Portal:Contents/Overview/Intro.[20] I have reverted it since it's already at Wikidata. If you add an interlanguage link to a template or other page used for transclusion then place it inside <noinclude>...</noinclude>. Otherwise it will also be applied to pages transcluding it. PrimeHunter (talk) 10:57, 25 October 2018 (UTC)
Editing notification list for “A link was made from…”
I get notifications occasionally that tell me of links being made from one page to any of the pages I have created. However, this includes many pages I don’t care about, like an AFC draft that i moved from user space, and excludes pages I want to monitor, like attempted deorphanings. Is there a way to edit this list, like with the main Watchlist? Could this feature be added? --Nessie (talk) 16:56, 24 October 2018 (UTC)
Is there a way to enable this feature for other pages, then? This seems like it would be technically feasible. --Nessie (talk) 14:25, 25 October 2018 (UTC)
When deleting a batch of pages, most of my deletions went normally (although a little slow), but one said something like "Deletion has been requested; please be patient". [I forgot to save the exact wording]. What MediaWiki page is this? I have activated the "message names" option for my toolbox, but when I clicked it, I merely got the normal "this page has already been deleted" warning, since my deletion action had gone through. Nyttend (talk) 19:14, 20 October 2018 (UTC)
I've been getting that when deleting templates; I'm wondering whether whatever the WMF broke this Thursday has confused whatever searches for transclusions. ‑ Iridescent19:20, 20 October 2018 (UTC)
(The rationale for T198176 is so that we can delete pages with long histories. You can review the linked tasks and discussion. --Izno (talk) 20:40, 20 October 2018 (UTC))
In the last several days, deleting or undeleting or moving a page has taken much longer than formerly, even if the page has only a few edits. Is there a reason for this? Anthony Appleyard (talk) 07:30, 23 October 2018 (UTC)
I experienced this yesterday: XfD closer timed out while trying to delete a page (and it wasn't my internet, which was streaming youtube just fine a little later). So this appears to still be an issue. Vanamonde (talk) 15:16, 23 October 2018 (UTC)
I just came here to ask about this too. It's been happening on pages even with only a handful of revisions. I'm getting normal speed page loads otherwise, so it doesn't seem to be an issue with my connectivity or machine. SeraphimbladeTalk to me17:29, 23 October 2018 (UTC)
I believe that they altered the software so that when deleting a page, it will update the link tables straightway, and not report back as a done task until all updates are complete. Previously, it was asynchronous: the deletion was carried out, a request to update the link tables was queued, and it would report back as a done task even if the link update job was still pending. --Redrose64 🌹 (talk) 20:13, 23 October 2018 (UTC)
If that is the reason, could it not be changed back? The current delay is seriously inconvenient in terms of wasted time. I am accustomed to spending a lot of time in CSD.----Anthony Bradbury"talk"20:17, 23 October 2018 (UTC)
Can anyone explain, and ideally correct, the current problem which appears to affect certainly several surveyed admins, if not all of them. Speedy deletion has, for the last three days, been painfully slow. When formerly the process took around 2 or 3 seconds it now takes 12 to 15; this seriously impedes my work here, which focusses on the WP:CSD page. Batch is also slowed, in what appears to be a similar ratio. I suspect a software bug, which I am wholly incompetent to root out; But it would be helpful if some could.----Anthony Bradbury"talk"19:50, 23 October 2018 (UTC)
Presumably you are asking about ((WikiProject Basshunter)). The problem that you describe usually occurs for one of two reasons.
when the WikiProject banner on a talk page is given an explicit |class=na. The |class= parameter should normally be omitted, except for the talk pages of articles and of disambiguation pages. Other than those two cases, it will autodetect the class; what it autodetects it as will depend upon the quality scale (see next point).
when the WikiProject banner template is configured to use the standard quality scale - this is certainly the case here, since it has |QUALITY_SCALE=yes, so it will only autodetect whether the talk page is that of a page in mainspace or not - if not, it sets NA-class.
To allow for Category-class and Template-class to be allowed (and autodetected), you need to either use the extended quality scale, which will give you Category, Disambig, Draft, File, Portal, Project and Template; or use a custom scale - for the latter, you can turn on Category and Template individually, leaving the other five turned off. --Redrose64 🌹 (talk) 20:14, 24 October 2018 (UTC)
Alter |QUALITY_SCALE=yes to |QUALITY_SCALE=extended and save. Wait for the template's transclusions to work through the job queue, and you should see Category:Category-Class Basshunter articles, Category:Template-Class Basshunter articles and five others being populated. --Redrose64 🌹 (talk) 22:39, 24 October 2018 (UTC)
A WikiProject with one participant is a waste of time. You should have recruited at least half a dozen editors before attempting to create that project. Anyway the technical aspects have been well answered by Redrose above and the template is functioning correctly. — Martin (MSGJ · talk) 07:48, 25 October 2018 (UTC)
@Redrose64: Yes I read it but it's not mandatory. I could make request to "advertise project to people who might like to join, and to get suggestions from knowledgeable editors about improving". This project will work better than these with even over 20 members which are practically daed. Isn't it possible to add redirect class to extened version? I think it would be simplest way and would be helpful in the future as include all classes.Eurohunter (talk) 09:24, 25 October 2018 (UTC)
Should this project be scrapped and just sent to WP:MFD? Then maybe start over when a proper knowledge base of creating a WikiProject is obtained. StarcheerspeaksnewslostwarsTalk to me17:36, 25 October 2018 (UTC)
Two features I particularly love while browsing WP are (1) Page Previews visible by hovering over wikilinks and (2) flipping through an article's images by double-clicking on one.
Unfortunately, it is not possible to use these features together. Specifically, when I am flipping through an article's pics, and read a caption with a wikilink, I cannot hover over it to preview the linked article.
Enabling previews in this environment would benefit WP readers like me who enjoy browsing the encyclopedia for the joy of learning new things.
@YBG: While I think it's a good idea, it would probably be more trouble than it's worth to develop this (very small actual benefit to readers, no benefit to editors). Also, because only ten proposals are picked for Community Tech to work on, I think this probably won't get selected (I made three unsuccessful proposals last year but am planning to make three more this year anyway) and it would probably be better to propose something that would have a bigger impact for readers or for editors. Jc86035 (talk) 07:58, 25 October 2018 (UTC)
@Jc86035: Yea, I recognize it is a small thing - but I was kind of hoping that it might not require much effort. Maybe rather than considering it a new feature, it could be characterized as a bug in the implementation of Page Previews. Thanks for your input! YBG (talk) 15:56, 25 October 2018 (UTC)
So, the image in this userbox should have the Wikipedia logo against a gray (i.e. transparent) background, but instead the PNG image is rendering the Wikipedia logo against a white background. This issue has been posted here before – see Wikipedia:Village pump (technical)/Archive 167#Transparent png files – but it looks like the phab ticket associated with the post has been closed as "Resolved": see phab:T198370. This obviously isn't a high priority issue, but it's a little bit of an eyesore. Ekips39noticed that the issue only occurs at specific sizes:
I'm not seeing it. For me, the image in the userbox has a grey background, and the images in the black div have a black background. If I click through to the image page, I'm seeing the checkerboard background that means it's transparent. Could it be an issue on your end? What browser/OS are you using? I'm on Chrome 70.0.3538.67 (and also tested on Firefox 59.0.2) on Ubuntu 18.04. rchard2scout (talk) 07:59, 25 October 2018 (UTC)
Funny, now the image in the userbox has a transparent background and only the 40px, 50px and 100px versions in the black div have white backgrounds. Previously only the unsized and 10px had transparent backgrounds. Progress? ♫ekips39(talk)❀08:26, 25 October 2018 (UTC)
Same as Redrose64: 10px only is white-backgrounded. Chrome 69.0 on Win7. Zooming in and out or changing browser's window display size does not change anything (even if reloading the page). TigraanClick here to contact me09:08, 25 October 2018 (UTC)
I'm not seeing this bug either. Mop in the userbox has a grey background and the same goes with the mops in the black background. Even if I tried experimenting with a Wikipedia meme vector image here, which goes up to 200px, and with a different background color, I'm still not seeing it, even though they are just PNGs automatically generated from an SVG. Windows 10 user here using Firefox 63. theinstantmatrix (talk) 10:20, 25 October 2018 (UTC) edited 10:26, 25 October 2018 (UTC)
From what I see the one in the userbox has a white background (I'd always assumed that was intentional). Of the ones in the black box, the 2nd and 4th smallest ones have white backgrounds, while the rest have black. I'm on Firefox. ~ ONUnicorn(Talk|Contribs)problem solving13:59, 25 October 2018 (UTC)
Windows 10 with Google Chrome here; the userbox mop has a grey background, the line of different sizes all has black backgrounds. Home Lander (talk) 15:31, 25 October 2018 (UTC)
Should have mentioned that if I load the image on WP, the background is white, but if I go to Commons, the background is white/grey checkerboard. Interesting. Home Lander (talk) 15:50, 25 October 2018 (UTC)
@Home Lander: This has been the case for several years. It's because the site CSS at English Wikipedia and at Commons have different rules for file description pages (in particular, the background at English Wikipedia is plain until you hover over it, when it becomes chequered; at Commons it's always chequered using this pattern repeated on both axes). These rules do not affect how an image is displayed elsewhere. --Redrose64 🌹 (talk) 23:43, 25 October 2018 (UTC)
BTW I'm currently using Firefox 62 on Ubuntu 16.04. Will report back shortly with results from my other browsers and computer. ♫ekips39(talk)❀16:21, 25 October 2018 (UTC)
Actually I flushed my cache and the transparent backgrounds are all there. The white doesn't display in any other browser or here on Windows 10. Cleared up for anyone else? ♫ekips39(talk)❀16:31, 25 October 2018 (UTC)
It also happens for me in all four tested browsers. Simplified example showing it happens for the longest group name "President of Ethiopia":
Presidents of Ethiopia
Flag of Ethiopia Ethiopia
Meles Zenawi
Flag of Ethiopia President of Ethiopia
Meles Zenawi
Group names in a navbox are right-aligned. I guess that works poorly when text is combined with a right-aligned floating image. An inline image (no |right) to the right of the text works for me:
Your edit was not saved because it contains a new external link to a site registered on Wikipedia's blacklist.
...
The following link has triggered a protection filter: whale.to
...
This isn't even true; I haven't added any new external links to whale.to, and these whale.to links have sat around undisturbed since April 2006, or more than 12 years. It should be possible to save an article with existing blacklisted external links without removing the blacklisted external links. Is there a class of Wikipedian who has that power, and if so how do I submit edit requests? And if not, please consider this a request to create a way to edit articles with protection filter violations. —Anomalocaris (talk) 00:35, 26 October 2018 (UTC)
@Graham87: oops, thanks for the note - looks like phab:T36928 to allow overrides has been stalled for years that could enable this one day. So Anomalocaris, the only current bypass is for an admin to remove the blacklist item or add a whitelist override. — xaosfluxTalk04:43, 26 October 2018 (UTC)
A case like this solved by putting nowiki tags around the offending URL. I had a very quick look at the page history ten minutes ago and it looked like you had done that. Johnuniq (talk) 04:59, 26 October 2018 (UTC)
Error handling, parser functions and #iferror on Wikipedia
Is it technically possible, in case of error for parser functions on Wikipedia to write also incorrect input (or error message given at input [#iferror]), to instead
Expression error: Unexpected < operator
((#expr:<span class="error">Error of the Template</span>))
get something like
Expression error: Unexpected < operator: Error of the Template
((#expr:<span class="error">Error of the Template</span>)): <span class="error">Error of the Template</span>
#iferror is good only for one parser function at a time. If the functions/templates are nested (ex. parse input from ((Inflation/sandbox)) in #expr in formatnum/price), then it's sometimes hard to pinpoint the exact reason for the error (it would be easier if there was bad input/error message printed). Maybe it could be done only for preview? --89.25.210.104 (talk) 14:18, 26 October 2018 (UTC)
One particular article virtually impossible to edit on mobile?
So last night I tried to add a link to the hatnote on Rwandan genocidediff. I was using my cell phone, which is a virtually brand new Galaxy Note 8. I've never had a problem editing using it before (aside from the simple fact that WP's mobile interface sucks). However, editing that particular article caused it to be very, very slow - like I'd type one character and have to wait 30-40 seconds for it to appear. When I tried to take the underscores out and replace them with spaces the whole browser crashed. That happened 3-4 times until I gave up and left them in. This morning I'm on a desktop computer and I had no problem with it diff. Anyone have any idea what the problem could be? Again, it's just that article I had trouble with. I was able to make othereditsfine. ~ ONUnicorn(Talk|Contribs)problem solving13:34, 25 October 2018 (UTC)
One thing that slows the editor a lot (on large pages) is syntax highlighting. So you can disable it by tapping the highlighter/marker icon in the toolbar at the top of the editor window and that should make the editor a lot faster. Galobtter (pingó mió) 15:15, 25 October 2018 (UTC)
Thanks for the tip Galobtter. I didn't have syntax highlighting enabled, didn't even know it existed. However, I can see how it'll make editing some pages easier. If it causes the editor to slow down I'll leave it off unless I need it. ~ ONUnicorn(Talk|Contribs)problem solving17:21, 25 October 2018 (UTC)
@ONUnicorn: I can't find it right now, but I also remember that there was a specific bug in Chrome/Android recently which had this behavior. It was fixed since, but if you happen to be on that specific version of the browser, than you could be encountering that as well of course. —TheDJ (talk • contribs) 20:52, 27 October 2018 (UTC)
Hello
I've got this issue with a Timeline on this page Nuclear Assault
Timeline chart
I've tried to edit the LineData since the problem seems to come from there, but I can't find any sense in it. It seems to work with 1-5 items, but not with 6.
Just wanted to report it, maybe it's a bug in the chart generation, maybe it's an error in the page source and I just don't see it. — Preceding unsigned comment added by Diskad (talk • contribs) 18:35, 26 October 2018 (UTC)
@Diskad: This will usually be fixed by any change which forces a new version of the image. Just try to change one character somewhere. If it works in preview then save. PrimeHunter (talk) 18:54, 26 October 2018 (UTC)
Thanks, that seems to do the trick. So this can be closed now. Should i remove this conversation and the screen dump? What is the proper etiquette? Diskad (talk) 21:50, 26 October 2018 (UTC)
I posted this at Talk:Babe Ruth but got no response so I thought I'd try here. When I click on "show" on the navboxes at the bottom of Babe Ruth, they don't open (Chrome, Windows 10). Is this happening to others? I tried cutting everything *except* the navboxes out of the page in preview mode, and they worked fine when I did that, so I'm not sure the navboxes themselves are to blame. Mike Christie (talk - contribs - library) 01:21, 27 October 2018 (UTC)
I don't think it's that -- I do have a couple of dozen tabs open, but I just tried it on Safari on my iPhone and that also fails. Must be something in my gadgets; I'll try disabling things and see if that fixes it. Thanks for the replies. Mike Christie (talk - contribs - library) 08:14, 27 October 2018 (UTC)
Historic flag
I'm trying to get the former Greek flag as a flagicon, using the code ((flagicon|Greece|1906)). This should display the Greek flag as of 1906 (i.e. the former flag), but it's instead showing the current Greek flag, which wasn't adopted until 1978. Can anyone please help? Joseph2302 (talk)08:40, 27 October 2018 (UTC)
You should use ((flagicon|Greece|old)); that's how it was named. 1906 is quite arbitrary not sure how you got it, since the flag was has been in use before and after that year. –Ammarpad (talk) 09:11, 27 October 2018 (UTC)
Hello Wikipedians! Recently, when using English version Wikipedia via mobile app, I noticed a really nice feature that displays little speaker button next to the page title in the app. The button allows the reader to hear the pronunciation of page name without leaving the article itself. Sorry for asking for help here, but I'm just really conscious how can I achieve such a simultaneous feature in Polish Wikipedia, which I mainly edit. I was looking for some CSS class that would force the app to detect proper .ogg file and display the button, but, unfortunately, nothing that I tried was successful. Thank you in advance, Karol Szapsza (talk) 15:00, 26 October 2018 (UTC)
Fixed. The words were within templates for emphasis but the tool was unable to discern that. Now changed to use plain HTML element, it's showing. –Ammarpad (talk) 17:42, 27 October 2018 (UTC)
The target of WP:BLP said: adding ((strong|information about living persons)) to ((em|any)) Wikipedia page. It's normal that popups skips templates. It skipped "any" for the same reason. Wikipedia:Tools/Navigation popups#Options has popupPreviewKillTemplates: "If true, templates referred to in an article are simply deleted from previews; otherwise, they're shown as raw wikitext." It defaults to true. Popups is a gadget disabled by default and not a part of MediaWiki. I'm not sure we should be writing wikitext with popups in mind. A few sections up at #Some technical questions on a thread at VPP is a link to a discussion where some people think we should write wikitext to control which image is chosen for an article by some features. There are also people who think we should write to try to control Google search results pages. I think we should focus on the how the actual page looks. PrimeHunter (talk) 17:51, 27 October 2018 (UTC)
There is a (inactive-ish) ongoing project to rewrite Popups to use an actual parser (Parsoid) instead of the current (mildly broken) one. When that project is done, this and other Popups parsing issues should be resolved. In the meantime, I agree with PrimeHunter that it may not be the best idea to rewrite wikitext for Popups usage. Enterprisey (talk!) 18:59, 27 October 2018 (UTC)
Thanks for the insight, everyone. I didn't know a template existed for emphasizing text when simply bolding it works just as well. Home Lander (talk) 21:03, 27 October 2018 (UTC)
@Home Lander: Bold using wikitext and the template ((strong)) are not the same under the hood; one applies the <b>...</b> element and one applies the <strong>...</strong> element. (I assume you can tell which is which.) These apply different meanings to the text in-question. Here is a brief discussion on their meaning. --Izno (talk) 23:10, 27 October 2018 (UTC)
Huge Script errors in console log
I was recently testing my own JS and noticed that there are four or five huge red warning messages(which seemed unrelated to my script) at the console log which I had not seen until very recently.
Text of one of the warnings
load.php?debug=false…ipts&skin=vector:10 [Report Only] Refused to load the script 'https://tools-static.wmflabs.org/meta/scripts/pathoschild.templatescript.js' because it violates the following Content Security Policy directive: "script-src 'unsafe-eval' 'self' meta.wikimedia.org *.wikimedia.org *.wikipedia.org *.wikinews.org *.wiktionary.org *.wikibooks.org *.wikiversity.org *.wikisource.org wikisource.org *.wikiquote.org *.wikidata.org *.wikivoyage.org *.mediawiki.org 'unsafe-inline'".
Could someone with more technical knowledge confirm if everything is okay.
Seems like a new mediawiki report that shows you and apparently others on this wiki and several others are loading javascript from a "third" party site. That is a massively bad and naive approach. The person who controls access to that script can easily modify it and change your settings or collect your data. An admin or global administrator who is dumb enough to do that that would immediately provide the ability to do stuff like block, remove rights and have a party. Their account could easily be hacked. Funny thing is that some admins have naively been loading this tool.
Worse thing is that there is at least one wiki that users were dumb enough to enable such a nasty hack has a javascript tool for all users.08:11, 26 October 2018 (UTC)— Preceding unsigned comment added by 197.235.132.239 (talk)
Hi folks. Yes, we just enabled CSP in report only mode on the sites. You don't have to worry too much about this now, we'll definitely be giving people notice before enabling in strict (blocking) mode. Additionally this currently will warn both for external resource loads and external scripts (your example above is an external script load). We eventually plan to have some sort of preference opt-in to allow external (non-script) resource loads for people who want that while still protecting the vast majority who don't use that (details of that are yet to be worked out). With all that said, we strongly encourage you to only load scripts that are saved on wiki. Remember, any external script that you load has full control of your account and can do anything you could do. BWolff (WMF) (talk) 08:25, 26 October 2018 (UTC)
Isn't the issue in this particular case (for meta:TemplateScript) that wmflabs.org is considered "external" unlike from meta (or wikidata or...)? Seems like *.wmflabs.org is missing from the 'self' list. ~ Amory(u • t • c)11:22, 26 October 2018 (UTC)
A question like that undeniably proves the point made above. It should also be worrying that someone who seems to lack basic knowledge of sofware / programming security practices not only has the interfaceadmin rights, and also very likely has the exact same script loaded. Generally, this proves that this measure will resolve very little as long as some sort js review , much like flagged revisions doesn't exist. Chances are that most (if not all) other interface administrators also load resources from javascript pages that they don't control, e.g. normal user's js pages. If it is this bad here, smaller wikis must be infinitely worse.197.235.132.239 (talk) 12:22, 26 October 2018 (UTC)
Hey, Amory meant no harm. There's a lot of good stuff on Toolforge. Indeed the barrier of entry to creating things there is relatively low, but not everyone knows that. As it stands now, on-wiki JS is not necessarily safe either, for the record. No need to go into details.
There are a hoard of scripts heavily relied upon that source Toolforge. It is a no-brainer that this will definitely be a problem. The preference opt-in to allow external resource loads will have to be a requirement. Some things you just can't do on-wiki. That being said, on the surface it would seem TemplateScript operates fully client-side and doesn't need to live on Toolforge at all?
My hope is we can selectively whitelist specific patterns, such as xtools.wmflabs.org/* (or tools.wmflabs.org/my-tool/*). The ArticleInfo gadget for instance is very popular. This functionality cannot feasibly be implemented on-wiki, and there will be a lot of unhappy people if this can no longer be sourced :( — MusikAnimaltalk17:36, 26 October 2018 (UTC)
@Musik , Let me put it bluntly, Amorymeltzer (an interface administrator) seems to be using (at the time of this writing) the script above in their account. That should immediately be something of great concern to all interface admins because it essentially means that anyone who has access to that repository is essentially a interface administrator (as implied by BWolff), and that's not a right that the presumed script author is currently allowed to have on this wiki. With just a bit of random searching, it was easy to identify that the same script is being used by some users with significant cross-wiki rights, so by the same token the user also has access to those rights, and that should be very concerning to Wikimedia system administrators.
Going by wikia's help pages, they seemed to have reacted in a somewhat more effective way to similar incidents, and even developed (some) sensible rules and extensions (similar to https://github.com/TK-999/mediawiki-extensions-ContentReview) governing the use of javascript.
About the scripts you mention, considering that most deployed scripts neither validate the input properly nor escape the output (nor is there even enforcement of such practices), even if they are loaded as text or json they will still pose a significant risk. The ArticleInfo can mostly be implemented locally using the mw:API:Revision, for pages without too many revisions. 197.235.144.129 (talk) 21:35, 26 October 2018 (UTC)
I'm the script author for the TemplateScript mentioned in the error. Since I'm already an admin and global interface editor, repository access doesn't give me any new access. I agree that pages in the MediaWiki namespace are more secure than a third-party site, but keep in mind the historical context: way back when these scripts were written, scripts were routinely saved in the user namespace where anyone could edit them. Moving them to Tool Labs was a big step up for security at the time. For example, the user scripts that use TemplateScript in this case are still stored in an unprotected user page, so they're much less secure than the Tool Labs scripts. —Pathoschild 21:45, 27 October 2018 (UTC)
I'm the IP who made the notes above. Thanks for the historical context, it was very enlightening. It was certainly a worse situation than what currently exists , and it explains the easy going manner of admins and even wikimedia system administrators here. Although a more sensible approach at the time would have been to place the script in a page, request that admins protect it, and maybe allow custom trusted groups access to it. Alternatively, it could have been stored in the mediawiki namespace which presumably has always been protected.
I stand corrected about the account not having interface admin rights, but it still doesn't eliminate the fact that you currently have access to steward, and global sysop rights which your account shouldn't have and isn't listed in Special:CentralAuth/Pathoschild. This access exists because some stewards, global sysops and even mediawiki developers load your script. It was trivial for me to find this. Hackers would find even simpler ways of escalating privilege, e.g. by hacking regular accounts (with lower account security) that have editing rights to scripts loaded by privileged accounts. Looking at the logs in central wikis like commons and meta, it seems that some admins and even stewards have been compromised, and it is very likely that javascript was the tool used to hack some of these accounts. 197.218.90.129 (talk) 11:33, 28 October 2018 (UTC)
(edit conflict) Much as I appreciate the brow-beating, the question asked was why this showed up in the console and whether everything was okay. After being insulted, Brian gave the correct, jargony answer ("enabled CSP") that it was "external" but did not explain why this particular error showed up. I attempted to be helpful and explain that, unlike other sites (i.e., everything) which could reasonably be understood to be external to enWiki but do not provide warnings, wmflabs is indeed missing from the list in the console message, thus it provides a warning where as those listed (meta, wikidata, etc.) do not. ~ Amory(u • t • c)17:43, 26 October 2018 (UTC)
hi User:Amorymeltzer - my apologies for being jargony. Sometimes i slip into it without meaning to. For the purposes of this, we consider labs to be external. Basically because anyone can create a labs project. While of course many (most) labs project are done by trusted people, we ultimately want to limit script loads to places WMF truly controls, and has a publicly auditable view of previous versions of the script (i.e. on wiki page histories). Additionally we consider labs to be a seperate "security domain" and want to avoid mixing in executable code across security domains. Last, with all due respect to the wikipedia community, the people who use user scripts are of course not security experts, nor should they be (nobody can be an expert on everything). Part of the reason we want to eventually limit freedom here, is to prevent users not familar with the intracsises of web security from making insecure choices they might not understand the consequences of (or prevent the people who write the scripts other people rely on from making poor choices for people who use their user scripts). Re Musikanimal: indeed, we are not sure what the solution for loading (non-script) data from labs/external data looks like. I agree we will need to allow it for certain classes of users, we will be writing up a plan for that some point in the near future. BWolff (WMF) (talk) 21:48, 26 October 2018 (UTC)
I'm sorry, but I cannot parse what is going on, what error my script is responsible for making, and thus the nature of problem being created for the wider Wikimedia project that is implied. If any expert can tell me what changes I can make to my script to address the issue, I would be most grateful. -- Ohc ¡digame!20:06, 27 October 2018 (UTC)
Ohconfucius, It is loading code from a domain that is considered to be not trusted: "tools.wmflabs.org". As JS attacks have been increasing, the WMF security team have added a new measure to the website, which tells the browser which domains and which technologies should be trusted (CSP) by your browser. This measure is currently in report-only mode, meaning that it will show a warning if you load code from those domains in your Javascript console, as well as reporting the domain violation in question back to WMF (so that WMF gains insight into what will break when they make this more than just reporting). Eventually this reporting will probably be turned into a blocking mode, at which time the script would actually stop working. I have created a ticket for Pathoschild, to move that scripting code back into meta.wikimedia.org. —TheDJ (talk • contribs) 20:48, 27 October 2018 (UTC)
While it's easy to copy the scripts to Meta (presumably in the MediaWiki namespace), there are countless references to the scripts across all wikis that will need to be updated. —Pathoschild 21:45, 27 October 2018 (UTC)
All I want for Christmas
is my two front teeth; and maybe one other thing (depending on cost).
I use "ExpandTemplates" quite often, and badly wish the page had an intuitive set of extensions like search and replace, some "CharInsert"-ability (for parsers and magic words that are so painstakingly, otherwise, generated by hand) and maybe a table generator like what we already have on editing NavBars, except for the page where they, arguably, are needed most. For me, such functionality would be hugely beneficial, and immediate dividends (in salvaged lost time) could be reaped. Is this something that could reasonably happen? Thank you — and Merry Christmas.--John Cline (talk) 05:18, 28 October 2018 (UTC)
The # of watchers a page used to be clearly-delineated at the top of the editing window but it's sort of disappeared-ish...
The # of watchers on a particular page used to be clearly laid-out at the top of the editing window but it's been moved - duh on me, I suppose - and I had to dig through some of the editing tools to find it. Just in case anyone else around here had lost the watchers function FYI - go to "Pageviews" in the page history window, click on that and "Pageviews analysis" comes up and right there! Over on the right! Under Basic information you will see "Watchers". Am so happy I found it I just had to share. Shearonink (talk) 06:47, 28 October 2018 (UTC)
If you click "page information" under "tools" on the left navbar, it includes the number of watchers, and some other good stuff, for a click or two less. In case that is helpful as well. Cheers.--John Cline (talk) 07:08, 28 October 2018 (UTC)
@Shearonink: I don't recall ever seeing the number of watchers displayed at the top of the editing window. Perhaps you had some gadget or other user script that did that, which has been amended recently. --Redrose64 🌹 (talk) 08:41, 28 October 2018 (UTC)
I believe Shearonink is referring to the "number of watchers" link formerly present on history pages. That was indeed removed after a lengthy conversation at MediaWiki_talk:Histlegend. Shearonink, it was removed because if you look in the left side toolbar on wiki, you'll see a link titled "Page information." That is the same link that was originally provided on every page's history legend, so there was no need to also have it in the history legend. ~ Amory(u • t • c)11:00, 28 October 2018 (UTC)
Yes thank you Amorymeltzer that's exactly it. Went to use it yesterday and *poof* it seemed to be gone. Thx to everyone for mentioning the "Page information" link - been here for years and still don't know where to find all the stuff... Shearonink (talk) 15:45, 28 October 2018 (UTC)
Cannot edit the article Bodacious which is protected from IP and anon users, but I am registered user
Hi, I'm definitely getting run around from different places on this issue. The issue is I can't edit an article that had protection placed on it to prevent IP and anon users from editing. If it's not addressed soon, it will run out of the 2 days placed on it anyway. I am a registered user so I should be able to edit it. When I make an edit it just acts like it is accepting it, but then when I look at it, the edit is not there. There is no message or error message. See the message below, which leads to another message with all the details from the admin who set up the protection. They were unable to help. dawnleelynn(talk)15:51, 28 October 2018 (UTC)
Please see this conversation with the admin I just had who referred me to post at "ANI" for Bodacious (bull). When I came to the noticeboard area, I wasn't really sure which noticeboard to post it at. [24] The issue is explained clearly at the admin's talk page. But the theme is that I cannot edit an article that was protected to keep IP users from editing it temporarily. I have a login and should be able to edit it. Thank you! Going to bed but hope it be resolved w/o me. dawnleelynn(talk)06:15, 28 October 2018 (UTC)
It is odd, the article is semi-protected and you are an extended confirmed user, so you should be able to edit it. I suggest posting at Wikipedia:Village pump (technical) instead of here. 173.228.123.166
@Xaosflux: Hey, it worked that time! It was the exact edit I tried several times yesterday. I can't explain that, but I swear it didn't work several times yesterday. Thank you for looking at it. Well, I still have a bit over a day left of protection. Who can explain things like this? Thanks a bunch. Maybe since I rebooted the computer? dawnleelynn(talk)16:36, 28 October 2018 (UTC)
If you click on "email this user" on my userpage, you'll see that there is no email notice, even though I have one set. Even though I have one, it's not showing up. ThegooduserLet's Chat🍁20:37, 28 October 2018 (UTC)
I don't think any parameters are passed to that message. It looks like ((FULLPAGENAME)) needs to be changed to ((PAGENAME)). It seems that the former doesn't work in the special namespace any more. "((FULLPAGENAME:Special:EmailUser/Example))" → "Special:EmailUser/Example"; "((PAGENAME:Special:EmailUser/Example))" → "EmailUser/Example" — JJMC89 (T·C) 23:15, 28 October 2018 (UTC)
Replacing ((#titleparts:((FULLPAGENAME))|1|2)) by $1 seems to have fixed it.[25] But I don't know why the old version was broken. It worked when it was entered at Special:ExpandTemplates with Special:EmailUser/Thegooduser as context title. Maybe ((FULLPAGENAME)) is currently broken on special pages and doesn't actually return Special:EmailUser/Thegooduser on Special:EmailUser/Thegooduser even though Special:ExpandTemplates says it does. PrimeHunter (talk) 23:27, 28 October 2018 (UTC)
We'd like to invite you to the weekly Technical Advice IRC meeting. The Technical Advice IRC Meeting is a weekly support event for volunteer developers. Every Wednesday, two full-time developers are available to help you with all your questions about Mediawiki, gadgets, tools and more! This can be anything from "how to get started" over "who would be the best contact for X" to specific questions on your project.
Currently, whenever a threaded discussion is moved to an archive all of the incoming links in the form of
[[Page name#Threaded discussion]]
are left broken. What could we do, similar to leaving a redirect, to prevent these links from breaking? In my opinion, it is a worthwhile consideration. Thank you.--John Cline (talk) 13:41, 24 October 2018 (UTC)
If Page name#thread is on the same page as the wikilink, then just [[#Question about breaking links]]{test) or [[((FULLPAGENAME))#Question about breaking links]](test) should work (but I don't know what you mean by threaded, example would be nice). --89.25.210.104 (talk) 14:20, 24 October 2018 (UTC)
@John Cline: It's not a problem with Legobot (talk·contribs) (for which the bot operator (not "caretaker") is Legoktm (talk·contribs)). Anchors like rfc_F132519 are set up by the ((rfc)) template, based upon the value of its |rfcid= parameter - in this case it would be |rfcid=F132519; hence, if that parameter is altered or removed, or if the ((rfc)) template itself is removed (as happened in this edit), the anchor no longer exists within the page. Remember that once thirty days has elapsed (or an ongoing RfC is moved to a talk page's archive), Legobot removes the ((rfc)) template; and once this is done, Legobot no longer sends out WP:FRS messages for that RfC. However, it does not remove old FRS messages, which remain on user talk pages indefinitely, so care must be exercised if the FRS message was sent out more than a month earlier. --Redrose64 🌹 (talk) 20:56, 24 October 2018 (UTC)
Then maybe it's worth adding the info about possibility of link expiration after 30 days (not to mention that thread title would also be helpful [don't know if it's possible])? --89.25.210.104 (talk) 14:11, 25 October 2018 (UTC)
The link does not expire. The RfC expires, thirty days after the timestamp that next follows the ((rfc)) template. RfC expiry is followed up by Legobot removing the ((rfc)) template, with the consequence that any anchors that are internal to that template are lost. If you want to get the behaviour of Legobot changed, I suggest that you look through the bot's talk page (and archives) to see just how often a change request is accepted by Legoktm. --Redrose64 🌹 (talk) 23:51, 25 October 2018 (UTC)
It doesn't really matter which one of them expires, link still stops working without indication as to why (and you can't even search archives without thread name). It's also not clearly stated in the message and on pages linked from it (unless Talk page and "archive page no. x" are counted as documentation).
If it's a common (not recently introduced) situation that links could be dead after a while (and the thread name isn't known), it should be stated on the message/bot page (or in pages linked from that message). --89.25.210.104 (talk) 15:06, 26 October 2018 (UTC)
Links do not "expire". They cannot expire - either the anchor that they are directed at exists in the destination page, or it does not. Anchors cannot be given dates (expiry or otherwise) - they take the form of an id="identifier" attribute on a HTML element. There is no concept of an expiry date in HTML attributes, or the tags that they are within.
The anchor generated by the |rfcid= parameter has been a feature of the ((rfc)) template ever since this edit more than seven years ago. This anchor is normally only used by two things: the FRS messages mentioned above; and entries in the RfC listings, which are removed at the same time that the ((rfc)) template is removed from an expired RfC. Hence links to nonexistent anchors for expired RfCs should only exist within old FRS messages; it is not a priority to ensure that these are maintained in a working state indefinitely. If you really need people to still reach an RfC-that-isn't-any-more, add an ((anchor)) template, like this. --Redrose64 🌹 (talk) 19:00, 26 October 2018 (UTC)
Links do not "expire". They cannot expire - either the anchor that they are directed at exists in the destination page, or it does not. If a link target disappears after 30 days, then that link expires to me.
I just describe what the bot does, I don't know the whys. I suggest that you ask the bot operator, who is Legoktm (talk·contribs), not me. Also, it is absolutely nothing to do with a session, either. Honestly, I think that you're deliberately being difficult, refusing time and again to accept the explanations that are provided. --Redrose64 🌹 (talk) 21:54, 27 October 2018 (UTC)
I just describe what the bot does, I don't know the whys. If you mentioned that earlier, maybe this discussion wasn't so "difficult" (and was a bit shorter).
I think that you're deliberately being difficult, refusing time and again to accept the explanations that are provided.
And I think that we misunderstood each other - after first explanations of how bot works, all I really wanted to know is why it includes only anchor id (without thread name). --89.25.210.104 (talk) 17:39, 28 October 2018 (UTC)
This proves that you are deliberately ignoring my explanations. I stated in my post of 20:56, 24 October 2018 (UTC) that the bot operator is Legoktm (talk·contribs). That person is not me, and I did not write any of Legobot's code. If it was, and I did, I would have said so. Now, just WP:LETITGO and get out of here. --Redrose64 🌹 (talk) 20:02, 28 October 2018 (UTC)
This proves that you are deliberately ignoring my explanations. No, I'm not ignoring your explanations, I'm even thankful for explanations of HOW the bot works. But it just isn't really the thing I want to know. I'm sorry if my questions aren't clear enough (and ignore the whole "expire" topic, it wasn't supposed to be a main thing).
I stated in my post of 20:56, 24 October 2018 (UTC) that the bot operator is Legoktm (talk·contribs). And if you added in that post the sentence from your post of 21:54, 27 October 2018, then this discussion would be shorter by 3 days. Misunderstandings happens.
In general - there isn't a technical fix for this challenge with wikitext. If linking to a discussion from another page that is meant to be kept for reference etc, instead of linking to the page you should link to the version#section such as this: WP:VPT original question on links. — xaosfluxTalk14:34, 24 October 2018 (UTC)
You could also craft link to archive by ((#ifexist)) (if archive number is permanent), but it's expensive. And if it's in archive it probably shouldn't be edited anyway, so permalink sounds better. --89.25.210.104 (talk) 15:10, 24 October 2018 (UTC)
ClueBot occasionally fixes these for the pages it is scheduled to archive, but it does so based on a database that is very large, so whether it actually gets around to a timely (or untimely) fix is usually some manner of luck... There isn't really a good technical fix besides that, and that requires some pain on the part of the bot maintainers. --Izno (talk) 16:57, 24 October 2018 (UTC)
Are there any other bots to mark unsigned comments with ((unsigned)) except for Sinebot? I contacted the owner to add it to ckb but it appears that they're inactive. I was wondering if there were other bots with active owners who are willing to activate them in other wikis.--▸ épinetalk♬14:34, 29 October 2018 (UTC)
Bias can only come from humans. A machine can't have any. It only has certain patterns of actions governed by certain rules. The other thread seems to have been deliberately written to incite emotional responses. It is easy to overwrite those cases and even make it useful. For example, create an optimal image as described by the extension page, and embed it in a page to make it always selected in such cases. The image could be designed to encourage contributors to come up with a better image instead.
Some third party wiki sites even have a placeholder syntax for the [[file: markup that encourages people to pick an image, although it is not even necessary here.197.235.144.129 (talk) 22:37, 26 October 2018 (UTC)
A machine can't have any [bias]. That's not really untrue, particularly when we start getting into machine learning. The bias may be intentional in the choices made to train the AI, or the AI might reflect unintentional bias learned from the humans who trained it, or it might be completely unintentional but still be present. As an example of the latter, there's the anecdote about someone wanting to train an AI to detect tanks by giving it a corpus of images of terrain with and without tanks present, but since it turned out the weather was also different in the "with" and "without" sets it learned to detect that instead. As another example, it's fairly easy to have a pseudorandom number generator that generates numbers that aren't perfectly random in some way.
But even that is not terribly relevant to the linked discussion, since whether or not there's actual political bias in the algorithm used to choose the page images there's still an appearance of bias in the fact that an image of just one candidate is selected to represent the article about the election. And the algorithm is such that it may not be clear if someone's edit that changes the selected image did that intentionally. The current algorithm is also such that the image selected must actually appear in the lead of the article, which might complicate matters if the "placeholder" doesn't make sense for the article itself. Anomie⚔17:11, 29 October 2018 (UTC)
Unhelpful diffs
I refer to [| this diff] only as an example; please hover to see what I mean, clicking through produces a more helpful result.
The referenced diff is shown in my watchlist as '-55', i.e. the editor reduced the length of the article by 55 characters. But a cursory inspection of the diff shows a much larger block of characters
removed, and then reinstated, apparently without change. Reading the diff, it's effectively impossible to work out what was actually changed.
I was a programmer; I don't recall ever seeing a code diff that made a minor change appear as a complete rewrite. And indeed that would have been unacceptable; forcing programmers to wade through inadequate diffs would have annoyed both the coders and their paymasters alike.
I don't think it was always so; I don't think Wikipedia always showed hover diffs that were so unhelpful. Perhaps I am wrong. Is the hover diff produced by the same software as the click-through diff? Does anyone else find the hover diffs problematic when reviewing edits from the watchlist? MrDemeanour (talk) 11:31, 28 October 2018 (UTC)
One was moved and its caption was changed. Three other images were removed and instead a gallery including them was added. It was rather substantial change. Ruslik_Zero13:06, 28 October 2018 (UTC)
Speaking of diffs, there's a third option, which is the visual diff. You can enable it at Special:Preferences#mw-prefsection-betafeatures. Some diffs make more sense in wikitext mode, and others in visual mode.
It occurs to me that work on visual diffs is winding down. Are you all basically happy with it? Should I poke the product manager to get it out of Beta Features, and into some (which?) normal section of Special:Preferences? Whatamidoing (WMF) (talk) 17:35, 29 October 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Recent changes
You can post proposals for the Community Wishlist Survey. The survey decides what the Community Tech team will work on. You can post proposals until 11 November. You can vote on proposals from 16 November to 30 November.
The wikis now have a content security policy report. This means that you might get a warning in your javascript console when you load external resources in your user scripts. For security reasons it is recommended that you don't do this. It might not be possible to load external resources in your scripts in the future. [27]
Problems
Your watchlist can show which changes you have already seen. This did not work for a few days. It has been fixed. [28]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 30 October. It will be on non-Wikipedia wikis and some Wikipedias from 31 October. It will be on all wikis from 1 November (calendar).
The 2006 wikitext editor is no longer available. It will be removed from Special:Preferences. It has not been the standard editor for a long time. It was replaced by the 2010 wikitext editor. [29][30]
Meetings
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 31 October at 15:00 (UTC). See how to join.
Can the following message be disabled?
This is the current revision of this page, as edited by Nizil Shah (talk | contribs) at 09:57, 18 October 2018 (ce1). The present address (URL) is a permanent link to this version.
(diff) ← Previous revision | Latest revision (diff) | Newer revision → (diff)
Thanks PrimeHunter. I added that to common.js but it only shortened the message. The it "url" still has a suffix "&oldid=865480201" and message still appears as "(diff) ← Previous revision | Latest revision (diff) | Newer revision → (diff)" Further, if I click on edit, the edit screen doesn't work on smartphone. Capankajsmilyo(Talk | Infobox assistance)07:32, 24 October 2018 (UTC)
I see you correctly added it to User:Capankajsmilyo/common.css and not common.js. I don't know a way to only hide "(diff) ← Previous revision | Latest revision (diff) | Newer revision → (diff)" on the current revision. Are you sure your edit screen problem is related to the code? Is it a mobile edit or desktop edit? What is the url and what goes wrong? PrimeHunter (talk) 11:16, 24 October 2018 (UTC)
To hide the box beginning "This is an old revision of this page, as edited by ...", use the rule
div#mw-revision-info{display:none;}
To hide the box beginning "This is the current revision of this page, as edited by ...", use the rule
div#mw-revision-info-current{display:none;}
To hide the row that includes "(diff) ← Previous revision | Latest revision (diff) | Newer revision → (diff)", use the rule
Thanks Redrose64, but the issue is more than the visible box. oldid in url disable the ability to edit. When I click edit from my smartphone on such a page, it doesn't save anymore. The incorrect url comes when I open an article by clicking name on any edit diff. Capankajsmilyo(Talk | Infobox assistance)07:38, 27 October 2018 (UTC)
CSS cannot be used to amend the URL in the address bar of your browser; JavaScript can be used to do this, by a form of redirection. I do not think this is desirable. How many times have you visited a website, been redirected to another page immediately, then found that you cannot use the "back" button of your browser?
If the oldid is that of the current version, you should get a full set of edit links; if it is not the current version, you will only have the Edit tab at the top, no section edit links. This is intentional: we need to deter people from editing old versions unless they really need to. There was a stink a few years ago when a new MediaWiki release provided section edit links in old versions of pages - as a consequence, that MediaWiki release was rolled back within two or three days. --Redrose64 🌹 (talk) 19:33, 27 October 2018 (UTC)
Iv clicked a few screenshots to explain my issue.
When I clicked on Patan Devi in my watchlist (latest revision)
This diff screen appears, which is good, but when I click Patan Devi here
This appears instead of the article (Till one month ago it was ok)
This is the url (now, 1 month ago url did not have any oldid in such flow)
Right, so you are using mobile (Primehunter did ask "is it a mobile edit or desktop edit?" at 11:16, 24 October 2018 (UTC), and I have been working on the assumption that you use desktop), in which case I cannot help any further. --Redrose64 🌹 (talk) 20:57, 29 October 2018 (UTC)
@Magnolia677: I've moved the page. You shouldn't have had any trouble with it (if you're moving a page over a redirect, it only works if the redirect hasn't been edited since its creation). Jc86035 (talk) 11:28, 30 October 2018 (UTC)
1. How can I change the alert number in my bell, and the color of my bell?
2. How can I change the "You have new messages" and the "You have new messages from another user (last change) colour? ThegooduserLet's Chat🍁00:53, 30 October 2018 (UTC)
@Thegooduser: You can do this by editing your common CSS to add the appropriate text below, replacing "colorname" with your chosen color.
Alert number in bell:
/* Number in bell */#pt-notifications-alert.mw-echo-notifications-badge.oo-ui-flaggedElement-unseen::after,#pt-notifications-alert.mw-echo-notifications-badge.mw-echo-unseen-notifications::after{color:colorname;background-color:colorname;}/* Number in inbox */#pt-notifications-notice.mw-echo-notifications-badge.oo-ui-flaggedElement-unseen::after,#pt-notifications-notice.mw-echo-notifications-badge.mw-echo-unseen-notifications::after{color:colorname;background-color:colorname;}
How technically difficult would it be to update how the media player looks (for all users, not just me)? Would it just be a matter of submitting a pull request to update this picture? Jc86035 (talk) 16:03, 30 October 2018 (UTC)
@Jc86035: We've been working for years on replacing the entire media player with VideoJS, a much more modern and streamlined player (which has the advantage of working on mobile, which given that's more than half of all our readers…). A Beta Feature with this is planned as soon as we replace the subtitles system so that the new player can work with them. Top-level task is phab:T100106. If you want, you can test the Beta Feature out on the Beta Cluster right now (e.g.this file; note that you'll need a different account for the Beta Cluster system and shouldn't re-use a password, then enable the new video player). Jdforrester (WMF) (talk) 16:49, 30 October 2018 (UTC)
@Jc86035: Yup, the old Wikimedia bespoke system for doing subtitles isn't very good, which is why we're replacing it with something a lot closer to the way everyone else now does it. A decade of technical change on the Internet means things de facto standardise, which is helpful. :-) Jdforrester (WMF) (talk) 17:54, 30 October 2018 (UTC)
Minor improvements to the article Pegasus (spyware) have been repeatedly undone, despite references and increase of ease-of-reading(methinks). Edit summary attempt was made to tag for education and expert-needed. Anyone mind fixing the article up to reflect the reality of malice?126.3.10.162 (talk) 21:35, 30 October 2018 (UTC)
On redirect pages with the "redirect=no" at the URL's end, it previously would have ALL the text hidden below the redirect target page; a while ago, it shows all the text. Is there any way I can remove it for myself and put it back the way it was for me only? I'm using uBlock Origin for element hiding, and had the filter ##.ambox-move.ambox.metadata.plainlinks to remove templates. For example, go on a vandalized version of a redirect page and try hiding the text. It won't work with an element hider unless it's the template. Dolfinz1972 (talk) 06:43, 30 October 2018 (UTC)
Also, how can I remove "(view filter log)" from the history of a page to the right of "View logs for this page"? with element hiding or any way? Dolfinz1972 (talk) 06:43, 30 October 2018 (UTC)
Script to show number of new notifications in titlebar
Hello,
I am looking for a user script which can show the number of new unread recent changes and/or new unread notifications from Echo in the title bar of the tab, as '(3,5) The normal title goes here' for example. Gryllida (talk) 01:27, 30 October 2018 (UTC)
document.title="("+document.getElementById("pt-notifications-alert").firstChild.getAttribute("data-counter-num")+","+document.getElementById("pt-notifications-notice").firstChild.getAttribute("data-counter-num")+") "+document.title;//Add number of notifications in titlebar
Thanks this picks up echo. Do you know how to make it update live without refreshing the tab in case there is a new notification? I think echo updates itself in a tab which is already open. Gryllida (talk) 18:45, 30 October 2018 (UTC)
@Gryllida: I've never seen notifications update without reloading (and a test with my alt account didn't update until a reload). There may be an API call that can be used to grab the number of notifications, but it would be significantly more complicated than just scraping the page as I've done above. --Ahecht (TALK PAGE) 01:14, 31 October 2018 (UTC) 20:43, 30 October 2018 (UTC)
On mobile site under summary edit, can we just remove the watermark "Example: Fixed typo, added content"? Currently there are many vandalism with "Fixed typo" summary. Hddty. (talk) 16:17, 24 October 2018 (UTC)
Probably not, because it's helpful to good-faith editors (who outnumber vandals).
There's been a proposal to automagically calculate edit summaries, at least for simpler cases. I don't know if it's a suitable size project for the m:Community Wishlist (which opens on Monday), but it might be. This would suggest an edit summary such as "Changed tyop → typo" for simple spelling changes, or "Added content: "Lorem ipsum dolor sit amet consectetuer adipiscing elit" for adding new text. If you're interested, remind me to find the Phab task number (or someone else can). Whatamidoing (WMF) (talk) 17:26, 24 October 2018 (UTC)
It looks like phab:T54859 is the central task, but see also the description in phab:T65142.
These were written for the visual editor (specifically, VisualEditor's visual mode on desktop), so I think I'd recommend creating a new one for mobile editing, where it might be especially useful. "Mobile" could be interpreted as mobile-wikitext-editing and mobile-visual-editing, as well as the Android and iOS apps, but the apps are relatively low traffic, so I'd focus on the first two. Back on desktop, I don't know if it's feasible/performant enough for the 2010 WTE, but it is feasible for the visual editor. Whatamidoing (WMF) (talk) 22:02, 24 October 2018 (UTC)
Alternatively, maybe we can change the watermark to something like "Briefly describe your changes here", same as the desktop site. Hddty. (talk) 23:25, 24 October 2018 (UTC)
I'm sorry, but I've lost the context. Which "this" will desktop users want? (Hddy's proposed change to the message, or automagical edit summary suggestions?) Whatamidoing (WMF) (talk) 06:29, 31 October 2018 (UTC)
Please leave the message alone! I always check canned edit summary changes because they are often junk and it's handy to have a "check this" flag. Johnuniq (talk) 00:41, 25 October 2018 (UTC)
Citation template gadget in wikitext editor
The citation template dropdown menu in the wikitext editor doesn't seem to be working. I fill in all the fields and then click insert and it just goes away. Anyone know what's up with that? I use it a lot. It's a pain to do it manually (though I can). ~ ONUnicorn(Talk|Contribs)problem solving22:40, 30 October 2018 (UTC)
Try previewing the page a few times? Not sure if it's related, but something that fixes the issue for me (non-safe mode though). Headbomb {t · c · p · b}13:39, 31 October 2018 (UTC)
Previewing the page brought back the citation dropdown menu in safe mode. However, the behavior of the pop-up box to fill in the fields did not change. When I clicked insert it just went away. ~ ONUnicorn(Talk|Contribs)problem solving13:58, 31 October 2018 (UTC)
The upcoming sequel It – Chapter Two has an extremely unique issue, in that searching the title as traditionally stylized (It: Chapter Two) redirects you to the Italian Wikipedia page for "Chapter two" (click here to see it). The hatnote explaining the issue due to technical limitations is already in place, but with the film's release next year there will undoubtedly be people unfamiliar with Wikipedia's naming systems who can't figure out why they keep ending up on an Italian page. Is there any decent way to perhaps indicate on the Italian Wikipedia that the user might be looking for the film? Again, never had an issue even remotely similar to this so I figured I would consult here. Thank you! Sock(tock talk)16:21, 31 October 2018 (UTC)
There's already a hatnote --and link-- on the Italian page explaining the issue. So when people land there, they'll see how to come back here. –Ammarpad (talk) 17:34, 31 October 2018 (UTC)
Not sure if this is already available somewhere, but I was trying to find something that would allow the top page buttons to "stick" to the top of the window while scrolled. Like the Talk, Edit, History, etc buttons/menus. I think most of it should be possible to accomplish with straight CSS, but I wanted to see if anyone had a solution before I started hacking away on it. zchrykng (talk) 19:43, 31 October 2018 (UTC)
You could stick these and float them over the page with:
Xaosflux, thanks, but definitely not what I'm looking for. It would definitely require more extensive css than that. Looking for something more like this in effect. zchrykng (talk) 20:33, 31 October 2018 (UTC)
@Zchrykng: you could also define the height of the "content" section of the page and have it scroll with css - think you will have to fix to a specific height though, example:
And yes, it could be done with more complicated coding - don't have a lot of time right this second so just giving you some ideas. — xaosfluxTalk20:37, 31 October 2018 (UTC)
@Zchrykng: try adding this to your skin css (e.g. User:Zchrykng/vector.css) I dont recommend adding it to your common.css as it could impact your use on mobile. The vh directive may not work with older browsers. — xaosfluxTalk22:38, 31 October 2018 (UTC)
Hey, I'm personally very impressed with ((Maplink)) and the Kartographer extension. I would like to know the views of Wikipedia community on usage of the same in Infobox templates like
((Infobox monument)), ((Infobox country)), ((Infobox settlement)). Are there any policies, RFCs, etc I should be aware of? Also are the Infoboxes and Maplink compatible with each other and whats the easiest way to use one in other? Thanks in advance Capankajsmilyo(Talk | Infobox assistance)08:19, 31 October 2018 (UTC)
Maplink is a wonderful tool and it's only used in less than 3,000 articles so most people have probably never even seen it before despite being a few years old. The best way to get the word out is use it as widely as possible. Why not use it in infoboxes see Template:Maplink#Infoboxes. -- GreenC16:29, 1 November 2018 (UTC)
The Editing team has begun a design study of visual editing on the mobile website. New editors have trouble doing basic tasks on a smartphone, such as adding links to Wikipedia articles. You can read the report.
The Editing team wants to improve visual editing on the mobile website. Please read their ideas and tell the team what you think would help editors who use the mobile site.
If you aren't reading this in your preferred language, then please help us with translations! Subscribe to the Translators mailing list or contact us directly. We will notify you when the next issue is ready for translation. Thank you!
Can anyone explain why this version renders differently from this version (specifically, in the link to "2018 Korea Open" in the first line of the infobox) when the only difference between the two is the arrangement of whitespace between the template parameters? --R'n'B (call me Russ) 16:11, 1 November 2018 (UTC)
The difference in rendering is the target of the wikilink. It was caused by the newline you removed at the end of the second unnamed parameter in [34]. Help:Template#Parameters says: "Whitespace characters (spaces, tabs, returns) are stripped from the beginnings and ends of named parameter names and values ... This does not apply to unnamed parameters, where all whitespace characters are preserved."
((Infobox tennis tournament event)) says: [[((#ifexist:(({1))} (({2))} (tennis)|(({1))} (({2))} (tennis)|... (({1))} (({2))}]]. The idea for 2018 Korea Open – Singles is to examine whether 2018 Korea Open (tennis) exists. If it exists then link it instead of linking 2018 Korea Open. But if the unnamed parameter (({1))} or (({2))} contains a newline then ifexist fails to get a match. It's OK if they contain spaces because ifexist automatically reduces consecutive spaces to a single space when the match is made, and the generated wikilink does the same. The general solution if you want newlines in the unnamed parameters to work is to add a bunch of ((trim)) or similar in ((Infobox tennis tournament event)) to strip the newlines. It has to be done both in ifexist and when the wikilink is generated. PrimeHunter (talk) 17:04, 1 November 2018 (UTC)
Yes. Later editors might change it but that is a risk with any fix made in the call and not the template. I think the common practice for this template is to end the line with a pipe and place named parameters on new lines. PrimeHunter (talk) 17:50, 1 November 2018 (UTC)
Have you had any trouble using XTools?
Have you seen this error?
Hi, I don't know of a better place to ask this question. I'm trying to find out if users of XTools have seen the "XTools is currently overloaded" error a lot recently? I see 1,000+ instances of it in the logs for the past 24 hours alone, but we have not received any complaints from users. Of course, many users don't report these things.
To be clear, I'm talking about the "XTools is currently overloaded" error, specifically, not any ole error that shows the bunny rabbit. If it is true that it's happening a lot, I can see to it that it gets fixed, but I need some confirmation first.
Well, the nature of this error would mean everyone else sees it at that point in time as well. I don't really know what's going on, but so long as y'all aren't seeing it, I guess it's fine... :) Thanks again — MusikAnimaltalk19:53, 1 November 2018 (UTC)
Advisor not working
Hey, somehow "Advisor" is not working anymore for me since a few days/weeks. Nothing was changed in my preferences or scripts (I removed and added it back but no success). Any ideas? Kante4 (talk) 15:16, 1 November 2018 (UTC)
That script is very old and the account hosting it (Cameltrader) had not edited since 2008. The script in question was recently hijacked and was therefor deleted (by me) as a countermeasure. The security team is aware and has made sure that affected accounts have been notified and secured. There will be some public communication on this soon. —TheDJ (talk • contribs) 15:11, 1 November 2018 (UTC)
Kante4, I'm currently in the process of figuring out what will happen to the script and all of its transclusions by various users. Check back in a couple days please ? —TheDJ (talk • contribs) 15:22, 1 November 2018 (UTC)
Ditto the ping me. I only rarely get to use the script any more since WTE2017 doesn't support it, so I'd only bump into it when I somehow got dropped into the 2010 editor. I mostly used it for regularizing headings, removing trailing whitespace, and [[A|a]] changes, since I didn't trust the "turn everything into Unicode" changes as desirable changes. --Izno (talk) 03:00, 2 November 2018 (UTC)
No, I'm not sure what what you mean but you already did it correctly. I just pointed out that you used the term "my userpage" in a confusing way. PrimeHunter (talk) 17:55, 1 November 2018 (UTC)
@TheRedBox: normally loading someone else's userscript to your account requires you trust they won't break the script in the future, however as this is a popular script it has had additional page protection applied that only allows it to be edited by interface administrators who are highly trusted in managing scripts. The current version (Special:PermaLink/844587740) of this page appears "safe" upon review. — xaosfluxTalk18:40, 1 November 2018 (UTC)
Hi!
Where may I ask for a medium large external link fixing issue? At Wikipedia:Link_rot I did not find any page for that. User:MerlLinkBot seems to be offline for years now.
I could use my bot (de:user:CamelBot), but it's trained for dewiki and their templates and so would need some adaptions.
Can someone tell me about the input box on Special:ExpandTemplates labeled: Context title, for ((FULLPAGENAME)), etc.: ? I have no idea what it is intended for and the instructions for that page are a bit lacking; to say the least. In my opinion, a page like Help:ExpandTemplates would be a wonderful compliment to that special page.--John Cline (talk) 18:14, 2 November 2018 (UTC)
The entered code will behave like it was placed on the page given in the Context title field. Many templates and other features behave differently in different namespaces and sometimes on different page names in that namespace. The example ((FULLPAGENAME)) returns the name of the current page. Here it returns: Wikipedia:Village pump (technical)/Archive 170. See more at mw:Help:Magic words#Page names. PrimeHunter (talk) 18:38, 2 November 2018 (UTC)
Thank you for that information PrimeHunter. It is valuable! I'd like to ask another question; now that I have new ways to "play" with that special page.
Because I have heard about "cost" in terms of all that is involved to render a template's output, and seen things about "truncated diffs for performance reasons" and "expensive" automation, I wonder if there is any such "cost factor" that I should be cautiously aware of before I engage in any, so called, "playing" on that page? I've been using the page as if "cost" is not a concern, but want to be more informed. Thank you.--John Cline (talk) 19:49, 2 November 2018 (UTC)
I was looking at Takotsubo cardiomyopathy after seeing an article about it on another site and wondered why, if I got there because Takotsubo is a redirect, someone put a link to the Japanese meaning of the word in the article. I clicked on it and got redirected to the same article I was reading. I didn't know what the best way to fix this was, but when I edited, I saw the redirect was supposed to be an interlanguage link. So why did it go right back to English article?— Vchimpanzee • talk • contributions • 20:17, 1 November 2018 (UTC)
@Vchimpanzee: sorry, think we need a little more details to help you here. Can you give step by step what page you started on, exactly what you clicked on, what you ended up with, and what you expected to end up with? — xaosfluxTalk20:27, 1 November 2018 (UTC)
Okay, thanks. I didn't know that ja link was there when I first discovered this. But do you really want the English language link to redirect to this same article?— Vchimpanzee • talk • contributions • 15:39, 2 November 2018 (UTC)
Yes, since this is the English Wikipedia, if someone goes to that link here or searches for it here we want them on our article, not on another site - for the most part articles on one Wikipedia should not depend on articles on another. — xaosfluxTalk19:16, 2 November 2018 (UTC)
Automatically capitalize first alphabetical character of article-title
Right now, the first character of an article title is capitalized. So "bromomethane" is automatically handled as "Bromomethane" and an editor can link to it either way. But if the first character is a number (common in science) or punctuation (common in various proper-noun titles), this does not happen, even though MOS for sentence capitalization says it should. As a result, we wind up having a ton of manually created redirects so that, for example, both 2-bromobutane and 2-Bromobutane can be linked.
This would also improve searchability because some automatically generated lists are sorted by ASCII-string order rather than lexigraphically. Unless we strictly and manually follow a standard of capitalizing or manually adding sort-keys, we get orders like "2-Bromobutane 2-Chlorobutane 2-bromopentane". I couldn't find a phab ticket or previous VPT item about this, but my search-fu is weak here:/ DMacks (talk) 08:31, 3 November 2018 (UTC)
I don't know which MOS you refer to but we don't capitalize words after a year. See e.g. Special:PrefixIndex/2016 in or Special:PrefixIndex/2016 shooting. Special:PrefixIndex/- is mostly suffixes we don't capitalize. I tried some random digit-letter combinations like 5a/5A or 3e/3E, and it isn't even clear whether capitalization is more common than not. I think rules for automation would be too inflexible, complicated and confusing. The search box automatically matches any capitalization. Wikilinks can be handled with redirects as now. PrimeHunter (talk) 09:37, 3 November 2018 (UTC)
Wikipedia:Article titles#Article title format says article titles should be in sentence-case, which is consistent with Wikipedia:Naming conventions (chemistry)#Prefixes in titles (my title-genre of interest, where there are probably thousands of articles that begin with numbers and/or parenthesis). It relates more to discoverability rather than searching. If someone forgets to create a redir, it's a redlink. If someone is scanning by eye the contents of a category (for example, to see "what else do we know", not "do we know X specifically?"), one might not remember to look at other capitalizations because things are out-of-order (and we don't categorize redirects). On en.wp in chemistry, the redirs seem to usually exist. But on commons and other places where discovery is a key idea (and for commonscat linking from en.wp) but where editors are less disciplined, it makes a lot of redlinks and duplication of work. DMacks (talk) 10:20, 3 November 2018 (UTC)
We are a general encyclopedia, not a chemistry encyclopedia. A chemistry convention shouldn't be forced on all other articles by the software. You could make a request at phab: for a new MediaWiki configuration setting similar to mw:Manual:$wgCapitalLinks. But it should be disabled by default and I wouldn't support enabling it at the English Wikipedia. PrimeHunter (talk) 10:57, 3 November 2018 (UTC)
The MediaWiki software is working as designed: it capitalises the first character of the page title. Characters include letters, but they also include digits, punctuation, a number of other symbols, marks and spaces. Simply because capitalising a digit is a null action, this does not mean that the capitalisation should proceed to the next non-digit character. Having an article named 2-Bromobutane is not a crime, and you may also have a redirect to that page from 2-bromobutane without controversy. --Redrose64 🌹 (talk) 13:18, 3 November 2018 (UTC)
Search on en.wikipedia brings results from the Italian Wikipedia
Well, this never happened to me before. Is this a new feature? In the en.wikipedia search box I typed the name "Paconio" (which, by the way, is from the Chumash people) There is no article on English Wikipedia for this name, which it does tell me. Below that, it give me the search results for the Italian Wikipedia. 1 I thought it was a new bonus feature for us, so I randomly tried Italian names, but there are either articles are dabs for the names I can think of. But I can still get it to do its Italian thing by searching for Paconio . As Arte Johnson would say, "Verrrry interesting...", — Maile (talk) 00:49, 3 November 2018 (UTC)
How does the U.S. time change affect your preference to see your own time?
I've noticed on the Help Desk and Teahouse that some people choose to see their local time on this site. I guess that means signatures and histories. What happens if someone contributes during that hour which gets repeated this time of year? I almost forgot to add "U.S." but a site where I go has British posters who have already had a time change.— Vchimpanzee • talk • contributions • 16:04, 3 November 2018 (UTC)
It just means histories and watchlist and similar other system-controlled times. Signatures do not change. As for local times, it requires user action to shift the time back to the correct hour; it is not automatic. (There probably should be a task to improve that to automatic, but I guess that would be quite low priority.) --Izno (talk) 16:20, 3 November 2018 (UTC)
@Vchimpanzee: There are two distinct ways in which times are used in Wikipedia.
Page histories, contributions, watchlist, recent changes etc. read the times directly from the data tables on the servers, which are held in Unix time, relative to 00:00, 1 January 1970 (UTC). When a page is served, these times are converted from Unix time (which is a 32-bit integer) to a human-readable form based upon the "Date format" and "Time offset" settings at Preferences → Appearance.
The signatures in discussions such as this one are stored as plain text, with the times in UTC; the settings mentioned in the previous paragraph are ignored. There is a gadget ("(U) Change UTC-based times and dates, such as those used in signatures, to be relative to local time (documentation)") which can reformat these for display, but this does cause problems if somebody using this gadget copies a timestamp from outside an edit box and and pastes it into an edit box.
To add on to that, just think of summertime changes being the same as changing any other timezone, if you move several degrees east or west the time zone may change - but that is just what you call the time "locally". — xaosfluxTalk20:10, 3 November 2018 (UTC)
A lot of what was said above is incorrect. In Special:Preferences, you can set the timezone based on either offset from UTC (as I have) or a location. If you set it by offset from UTC, daylight saving will never apply and you will have to change the offset manually. If you set it by location, the times will display in page histories/watchlists/RecentChanges with daylight saving applied, if it was active during the time in question. For the purposes of this message, I changed the settings from the offset to my location of Perth, the capital of Western Australia, which normally has an offset of UTC+8 but had a three-year experiment with daylight saving from late-2006 to early-2009. During the time we had daylight saving, when the time offset was UTC+9, I created User:Graham87/sandbox10 to test what would happen when various numbers of tildes were entered. The edit times in the page history show for me as 15:50/15:51, 26 February 2009 with just an offset but 16:50/16:51 with the location set to Perth; the API shows that the edits were made at 07:50/51 (UTC). Also, MediaWiki doesn't store its timestamps in Unix time; the storage format for timestamps is described in the user guide on the MediaWiki wiki. Graham8707:06, 4 November 2018 (UTC)
I sometimes get this message when doing a page move:
Internal error
From Wikipedia, the free encyclopedia
Jump to navigationJump to search
[W957FwpAIDAAAG-f9a0AAAAV] 2018-11-04 04:52:40: Fatal exception of type "MWException"
Going back to the page that I was trying to move, I see that the page move did take place anyway, so that the error is after moving the page. What causes this error? Is there anything that I can do to prevent it (other than not moving pages)? Robert McClenon (talk) 04:57, 4 November 2018 (UTC)
The error means that for some reason, MediaWiki could not create a dummy entry for the page move in the history of the page. I have no idea why.
MWException from line 540 of /srv/mediawiki/php-1.33.0-wmf.1/includes/MovePage.php: Failed to create null revision while moving page ID 16153763 to Neil_Giraldo_(0)
#0 /srv/mediawiki/php-1.33.0-wmf.1/includes/MovePage.php(271): MovePage->moveToInternal(User, Title, string, boolean, array)
#1 /srv/mediawiki/php-1.33.0-wmf.1/includes/specials/SpecialMovepage.php(603): MovePage->move(User, string, boolean)
#2 /srv/mediawiki/php-1.33.0-wmf.1/includes/specials/SpecialMovepage.php(128): MovePageForm->doSubmit()
#3 /srv/mediawiki/php-1.33.0-wmf.1/includes/specialpage/SpecialPage.php(569): MovePageForm->execute(NULL)
#4 /srv/mediawiki/php-1.33.0-wmf.1/includes/specialpage/SpecialPageFactory.php(568): SpecialPage->run(NULL)
#5 /srv/mediawiki/php-1.33.0-wmf.1/includes/MediaWiki.php(288): MediaWiki\Special\SpecialPageFactory->executePath(Title, RequestContext)
#6 /srv/mediawiki/php-1.33.0-wmf.1/includes/MediaWiki.php(860): MediaWiki->performRequest()
#7 /srv/mediawiki/php-1.33.0-wmf.1/includes/MediaWiki.php(517): MediaWiki->main()
#8 /srv/mediawiki/php-1.33.0-wmf.1/index.php(42): MediaWiki->run()
#9 /srv/mediawiki/w/index.php(3): include(string)
#10 {main}
User:Bwolff - Thank you. It happens maybe less than one time out of ten, but often enough that I recognize it, so I would say it seems to be timing-dependent. It doesn't interfere with my moving a sandbox to draft space. Robert McClenon (talk) 02:04, 5 November 2018 (UTC)
Non-indexed articles showing up in Google Knowledge Panels
Search results on 4 Nov 2018 for the word Aspidomancy
While doing NPP at Aspidomancy, an article that had not been reviewed or patrolled, I did a google search. Imagine my surprise when the page showed up in a knowledge panel. It does not appear to show up in the main search results only the knowledge panel. This does not seem to be something on our end, but still felt worth raising in the community. Best, Barkeep49 (talk) 01:47, 5 November 2018 (UTC)
Barkeep49, Google doesn't obey our command. They have often stated that no matter what you tell them (for instance: NOINDEX), their priority is on indexing information that people use. So as soon as a website has linked to a page of ours, they will ignore what they consider to be our 'advice' and index it anyways. We have no clear concept of how their knowledge panels work, so influencing it is very difficult. —TheDJ (talk • contribs) 13:02, 5 November 2018 (UTC)
Gap caused by templatestyles
Is there a phab issue for the gap between navboxes that use templatestyles? in particular, see the two navboxes at the foot of Fort Simpson. As far as I can tell, there is no way to remove the spacing between the two boxes without changing Module:Navbox Canada to not use templatestyles. thank you. Frietjes (talk) 15:25, 4 November 2018 (UTC)
WOSlinker, I thought about that, but at one point in time, I was told that the templatestyles must come before the first use of the class. I don't know if that is true. Frietjes (talk) 17:44, 4 November 2018 (UTC)
Izno, having ((navbox)) take a |templatestyles= arg would also fix the problem, since navbox could insert it in a place that doesn't cause a problem (e.g., inside the first div tag). Frietjes (talk) 17:45, 4 November 2018 (UTC)
@Izno and WOSlinker:, okay, I see, the 1px spacing between the navboxes comes from the '.navbox + .navbox' rule in MediaWiki:Common.css. so the templatestyles are causing the two navboxes to appear "non-consecutive". putting the templatestyles inside the outer navbox div, or in the title looks like the only simple solution. Frietjes (talk) 18:23, 4 November 2018 (UTC)
thank you, very useful. for now, I have implemented WOSlinker's fix, which is probably nearly optimal given the comments about navboxes being removed from mobile view. better would probably be to pass the templatestyles in to Module:navbox and have navbox put them inside the outer div, but this pretty close. Frietjes (talk) 14:38, 5 November 2018 (UTC)
Questions on Module:US House seniority box
I've been working off-and-on on a module that will automatically generate the "United States Representatives by seniority" navboxes found on US House members automatically. Right now, every time a member resigns, the ordinals in these tables need to all be updated (by AWB, normally). Using a module to generate these should allow only the data file to be changed when a member resigns.
I've come across a few issues:
LUA makes it very difficult to have indexed lists. In Python, it's normal to have x[2] or x[12], but expecting the members of a LUA table to be in a defined order seems to be working against the language (there's no way to even get the "length" of a table). Explicitly writing 1->Don Young, 2->Jim Sensenbrenner in the data file is something I'd like to avoid. Is there a different pattern that should be used?
Should the template rendering be in the Module (like I've recently added code for), or in an associated Template? It seems far easier to handle edge cases (like "no member with more seniority") in a programming language than trying to do conditional logic in WikiText.
Should the raw data be at Module:US House seniority box/data, or somewhere else? In my ideal world, we'd simply import a JSON file with this information (Wikipedia page name, display name, party, seniority order) but that doesn't seem to be an option. I don't see any namespace better than the Module namespace for this. Wikidata doesn't have this data (and is unlikely to have seniority order), so that's not an option even if it were possible.
Should this be more generalizable? A way to find a page's position in a list is not a terribly odd thing to want to generate automatically; most of the time these are done by hand (and updated by AWB) or are lists that don't change over time.
Questions about modules are normally at WT:LUA. Lua's tables are extremely convenient and I now find that going back to Python is a pain. It's true that Lua's tables have some irritating issues but there is always a way to organize the data to fit in with how Lua works—the result will be something that executes faster and with a smaller memory footprint than Python. x[2] works fine in Lua and with the current Module:US House seniority box/data that would give "Jim Sensenbrenner" (the second entry). Yes, all the code should be in the module with the template merely invoking the module, and the raw data should be in the data module. That should return a table with your current data as one of its members because it is inevitable that some future expansion will have another table of data in the data module. There probably will be JSON available at some time—I think experiments with that are occurring at Commons?—but a data module is good. Each entry should be a table with the data you mentioned (page title, name, party, seniority). I don't understand about the order/position issue but would have suggestions if an example were available. Johnuniq (talk) 04:29, 5 November 2018 (UTC)
The count of the integer keys in a Lua table is #table_name. If you're mixing integer and non-integer keys (which it does not presently look like you are doing), you're probably not storing all of your stuff sensibly. --Izno (talk) 13:11, 5 November 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Recent changes
You can now use TemplateWizard to edit templates. This works only with the 2010 wikitext editor and not in the visual editor or the 2017 wikitext editor. If you click on you can enter the information in a pop-up. You can turn on TemplateWizard in your beta feature preferences. [35]
Changes later this week
You can choose to see edit conflicts in a two-column view. This is a beta feature. You can find it in your preferences. The interface for the two-column edit conflict will change. You can read more.
When you edit with the visual editor you can use the "Automatic" citation tab. This helps you generate citations. You will now be able to write plain text citations or the title of a journal article or a book in this tab. This will search the Crossref and WorldCat databases and add the top result. [36]
The new version of MediaWiki will be on test wikis and MediaWiki.org from 6 November. It will be on non-Wikipedia wikis and some Wikipedias from 7 November. It will be on all wikis from 8 November (calendar).
Meetings
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 7 November at 16:00 (UTC). See how to join.
Is there a "sure-enough-way" to show "source code" on the surface of a page (like when presenting an edit request) where you can be absolutely confident that the code rendered for display is a true and faithful representation of the source code you meant to show? Until today, I had thought <nowiki>...</nowiki> was the "go-to" markup for this. I now know that this is not the case and I am thoroughly confused. So how would this best be done? Thank you.--John Cline (talk) 02:41, 5 November 2018 (UTC)
No Masem, that doesn't appear correct. Consider the following edit request (which is the actual edit request I posted yesterday) where I am now using <code>...</code> instead:
Addendum - <code>...</code> was replaced with <syntaxhighlight>...</syntaxhighlight> for the example below as of this timestamp.--John Cline (talk) 08:19, 5 November 2018 (UTC)
Please change the following code:
((#if: (({concern|(({reason|(({1|))))))))}
|<nowiki> </nowiki><!-- because of the following concern: -->
|.
))
<blockquote>(({concern|(({reason|(({1|))))))))}</blockquote>
<p class="verbose">~~~~
To instead be:
((#if: (({concern|(({reason|(({1|))))))))}
|<nowiki> </nowiki><!-- because of the following concern: --><blockquote>(({concern|(({reason|(({1|))))))))}</blockquote>
|.
))
<p class="verbose">~~~~
Addendum - <code>...</code> was replaced with <syntaxhighlight>...</syntaxhighlight> for the example above as of this timestamp.--John Cline (talk) 08:19, 5 November 2018 (UTC)
<syntaxhighlight>...</syntaxhighlight> or <pre>...</pre> can do the job; this one is <syntaxhighlight lang="moin">...</syntaxhighlight>:
((#if: (({concern|(({reason|(({1|))))))))}
| because of the following concern:<blockquote>(({concern|(({reason|(({1|))))))))}</blockquote>
|.
))
<p class="verbose">
<syntaxhighlight>...</syntaxhighlight> is great, <pre>...</pre>, in and of itself, is not able to get the job done. I was using it in this example which can be seen to have failed,--John Cline (talk) 08:19, 5 November 2018 (UTC)
It appears from the following tests that only pre will give you a true and faithful representation of the source code you meant to show.
Four leading spaces:
WP:1AM
| .--.
| ______.-------| |
| __ (_____( | |\\\\|
| __..-- ``--.._ __/ `-------| |---,
| __ ``--..____.--'| \ ___ | | ||
| __..-- ``--.._ | | | | | | | ||
| ``--..___| | | |___| | | ||
The plug is pulled. `--.|_/ | | ||
Ignored is the disruptive one. ____\ .-------| |---`
Feed him I will not. (_____( | |\\\\|
| `-------| |
| `--`
[37]http://www.example.com
ampersand is &
blockquote:
WP:1AM
| .--.
| ______.-------| |
| __ (_____( | |\\\\|
| __..-- ``--.._ __/ `-------| |---,
| __ ``--..____.--'| \ ___ | | ||
| __..-- ``--.._ | | | | | | | ||
| ``--..___| | | |___| | | ||
The plug is pulled. `--.|_/ | | ||
Ignored is the disruptive one. ____\ .-------| |---`
Feed him I will not. (_____( | |\\\\|
| `-------| |
| `--`
[38]http://www.example.com
ampersand is &
code:
WP:1AM
| .--.
| ______.-------| |
| __ (_____( | |\\\\|
| __..-- ``--.._ __/ `-------| |---,
| __ ``--..____.--'| \ ___ | | ||
| __..-- ``--.._ | | | | | | | ||
| ``--..___| | | |___| | | ||
The plug is pulled. `--.|_/ | | ||
Ignored is the disruptive one. ____\ .-------| |---`
Feed him I will not. (_____( | |\\\\|
| `-------| |
| `--`
[39]http://www.example.com
ampersand is &
kbd:
WP:1AM
| .--.
| ______.-------| |
| __ (_____( | |\\\\|
| __..-- ``--.._ __/ `-------| |---,
| __ ``--..____.--'| \ ___ | | ||
| __..-- ``--.._ | | | | | | | ||
| ``--..___| | | |___| | | ||
The plug is pulled. `--.|_/ | | ||
Ignored is the disruptive one. ____\ .-------| |---`
Feed him I will not. (_____( | |\\\\|
| `-------| |
| `--`
[40]http://www.example.com
ampersand is &
nowiki:
[[WP:1AM]]
| .--.
| ______.-------| |
| __ (_____( | |\\\\|
| __..--'' ``--.._ __/ `-------| |---,
| __ ``--..____.--'| \ ___ | | ||
| __..--'' ``--.._ | | | | | | | ||
| ``--..___| | | |___| | | ||
The plug is pulled. `--.|_/ | | ||
Ignored is the disruptive one. ____\ .-------| |---`
Feed him I will not. (_____( | |\\\\|
| `-------| |
| `--`
[http://www.example.com]
http://www.example.com
ampersand is &
poem:
WP:1AM
| .--.
| ______.-------| |
| __ (_____( | |\\\\|
| __..-- ``--.._ __/ `-------| |---,
| __ ``--..____.--'| \ ___ | | ||
| __..-- ``--.._ | | | | | | | ||
| ``--..___| | | |___| | | ||
The plug is pulled. `--.|_/ | | ||
Ignored is the disruptive one. ____\ .-------| |---`
Feed him I will not. (_____( | |\\\\|
| `-------| |
| `--` [1] http://www.example.com
ampersand is &
pre:
[[WP:1AM]]
| .--.
| ______.-------| |
| __ (_____( | |\\\\|
| __..--'' ``--.._ __/ `-------| |---,
| __ ``--..____.--'| \ ___ | | ||
| __..--'' ``--.._ | | | | | | | ||
| ``--..___| | | |___| | | ||
The plug is pulled. `--.|_/ | | ||
Ignored is the disruptive one. ____\ .-------| |---`
Feed him I will not. (_____( | |\\\\|
| `-------| |
| `--`
[http://www.example.com]
http://www.example.com
ampersand is &
samp:
WP:1AM
| .--.
| ______.-------| |
| __ (_____( | |\\\\|
| __..-- ``--.._ __/ `-------| |---,
| __ ``--..____.--'| \ ___ | | ||
| __..-- ``--.._ | | | | | | | ||
| ``--..___| | | |___| | | ||
The plug is pulled. `--.|_/ | | ||
Ignored is the disruptive one. ____\ .-------| |---`
Feed him I will not. (_____( | |\\\\|
| `-------| |
| `--`
[41]http://www.example.com
ampersand is &
tt:
WP:1AM
| .--.
| ______.-------| |
| __ (_____( | |\\\\|
| __..-- ``--.._ __/ `-------| |---,
| __ ``--..____.--'| \ ___ | | ||
| __..-- ``--.._ | | | | | | | ||
| ``--..___| | | |___| | | ||
The plug is pulled. `--.|_/ | | ||
Ignored is the disruptive one. ____\ .-------| |---`
Feed him I will not. (_____( | |\\\\|
| `-------| |
| `--`
[42]http://www.example.com
ampersand is &
I use pre a fair bit but bear in mind that it does not do anything magic to hide an html entity. For example, if "ampersand is &" is inside pre, you would see "ampersand is &". That is, you need to replace any literal ampersands with & if the ampersand and what follows could be interpreted as an entity. Johnuniq (talk) 04:35, 5 November 2018 (UTC)
I added that line to all of the examples above, and none of them display htlm entities as typed.
Let me try this: &
OK, a Zero-width space between each character also works, but of course doesn't meet the "display what I typed" requirement. I also tried a bunch of things like combining pre and nowiki, but no luck. :( --Guy Macon (talk) 05:34, 5 November 2018 (UTC)
I appreciate the replies. It looks like <syntaxhighlight>...</syntaxhighlight> is clearly the best way to go with this. I also tested it with simulations that included ~~~~, <nowiki>...</nowiki>, and <!-- --> and they all rendered just as they exist in the source code. My confidence is restored. I also tested the element without any < class="" or style=";"> modifications and non-enhanced element rendered the syntax just as if <pre>...</pre> was being used which is the effect I was hoping to achieve. I replaced the <code>...</code> elements above, with <syntaxhighlight>...</syntaxhighlight> and simulated using the other markup I mentioned moments ago; you can see the results, and that it worked very well. Therefore, my best to Trappist the monk for sharing that information, and also a big Thanks too.--John Cline (talk) 08:19, 5 November 2018 (UTC)
@John Cline: If your intent is to describe a proposed amendment to a template, the best thing to do is to use the template's sandbox. If the template has documentation, the sandbox is linked near the bottom of that documentation; otherwise, simply append /sandbox to the name of the real template. Copy all of the code from the live template to the sandbox, and save. Then amend the sandbox. In this way, people can see exactly what changes you want to make; and can also test those changes before putting them live. More at WP:TESTCASES. --Redrose64 🌹 (talk) 21:09, 5 November 2018 (UTC)
Syntactic conundrum
Warm greetings to all of you.
Would by any chance one of your devoted experts be in position to solve a baffling mystery?
I wouldn't use that either but it wasn't the cause here. A named reference which was displayed right before had ]] instead of )).[43] The fix reveals another error: The English Wikipedia has no Template:Citation étrangère. PrimeHunter (talk) 22:14, 5 November 2018 (UTC)
@Izno and PrimeHunter. Thank you so much for your prompt replies. It therefore seems that the fundamental issue which, before your judicious adjustments, had previously induced this "syntactic chaos"—by catapulting most of the existing references towards the Bermuda Triangle—was specifically due to these two irrelevant and accidental brackets → ]], i.e.((BNF|...]] instead of ((BNF|...)). I feel so grateful for your your timely updates as well as for your clear and detailed explanations together with your friendly and benevolent help. With kind regards. Sincerely, — euphoniebreviary00:44, 00:48, 00:50, 6 November 2018 (UTC)
@Izno The Noise (electronic) article used to have a French interwiki link to Bruit de fond. Now it has an interwiki link to Bruit, which does not appear to me to be the same or to have the same content. On en.wiki there are multiple articles on noise. So disambiguation is an issue. Right now there are editors making decisions on content, open-loop without discussion. --Ancheta Wis (talk | contribs)12:14, 6 November 2018 (UTC)
@Ancheta Wis: You should request help on d:WD:PC; there is a French page there if you prefer French or you can speak in English at the English page. There are a number of French contributors at Wikidata who should be able to help you. --Izno (talk) 20:28, 6 November 2018 (UTC)
Comment table and edit summaries
As of October 25 this year, we don't store edit summaries in the rev_comment column of the revision table anymore. Instead, we use the "comment" and "revision_comment_temp" tables. As a result, Sigma's edit summary search tool fails to display or detect edit summaries correctly for new edits made since October 25 this year, or old edits from April 2017 and earlier. Eventually, all edits will fail to display or detect edit summaries correctly. Is there any way we can update that tool to use the new tables? GeoffreyT2000 (talk) 03:11, 6 November 2018 (UTC)
Note also backfilling of the new comment table is still a month or so from completion. So right now, to be fully comprehensive, you need to query revision and comment. — MusikAnimaltalk01:05, 7 November 2018 (UTC)
Category has negative contents
This is a screenshot of Category:Wikipedia categories that should not contain articles. Note the contents of one of its subcategories, Category:Wikipedia soft redirected categories: 72,403 C, -6 P. How can a category have a negative number of contents? Isn't that like someone having a height of -5 feet? I'm pretty sure the number of contents is a scalar quantity, not a vector :-) I just checked another of this category's parents, Category:Wikipedia categories, and the soft redirected categories still have -6 pages. If you enter the soft redirected categories category and scroll all the way down, it looks just like a category with no pages. At first I thought this might be because of the number of subcategories (with 200+ subcategories that precede the first page alphabetically, would I have to navigate until I found a page in its proper alphabetical place), but Category:Wikipedia non-diffusing parent categories has two pages that appear on the first page of results, even though they come far after the 200th subcategory. Nyttend (talk) 03:32, 7 November 2018 (UTC)
At MV Hebridean Princess I replaced some fraction characters in the infobox (under "Installed power") with ((frac)) per MOS. Unfortunately ((frac|10|1|2)) (10+1⁄2) line-breaks after the "10". Looking at the page source I see no trace of the that the template explicitly adds, so the only thing preventing the break is the nowrap class - which, in order to alleviate excessive use making things look bad on narrow screens, is disabled on mobile. <span style="white-space:nowrap">...</span> doesn't help. What's going on? I don't understand why the nbsp is disappearing. Hairy Dude (talk) 15:03, 1 November 2018 (UTC)
The nbsp is hidden offscreen by design, but that also makes it useless, as the character is not adjacent to other characters that way. The nowrap class is indeed currently disabled in infoboxes, because of the reason you mentioned. —TheDJ (talk • contribs) 15:17, 1 November 2018 (UTC)
Isn't the nbsp in there so that environments lacking CSS (including copy-paste from the browser) gets "10 1⁄2" rather than "101⁄2"? And in that case, at least Firefox does avoid breaking at the non-breaking space despite it being inside tags. Anomie⚔21:18, 1 November 2018 (UTC)
Can this issue be fixed? This is yet another instance when mobile users feel like second class citizens because basic expectations like no-break spaces not breaking don't hold because a handful of people behave badly in a way that's easily fixed with a single edit. Hairy Dude (talk) 11:58, 7 November 2018 (UTC)
Hairy Dude, I have no real solution here for you. You want me to remove the line that disables nowrap on Mobile infoboxes, so that the infoboxes break again ? I can do that, but then I need to deal with complaints about that again. No matter which way I jump, people seem to be/get upset, making it a not so appetising topic to engage on. —TheDJ (talk • contribs) 12:41, 7 November 2018 (UTC)
Fair enough, but just let me say that I think a technical "fix" breaking basic functionality to solve a problem that's fundamentally social in nature is a very poor compromise. Hairy Dude (talk) 13:01, 7 November 2018 (UTC)
Remove the theme color meta tags!!
I noticed this change recently. A meta tag for theme-color was added which makes the Android Chrome browser change the toolbar color of the OS.
First off, a website changing UI colors is a serious violation of disability accessibility standards. Websites run in a container and should never modify UI objects outside the browser. People with visual disabilities are impacted by color changes like this.
Second, the color chosen (light gray) causes severe eye strain in people sensitive to it. I am one of those people. This renders Wikipedia unusable by people with this disability.
Google enabled this awful "feature" with no way to disable it, so it is encumbant on websites to not use it.
Please remove this meta tag from Wikipedia. A website should not be changing UI colors outside the browser, and the current implimentation on Wikipedia makes it literally painful to use. — Preceding unsigned comment added by 108.2.72.57 (talk) 21:57, 3 November 2018 (UTC)
Heya IP, sorry you're frustrated. There was a recent change (and discussion) to account for the lack of clarity between the updated status bar on Android and the web content. The design and engineering folks put a change into effect that made the status bar consistent and distinguishable from the actual content. I've raised your concern of accessibility with the team to have them review in light of your considerations. CKoerner (WMF) (talk) 15:46, 7 November 2018 (UTC)
I agree that the color used should be chosen so that it meet the WCAG contrast requirements, but in this case, #EAECF0 has a 17.75 contrast ratio against black. That far exceeds the WCAG requirement of 4.5 for AA or 7 for AAA. In addition, saying that websites run in a container and should never modify the UI is a bit rediculous when this is a docuemented feature explicitly built in to the android operating system and the chrome browser. If this is an accessibility issue, it should be brought up with the Chrome or Android development team. If you want to disable a documented browser feature, I'd recommend switching to a different browser if this is an issue for you (for example, I'm not sure if Firefox for Android supports this tag, but if it does and doesn't have an option to disable it, you could install a greasemonkey script to strip it off). --Ahecht (TALK PAGE) 16:05, 7 November 2018 (UTC)
Some category header templates now include automated links to portals, passing them to ((Portal)). They only check whether the portals exist; they don't check whether the portals are ready for use, and some are not, e.g. Portal:1910s has been started but is tagged with ((Portal maintenance status|date=October 2018|incomplete=y|subpages=checked|broken=major)).
Yes. Modify Module:Portal to read the unparsed content of the portal page and look for ((Portal maintenance status)). You can then have Module:portal take appropriate action whatever that might be.
Thank you very much, Trappist the monk. I did not realise there was a module:Portal. Please could you, or someone else reading this, insert a condition there so that the module will skip linking to a portal with status broken=major ? – FayenaticLondon08:25, 3 November 2018 (UTC)
I could, but I'm not sure that I should. I almost never venture into portal space, but your Portal:1910s example doesn't look all that broken to me; readers going there will find something so I guess that I'm not seeing this as a problem. If, as part of whatever construction is undertaken, the portal page displays huge red error messages, has broken html, or whatever, and if that condition will exist for a while, perhaps the best solution is to do portal development in a sandbox and then publish the sandbox version when it is ready for public consumption. This is, after all, why we have sandboxen. No new code to write, no new code to maintain.
Yeah, if a portal is bad enough (for more a week or so while it is under construction) that it shouldn't be linked, then it shouldn't be in portal space at all but in draftspace. Galobtter (pingó mió) 09:43, 3 November 2018 (UTC)
Sandboxing is nice in theory, but some editors have been mass-producing automated portals with zero curation or maintenance. See recent discussions at WT:WPPORT and WT:PORTG#Notability_Discussion:_Revived. Also, their construction techniques use ((PAGENAME)) which probably wouldn't work in sandboxes.
Draftspace isn't set up for portals, and portals don't work in draftspace. That is, their code doesn't work in draftspace, and when moved to draftspace, they become broken. — The Transhumanist00:46, 8 November 2018 (UTC)
You pinged me into this conversation and then removed the ((ping)) so perhaps you don't want my participation here. Regardless, the answer to the question that you asked me is: I don't know. What happens to the portal box is dependent upon what the community determines to be the appropriate action for Module:Portal to take when it encounters |broken=major.
Portals have features that may acquire regular readers who return to portals or regularly click on portal links for their news, DYKs, automatically updating content, and automatically added content. Having links disappear could be disruptive to those readers. If a portal becomes broken and its link disappears, regular readers may not know how to contact the right people to fix it. With a link to a broken portal in place, they would see the problem, and would have the opportunity to go to the portal's talk page to report it, or even collaborate with others to fix it. On the portal's talk page, they would see our link to the WikiProject, and could go there to report the problem as well. With no link, they lose those options. For veterans like us, that's no big deal. But for new Wikipedians, or readers who have limited editing experience, that could leave them almost helpless, scratching their heads. Having links disappear may also have other ramifications that we are not yet aware of. We don't want to hide the problems, we want to fix them, and we want anybody on the scene to be able to do it. A link to a broken portal should also be a route to the instructions on how to fix it, such as having a link to the portal instruction page on every portal's talk page. We need to enable people who use portals to be able to help maintain them, rather than remove the option to do so. — The Transhumanist01:06, 8 November 2018 (UTC)
A portal that becomes slightly broken, may still be helpful to readers, such as if its news feature is still working. Why remove access to that news if the image slideshow all of a sudden becomes empty because images were removed from the root article from where they were displayed? This creates a domino effect. We need to be careful not to fall into an all-or-nothing mindset. A partially functioning portal is still useful, especially to regular visitors of the portal system. — The Transhumanist01:15, 8 November 2018 (UTC)
How difficult would it be to import VisualFileChange over from Commons to en.wiki, so that users who are patrolling images can perform batch actions and avoid spamming user pages with multiple automatically generated Twinkle notifications? GMGtalk12:37, 2 November 2018 (UTC)
Convenience link. The script is stored in commons:MediaWiki:VisualFileChange.js and subpages, so to get it here we'd need a Meta sysop to first import it to Meta wiki and then an enwiki sysop imports it here. The main problems I could see is that the JS would need to be rewritten to accomodate different namespaces, templates etc (for example, on Commons file deletion discussions occur on dedicated subpages). and that there may be dependencies outside of the main JS file. Jo-Jo Eumerus (talk, contributions) 12:48, 2 November 2018 (UTC)
Does that mean we go to phab land I presume? I mean it seems like it could be beneficial to many projects, since it seems like doing this here would facilitate transfers to any other project also (if I understand correctly), and User:ShakespeareFan00 has already started spitballing improvements for our purposes, which if eventually implemented, could also cross pollinate back over to Commons. GMGtalk12:54, 2 November 2018 (UTC)
"Would it be possible to have a multi-image/issue selector tool? That way a "wall" of notications could be be avoided, as the tool would let various issues be selected en-masse, and an image patroller could "batch" up requests, into ONE notifcation, perhaps for some uploaders containing ALL the media with issues that had been identified or flagged.
Alternatively (And possibly combined with the above), perhaps an upload "dashboard" (An extension of Special:ListFiles]] )) for users would be helpful, so that rather than seeing a wall of notifcations on the talk page, you get a 'bell' style notification in the UI, which links to the relevant concern in the image dashboard. This would also allow for the consideration that some people don't read their talk pages often, but do respond to the 'bell' or 'tray icons. (Not all CSD notifications or image tagging operations generate notifications, my custom notifcations)
Importing VisualFileChange from Commons, would be part of the soloution, as VFC can at least do multiple tags per image, or per batch as I understood it..
A more advanced tool than VFC would need to go further though, as VFC works on a 'batch' of images to apply the "same" set of issues. What would also be needed is a tool that can see a batch of images, and if needed apply 'differing' tags (by user selection on a per image basis if needed) in the batch. Some CSD are applied slightly differently if you have different license tags,to give an example. Some Tags need date= tag applied ( or on one of mine placed= tags, and so on. Once a set of images and associated tags are logged, the tool I am describing should generate ONE BIG notification covering all the affected media and issues, rather than a wall of notifcations like TWINKLE does when patrolling a lot of images.
Related Discussion on a "media" dashboard, vs talk page notifcations.
Alternatively (And possibly combined with the above), perhaps an upload "dashboard" (An extension of Special:ListFiles for users would be helpful, so that rather than seeing a wall of notifcations on the talk page, you get a 'bell' style notification in the UI, which links to the relevant concern in the image dashboard. This would also allow for the consideration that some people don't read their talk pages often, but do respond to the 'bell' or 'tray icons. (Not all CSD notifications or image tagging operations generate notifications, my custom notifcations)
Related consideration, When media you uploaded gets tagged for CSD , currently you get a Talk page notification in most instances.. Is there are mechanism for this instead to be a 'notification' instead?
Suggested format " <User X> has edited [[File:]] which you uploaded or restored... The edit is <link to diff> " or " <user X> has tagged [[File: ]] which you uploaded for <CSD criteria>, Please review your upload and take appropriate action.". Moving image issue notifcations from tak pages to a dedicated dshaboard or log, might also increase visibility and response to the notifcations. Other things which are image notifications that could be a 'bell' styule notifcation, might be listing at FFD, pending commons transfer (different from F2, F8 as such) and so on...
Such notifications should ideally be based on what a user has uploaded (without them needing to be explicitly placed in a Watchlist), (and potentially on File: pages contained in a Watchlist for that user.)
ShakespeareFan00 (talk) 12:57, 2 November 2018 (UTC)
Umm...I'd say a good first step would be whether we can find a competent victim volunteer to just get the thing over here. GMGtalk13:22, 2 November 2018 (UTC)
I don't think that you don't need to get the script imported, it's possible to install JavaScript that is hosted on another Wikimedia project. For example, I find User:Anomie/previewtemplatelastmod very useful (thanks Anomie), but it's not set up on Wicipedia Cymraeg, so I import it cross-wiki, see cy:Defnyddiwr:Redrose64/common.js. In this case I think that the required code is
Most characters have an inverted glyph somewhere in Unicode... it's probably some utility that a vandal was playing with. I suspect that this reverses the order of characters in a string and then replaces each character with it's inverted equivalent:
Some (l, s) map to themselves; some (n/u) are easy mappings because the inverted character is a valid Latin letter. The first step is dead easy, we did it as one of the first programming assignments at school; the second step requires a lookup table. --Redrose64 🌹 (talk) 10:42, 7 November 2018 (UTC)
What is that mess? I cannot read the diff, nor this edit window (which has slowed to a crawl); even my watchlist is screwed with little squares all over the edit summaries for about twelve lines. Please don't do it again. --Redrose64 🌹 (talk) 19:10, 8 November 2018 (UTC)
I'm not sure it would be possible to edit filter Zalgo text; because it's made of the standard ASCII character set plus combined diacritics as opposed to specialist and obscure unicode, the diacritics used differ each time, and there are many legitimate uses for diacritics, I can't see how you could write a script that wouldn't hit too many false positives. (This is precisely why vandals—on all user-generated sites, not just Wikipedia—like it so much.) On the plus side, because it screws up the formatting and page history so visibly, it's very easy to spot, so it rarely stays live for long. On upside-down text, it probably wouldn't be possible to filter at all, as all the symbols used have legitimate uses. ‑ Iridescent19:32, 8 November 2018 (UTC)
And now the colors are back. I wonder what's going on. For example, the equals signs to the left and right of the heading (which is larger than the other text) are blue, and so are the Wikilinks and the colon and four tildes.— Vchimpanzee • talk • contributions • 20:16, 5 November 2018 (UTC)
This sounds like slow servers. Several times today I have failed to receive one or another of the CSS or JS pages that should be served to me when retrieving a Wikipedia page; I suspect that Vchimpanzee is experiencing the same problem. --Redrose64 🌹 (talk) 21:10, 5 November 2018 (UTC)
Can a user like me update a wmf lab tool? I am talking about Firefly Tools which is helpful in showing the Linter category count.Adithyak1997 (talk) 18:20, 8 November 2018 (UTC)
Accessing parent template params from a subtemplate
Resolved
Is a subtemplate able to access the parameters of its parent template without them being included in the subtemplate's call string? For example, suppose I call ((x1|a=1|b=2|c=3|d=4|e=5|...|z=26|aa=101|bb=252|...|zz=26926)) and the code of ((x1)) in turn calls ((x2|1=a|2=aa|3=zz)). I'd like for ((x2)) to have access to all of the ((x1))'s parameters, so that, for example, within ((x2)) I could invoke the following:
(({ (({1)))))) → (({a))} → 1
(({ (({2)))))) → (({aa))} → 101
(({ (({3)))))) → (({zz))} → 26926
(({ (({1))}(({1)))))) → (({aa))} → 101
The only solutions I can think of are:
Append all ((x1)) parameters when invoking ((x2)): ((x2|1=a|2=aa|3=zz||a=1|b=2|c=3|d=4|e=5|...|z=26|aa=11|bb=22|...|zz=2626))
Expand all of the parameters when invoking ((x2)): ((x2|1=(({a))}|2=(({aa))}|3=(({zz))))}
However, for my particular application, neither of these is workable. The first is unworkable because my x1 has about 200 parameters and invokes x1 multiple times, which I'd like to have appear on consecutive lines. The second is unworkable because my x2 mangles its input parameter values something like example 4 above so that each x2 parameter is actually used to select up to five different x1 parameters, and if include them all, then my x2 invocations won't fit on a single line. Any ideas? YBG (talk) 21:23, 9 November 2018 (UTC)
PHP fatal error: entire web request took longer than 60 seconds and timed out: Supposedly because of maintenance, but it's been doing this for hours. Any ideas? ——SerialNumber5412914:22, 10 November 2018 (UTC)
I've been bothered by the obvious move of the raw link only to where you have to load the pretty edit-watchlist page, so unless you know the raw page is there, you're screwed with super-long watchlists. Tempted to file a phab about it. --Izno (talk) 15:58, 10 November 2018 (UTC)
In many of the pages, I was able to see that there are multiple short descriptions occurring for a single article. Both of them are same. For example, consider the page Kondotty. I know there are two short descriptions, one from wikidata and other from wikipedia. But I guess the short description from wikidata was overridden with the one from wikipedia as requested by English wikipedia.Adithyak1997 (talk) 10:54, 11 November 2018 (UTC)
@PrimeHunter:So in such a case, shouldn't there be a condition set to display only one of them? I mean, in cases where there are two short descriptions, shouldn't there be a criteria to display only one of them?Adithyak1997 (talk) 14:25, 11 November 2018 (UTC)
Suddenly I'm seeing &ndash, (with a comma, followed by a blank space) where formerly I saw an actual en-dash. Is it just me or my laptop, or has this suddenly stopped working? There are probably over a million occurrences of this within Wikipedia. Michael Hardy (talk) 19:39, 11 November 2018 (UTC)
Interesting, thanks. I had no idea interwiki maps extended beyond the wikis. One more complication for bot writers :) -- GreenC00:46, 11 November 2018 (UTC)
I recall sometime back discussing on here about an IP talk page that could not be created because it was blacklisted. User talk:2600:1700:8680:E900:8C6F:CAC6:D0E0:A9EB just exhibited the same behavior; the message intended to be left there was left at the user page instead, and a sysop had to move the page to the correct location. I recall this issue having something to do with a rule on the title blacklist; can IP talk pages be exempted from whatever it is? Home Lander (talk) 21:42, 11 November 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Recent changes
Some old mobile browsers can use the watchlist again. This has not worked for a while. These browsers are called grade C browsers. This helps for example Windows Phone 8.1 with Internet Explorer and Lumia 535 with Windows 10. [47]
Problems
You can choose to see edit conflicts in a two-column view. This is a beta feature. You can find it in your preferences. Users who use this view saw the edit conflict resolution page when they wanted to see a preview. This has been fixed. [48][49][50]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 13 November. It will be on non-Wikipedia wikis and some Wikipedias from 14 November. It will be on all wikis from 15 November (calendar).
Meetings
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 14 November at 15:00 (UTC). See how to join.
Future changes
You can use the content translation tool to translate articles. The developers are working on a new version. One of the changes will be a maintenance category. Articles where users add a lot of text from machine translation without changing it will be in that category. This is so the community can review it. The users will also have been warned before they publish the article that it has a lot of unchanged text from machine translations. [51]
When i press shift+del in chrome on windows after highlighting the username, chrome only removes the password, and the username is offered the next time one clicks on "log in". --Espoo (talk) 17:58, 9 November 2018 (UTC)
Thanks. It would be important to provide a button after logging out that automates that for users, most of whom don't know they have to do that or how to do that. --Espoo (talk) 07:17, 10 November 2018 (UTC)
Those are controls about options in your browser which we don't control. One tip that might be good for your use case, use "private browsing" sessions, Microsoft, Google, and Mozilla browsers support these and won't remember anything from that session. — xaosfluxTalk17:06, 10 November 2018 (UTC)
...assuming that you don't mind logging in repeatedly. I accidentally had a wiki page in "private browsing" mode a few week ago, and it took me a while to figure out why it kept demanding that I log in over and over and over. Whatamidoing (WMF) (talk) 23:31, 12 November 2018 (UTC)
WP:SPI "Date filed" showing an error for some investigations?
A proposal to add responsive design to the main page was briefly implemented, then reverted after problems for certain browsers and gadgets showed up. Before trying to run it live again, it would be beneficial to do some testing with some different browsers and such. Anyone want to take a look at User:Yair rand/MPSandbox and report whether you see any issues? --Yair rand (talk) 21:38, 11 November 2018 (UTC)
Looks fine to me on several older iOS devices. I'd also highly advise to not rollback on every single bug that is being found when this only effects minute amounts of users. That leads to months long cycles that aren't very productive and actually hurt testing. Rollout and fix where issues are reported. The burden upon people working on this to have tested it against BB10s is unreasonable, there is no problem with such a small user groups NOT being able to use a single page for a couple of days, while others work to fix the problems. —TheDJ (talk • contribs) 09:02, 13 November 2018 (UTC)
Are there limitations on the number of versions that can be revision deleted in one action? I'm trying to respond to the RD one request at Mohamed Naguib, which involves more than 250 versions. I tried twice, and each time the browser tab crashed. I'm happy to do it in chunks, but thought I'd check here first in case there's something I need to know.--S Philbrick(Talk)15:58, 13 November 2018 (UTC)
For some reason, my Special:Contributions page looked like this earlier: (Imgur link). It was the case whether I was logged in, logged out, using Chrome or using Edge. The problem appears to have since been resolved, but still. What happened here? Narutolovehinata5tccsdnew20:46, 8 November 2018 (UTC)
This is apparently still happening when the page is transcluded (see my sandbox). I have a ticket open because it recently started breaking MathML rendering (see phab:T209446). — AfroThundr (u · t · c)01:34, 14 November 2018 (UTC)
Regular expression to add flag icons and country links to country list
Can anyone give me the regular expressions to quickly link all the country names, and to add the same flag icon wikitext that currently exist on the article pages:
@Timeshifter: I don't know your text editor syntax but our default source editor has regular expressions on "Advanced" followed by the search and replace icon on the right. There it works to search for (\|-.*\n\|\s*)([^\|\n]*) and replace with $1((flagcountry|$2)). Select "Treat search string as a regular expression" before clicking "Replace all". PrimeHunter (talk) 17:04, 13 November 2018 (UTC)
@PrimeHunter: Wow, this is going to help a lot of people making tables, and especially in updating tables. Thanks!!! See an example, along with some how-to info:
I am trying to use Earwig's Copyvio Detector to search a document in draft space for copyright violations, using the search engine. After it connects to wmflabs, it says that no sources were checked, and the probability of a copyright violation is 0%. Well, it should be 0% if there was no searching. What causes it not to search anything, and what do I do to correct this, to get it to search? Robert McClenon (talk) 07:09, 11 November 2018 (UTC)
This probably happened because there was discussion about copyright violations on drafts, and the reviewers were checking the drafts. The screen refers to a cache. Maybe the cache becomes full, and the tool doesn't handle that well. Does User:Earwig still maintain the tool? Robert McClenon (talk) 01:12, 12 November 2018 (UTC)
Thanks. Robert McClenon, I looked, and there doesn't seem to be any systemic problem with the tool right now. In general, you'll see that if every phrase from the article the tool attempted to search for in Google returned no results, and it has nothing else to go off of. The tool picks phrases at random throughout the article, so it could have gotten unlucky, or maybe the draft is truly unique, or maybe Google had some kind of problem. In my experience, if we've reached the daily request quota, you'd see a different error, but that may have changed. There shouldn't be any issues related to caches becoming full. If you give me the name of the draft, I can look more closely, but I can't be much more specific without further details. — Earwigtalk03:05, 13 November 2018 (UTC)
Robert McClenon, indeed, every sentence in that article is unique (at least as far as Google is aware). No sources were checked because no potential sources were found. (If you look at the top right, where it says "generated in X seconds using 8 queries", that tells us eight separate searches were made, and none of them turned up any hits.) — Earwigtalk02:23, 14 November 2018 (UTC)
Searching my contributions with a filter on a previous namespace?
I'm trying to find, in my contributions list, an edit I made to a draft. I don't remember the title of the draft, or specifically what I put in the comment field, but I'll recognize it when I see it. The edit was made sometime in the past few days.
If I filter on Draft namespace, I don't see it. I think the problem is that the draft was accepted and now lives in mainspace, so the filter on draft space no longer works. Is there some way to filter on, "It was in draft space when I made the edit"? -- RoySmith(talk)19:10, 14 November 2018 (UTC)
If you can't find it in your contributions, perhaps you would see it by reviewing the moving log; Ctrl+F "page draft:" and skim through those. It may be possible to find the edit you made using a database query at WP:Quarry. --Izno (talk) 20:52, 14 November 2018 (UTC)
I recently moved a page from a user sandbox to draft space. My User contributions shows the move:
(change visibility) 19:10, 14 November 2018 diffhist (0) m Draft:Highschoolers to Run Under Four Minutes in the Mile (RoySmith moved page User:Herg-derg-editor/sandbox to Draft:Highschoolers to Run Under Four Minutes in the Mile: Preferred location for AfC submissions) (current) [rollback: 1 edit]
But, if I look at the draft history, it claims that User:Legacypac is the one who did the move, one minute earlier:
(cur | prev) 19:09, 14 November 2018 Legacypac (talk | contribs | block) m . . (2,895 bytes) (0) . . (Legacypac moved page User:Herg-derg-editor/sandbox to Draft:Highschoolers to Run the Four Minute Mile: Preferred location for AfC submissions) (undo | thank)
Legacypac's contributions shows a confusing double-entry for this:
(change visibility) 19:09, 14 November 2018 diffhist (0) m Draft:Highschoolers to Run the Four Minute Mile (Legacypac moved page User:Herg-derg-editor/sandbox to Draft:Highschoolers to Run the Four Minute Mile: Preferred location for AfC submissions)
(change visibility) 19:09, 14 November 2018 diffhist (+78) N Draft:Highschoolers to Run Under Four Minutes in the Mile (Legacypac moved page User:Herg-derg-editor/sandbox to Draft:Highschoolers to Run the Four Minute Mile: Preferred location for AfC submissions) (Tag: New redirect)
Doh! Of course, thanks for figuring that out. There was indeed a race condition, but it was between when I decided to move the page and when I actually clicked the button. I'll move the redirect back -- RoySmith(talk)00:57, 15 November 2018 (UTC)
Script issues
Following a query (on my talk page), I tried the usual troubleshooting actions, but cannot understand why my Engvar script doesn't work in isolation and in the context of another user's monobook script (meaning the button doesn't dislay in the side bar when supposedly installed). I did find however, that all my scripts become callable from within my monoboook files. My workaround is to advise importing my monobook.js file by pasting the instruction:
importScript("User:Ohconfucius/monobook.js"); while removing importScript("User:Ohconfucius/script/EngvarB.js");
The consequence, however, is that it creates a very busy the sidebar above all for the users who have no utility for my other scripts, but at least it seems like an acceptable work-around for me. What could be the problem? -- Ohc ¡digame!09:35, 10 November 2018 (UTC)
This sounds like your Engvar script has some dependencies that are resolved by the presence of one or more of the other scripts in your monobook.js - try removing scripts one at a time from monobook.js until EngvarB.js fails. Then examine the last one removed, to see what it does that Engvar might depend upon. --Redrose64 🌹 (talk) 09:53, 10 November 2018 (UTC)
Thanks, @Redrose64: I tried loading a few scripts on my vector page. It seems that there is no dependency on my formatgeneral script. However, it seems to need either my foreigndates script or my MOSNUM script for example. How can I now identify what element in these scripts that ENGVAR requires and then to make the appropriae addition? -- Ohc ¡digame!16:53, 14 November 2018 (UTC)
As discussed some time ago (HT User:Plastikspork), we could do with a tool to facilitate the comparison of related templates to decide whether or not to merge them.
The tool would do the following, for two (or more?) templates:
determine the list of parameters in each (perhaps from raw code; perhaps from TemplateData; perhaps from ((Parameter names example)))
You could use the TemplateData GUI editor to produce a list of (most) parameters from the code. From there, I think that a quick round of grep -xvf would produce the list of unmatched options, for manual review. Whatamidoing (WMF) (talk) 17:23, 9 November 2018 (UTC)
Here is a module hack that, for the purposes of proof-of-concept compares ((cite book/old)) against ((cite journal/old)) (these are the pre-Lua cs1 templates). The output is merely the raw mw.dumpObject() return value and can be prettified to suit:
((#invoke:Sandbox/trappist the monk/template compare|compare|Template:Cite book/old|Template:Cite journal/old))
Pigsonthewing, I adapted some javascript that I wrote for "User:Frietjes/addcheckforunknownparameters.js" to create "User:Frietjes/templatecompare.js". once installed, you go to "Special:TemplateCompare" and put in the list of templates to compare. if you see a "Did not finish processing" alert, then let me know, and I will adjust the regular expressions to try to get complete processing for that particular template. I tested it on ((cite book/old)) against ((cite journal/old)) and it looks like it's working there. Frietjes (talk) 21:04, 9 November 2018 (UTC)
@Pigsonthewing:, it could be taking some time to process the templates. I put the pop-up alerts in there to show the progress through the script. if you have particular examples you want me to help debug, let me know. Frietjes (talk) 13:33, 13 November 2018 (UTC)
@Pigsonthewing:, you are correct. it was failing on the expandtemplates for some reason, which is only used to show a preview of the wikitext. I switched this to something else, it is working for me now. Frietjes (talk) 20:28, 13 November 2018 (UTC)
One computer can't load Wikipedia while another can
Hello, I've got two Windows computers, and yet one's able to access Wikipedia while the other's not. One's in Windows 7 with IE 11.0.9600.19155 and the other's in Windows 10 with IE 11.345.17134.0. I'm logged into the latter, and using my normal Monobook, while the other isn't logged in at all. This computer (obviously) can access the site, while the other is consistently returning a 404 error for Wikipedia. At the same time, the problem appears to be restricted to en:wp — using Windows 7, I can access other random sites fine (I run a Google search, find a website I've never seen before, and it loads fine), and I'm able to navigate to de:wp, fr:wp, Commons, and every other WMF wiki that I've tried. Any ideas? — Preceding unsigned comment added by Nyttend (talk • contribs) 17:54, 15 November 2018 (UTC)
Sometimes when the system clock is off by enough it prevents https from working correctly. And try clearing cookies and browser cache. -- GreenC18:38, 15 November 2018 (UTC)
Can't load it in Firefox 60.2.2esr or Chrome 70.0.3538.102; I get results identical to IE. I virtually never use Chrome or Firefox (except for loading sites that don't work in IE), so I doubt that there's anything problematic in the cookies or the cache. I've cleared both in Firefox and gotten the same results; I'm not immediately clear how to clear them in Chrome, so I've not (yet) done that. The computer I'm using has a system time of 15:40 on 2018-11-15, and the other one has a time of 3:40 PM on 11/15/2018 — can't see how this would be a problem in my specific situation. Nyttend (talk) 20:39, 15 November 2018 (UTC)
And now, somehow, I'm able to access en:wp on Windows 7. Nothing's different, as far as I can tell, while before I wouldn't have even had the chance to log in. Nyttend backup (talk) 20:54, 15 November 2018 (UTC)
In Chrome (at least on Mac; I'm assuming the same on Windows), you can open the Developer Tools window (Command-Option-I), select the network tab, and then reload your page. It'll show you the details of every network access your browser makes. Sometimes you can discover interesting things, like javascript files failing to load. You can also look in the javascript console for error messages. Even if you don't know how to interpret these messages yourself, you can copy-paste them and other people may be able make use of them. My hunch is that some javascript file fetch is timing out, but that's just a hunch. The network trace would help verify or disprove that. -- RoySmith(talk)22:57, 15 November 2018 (UTC)
It's displaying the wikitext exactly as expected for me. You can try editing the page in safe mode [53] or whilst logged out to see if it's a malfunctioning script you have, or try a different browser to see if it's your browser. --Deskana (talk) 00:12, 16 November 2018 (UTC)
(edit conflict) Hi Ivanvector, Xaosflux, yes, On checking the page again, I found it, I came here to self revert and got edit conflicted with you. thanks for your quick comments, neverthless. I have marked this as resolved --DBigXrayᗙ14:52, 16 November 2018 (UTC)
For the record, "Show only likely problem edits (and hide probably good edits)" at Special:Preferences#mw-prefsection-rc can hide edits on contributions. The option is on the "Recent changes" tab but the subheading says "Revision scoring on Recent changes, Related changes, and Contributions". PrimeHunter (talk) 14:56, 16 November 2018 (UTC)
Clearly the real problem here, was just my failure to notice the said edits which were there at the bottom and the next page of the contributions page. PrimeHunter I see, my preferences for these were unchecked. but this looks like a good feature, thank you for making this comment, since I learnt about this new feature today. --DBigXrayᗙ16:42, 16 November 2018 (UTC)
Revision scoring on Recent changes, Related changes, and Contributions
While testing this feature Mentioned by Primehunter I selected "Medium" threshold and checked "Highlight likely problem edits with colors and an "r" for "needs review"" . But I still dont see any kind of highlight on the Recent changes and contributions from accounts reported at WP:AIV.PrimeHunter, Is there something else needed to be done in order to make this work ? --DBigXrayᗙ16:42, 16 November 2018 (UTC)
With those selections I see around 10 highlighted edits in the 500 most recent changes. It appears that user contributions are not highlighted. PrimeHunter (talk) 19:31, 16 November 2018 (UTC)
PrimeHunter, yes, now I was able to see some orange highlighting on "Recent changes". But on clicking the contributions of problem editors (whose edits were highlighted on RC page, the same edits from the editor were not highlighted on his contribution page. Same as you observed. Is it the case that this feature is still not enabled for user contributions or is it a bug ?--DBigXrayᗙ19:44, 16 November 2018 (UTC)
DYKUpdateBOT malfunctioning
For some reason, the automated updates by the DYK bot have not been running for two days in a row. Requests for manual updates from admins have been posted at Main Page Errors. Any suggestions for a fix are appreciated. Flibirigit (talk) 05:16, 18 November 2018 (UTC)
NPR NEEDS YOU!New Page Reviewers operate the only firewall against junk, attack, spam, and undeclared paid editing which has aways made up the majority of a day's intake of new pages masquerading as articles. Community Wishlist Voting is taking place now until 30 November for the Page Curation and New Pages Feed improvements, and other software requests. The NPP community is hoping for a good turnout in support of the requests to Santa for the tools that are urgently needed. This is very important as the Foundation has been constantly asked for these upgrades for 4 years. The Page Curation suite of tools now stands a good chance of getting long awaited attention to the upgraded tools it needs, but it needs your help: whether you are an active patroller or just want a junk-free encyclopedia, the Community Wishlist Survey needs you: Vote NOW, and do also consider applying to become a New Page Reviewer. Kudpung กุดผึ้ง (talk) 11:57, 18 November 2018 (UTC)
Someone requested that an account be created. This was done and a temporary password assigned but they failed to login before the password expired. They've requested a new password but I don't know how to do that. Is it possible? if it matters, the username is Titanrich. ticket:2018111710004008 --S Philbrick(Talk)13:28, 18 November 2018 (UTC)
Xaosflux, Thank you. The sad thing is that I knew that but had a brain freeze and forgot that was possible. I've point to the user to this discussion so hopefully they will either follow the advice or get back to me and I'll help. S Philbrick(Talk)16:23, 18 November 2018 (UTC)
I'm afraid I opened a can of worms when changed the target of redirects Open-source and Open source toward Open source (disambiguation) and I suspect a bot might help but I neither know how to request nor build one. I was not even aware of this problem until it was pointed out to me, User talk:JasonCarswell#Open-source (moved to Talk:Open source), where the conversation continues. I will watch this space for answers but I'd prefer that you continue the conversation on my talk page and include/ping the others please. A bot would save days worth of editing. Thank you in advance. ~ JasonCarswell(talk)13:22, 18 November 2018 (UTC)
Ideally the bot would go through theses lists of thousands and find/add/replace disambiguation-pipe-links then no longer be needed, unless I'm mistaken about something I'm unaware of.
@EdJohnston:, I apologize for forgetting to update this section. My discussion was archived and copied to the better place Talk:Open source.
Problems using the VE interface
Today I set some global preferences which apparently have me using the VE interface instead of the source editor I am accustomed to. I figured I would give it a try but am dismayed by the interaction. Firstly, I only see a link to publish my changes; no preview, show changes, or cancel buttons exist. Secondly, the down arrow key between the publish changes tab and the pencil symbol gives a drop down menu which includes a source editing link but clicking on it does not escape the visual editor. I am experiencing the same thing in preparing this posting as well. I have attempted to include a file showing a screenshot but can not preview its positioning or size so if it's snafu-ed, please fix it for me. I'll be heading over to mw to undo my global preferences and have a look when I return. Doesn't that beat all? When I pressed the publish changes button an interim screen opened which contained therein a preview button, another publish changes button, and a resume editing button. Whomever thought it would be intuitive to press publish changes to get to the preview button is sorely mistaken. I'm still fixing to turn it off.--John Cline (talk) 12:40, 17 November 2018 (UTC)
Clicking Publish takes you to a modal popup that allows you to Preview, Show changes, and Publish, as well as provide your edit summary. Old hands hate the workflow. :) Pressing your Esc key will cancel an edit. --Izno (talk) 18:25, 17 November 2018 (UTC)
As for "escaping" the VE, you're looking for the Beta feature called "New wikitext mode". Turn that off. You may also have "Automatically enable all new beta features" turned on. You may not want that either. --Izno (talk) 18:28, 17 November 2018 (UTC)
Thanks Izno, I appreciate you. If this is a beta feature, where do you go to share feedback? There are aspects of this "new mode" that I could warm up to; assuming the bugs will be resolved.--John Cline (talk) 19:53, 17 November 2018 (UTC)
I have a whole series of bugs on that workflow. The workflow was designed for use in the visual editor, not in the wikitext mode, and it needs a second look. The reality is that nothing much is going to change there for a while – probably the middle of next year at the very earliest, unless it needs to be re-done in the mobile version, too.
If you otherwise like the "2017 wikitext editor" (main advantages: citoid, in-editor search for pages whose names you can't quite remember, and faster switching to the visual mode), then it's probably worth memorizing the Wikipedia:Keyboard shortcuts for preview and diffing (they're the same as you'd use in the other wikitext editors). Whatamidoing (WMF) (talk) 22:16, 19 November 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
There is an A/B test for sameAs data. This is to make it easier to find the right information with a search engine. This changes the metadata for a wiki page. It doesn't change how the page looks. [54]
Changes later this week
There is no new MediaWiki version this week.
Meetings
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 21 November at 15:00 (UTC). See how to join.
Future changes
The Wikimedia wikis use templates to show readers there are problems with the content on some pages. For example if there are no sources or the page needs to be rewritten. The mobile website will soon show more information when you use these templates. Some templates may need to be updated. [55]
The Education Program extension was removed from all Wikimedia projects. The database tables used by the extension will be archived. This will happen in a month. If you want the information on your wiki you should move it to a normal wiki page. [56]
User:WP 1.0 bot is a bot that creates wikiproject assessment tables and changelogs, a job which it has done for years. At the end of September, however, it started breaking down on the 'changelogs' task, and was unable to ever complete a run until it was blocked on November 5. The bot does not have an 'owner', and the project's talk page has little communication from the volunteer runners about fixing the bot or about an ETA for their planned revision of the bot altogether. I maintain an offline script that parses the changelog (ex.) to generate a weekly "new articles" report for the video games project (ex.), so the death of this bot has taken down my script as well. My question, therefore is:
Does anyone know of any datasource I can use to generate reports on assessment changes for articles tagged by a wikiproject? My ideal it would be all changes, but I'd settle for just new additions and deletions; so far I've just found User:AlexNewArtBot/VideogamesSearchResult, which has new page creations that match some heuristic but doesn't do deletions, moves from draftspace, or un-redirections. Thanks! --PresN21:39, 19 November 2018 (UTC)
Right now User:Audiodude is working with User:Kelson, writing and testing some code to try and fix the bot. Last I heard he was doing some testing but it was running too slowly. I'm afraid I don't know of alternatives, but hopefully in a few weeks the User:WP 1.0 bot will be working properly again. They are also looking to write some code for a new version of the bot; please leave your comments and ideas here. — Preceding unsigned comment added by Walkerma (talk • contribs) 04:08, 20 November 2018 (UTC)
You can't. The database Quarry queries from doesn't have page text, so there's no way to tell whether a given parameter is there or not. Search can (unreliably) find these, as above; or you could add a tracking category to the template if reason is present. —Cryptic17:04, 20 November 2018 (UTC)
Wikipedia links on Twitter - absence of twitter:card
Wikidata links on twitter result in a well-formatted link, arising out of the inclusion of <meta name="twitter:card" content="summary"/> in the wikidata source. Example: https://twitter.com/vrandezo/status/967076003041259520
So, as far as I can ascertain from phab:T157145, we're not implementing a 5-minute solution which would at least allow the full title of the article to be read on a tweet, because a much more complicated solution involving a new user-editable summary of the article can be imagined but not in any realistic timescale be implemented. phab:T157145 has been open for about 2 years whilst we contemplate this unicorn prospect. Meanwhile tweets show truncated URLs. Yay wikipedia! #smh.--Tagishsimon (talk) 18:53, 20 November 2018 (UTC)
We're not a social media site... whatever is a twitter card anyhow? There's no article at twitter card or even twitter:card, and no mention at Twitter. --Redrose64 🌹 (talk) 19:13, 20 November 2018 (UTC)
Did you trouble to look at and examine the two links I provided above, Redrose64? It's a question of how wikipedia chooses to represent itself on twitter; I grant, if that's not of interest to you, then it won't be of interest to you. But, you know, enough with the nihlism. "We're not a social media site". ffs. --Tagishsimon (talk) 19:17, 20 November 2018 (UTC)
But hey, let's take this slowly. twitter:card is an attribute that can be placed in the header of an HTML page. You can see an example of twitter:card code on the first line of this thread, as well as an explanation of the difference it makes. It affects the way in which a URL on twitter linking to a wikipedia page is formatted on twitter. Right now, links on twitter to wikipedia pages are bare and truncated; users cannot, at least without moving their mouse over the link, get an idea of the title of the article linked to. HTML pages having a twitter:card attribute cause the link to be formatted in a digital analogy of a card - a skeuomorphic representation of a small piece of paper, if you like, Redrose ... think business card. You can see the difference on this short thread - https://twitter.com/Tagishsimon/status/1064925227711369216
Twitter had ~336 million monthly active users in Q1-2018 [57].
It seems to me reasonable to seek to improve the way in which wikipedia links on twitter are formatted, such that they're a little more like wikidata links on twitter. But right now we seem to be hung up on a 'problem' that wikipedia does not have a short summary to provide to the twitter:card (whereas wikidata can provide the english language description (if it exists)). I'm seeking to make the case that we should not hang ourselves up on this, and simply provide the article title to the twitter:card ... because that would be very much better than the current bare truncated link. The article title is already an attribute of the HTML header. It took wikidata about 5 minutes to implement their twitter:card. It has taken us 2 years not to do so. --Tagishsimon (talk) 19:52, 20 November 2018 (UTC)
(I know that sometimes modifications in templates are visible only after some delay in the "What links here" functionality, but I've now waited for nine days...)
Also, in that same vein; category counts. Is there some way to generate a table listing how many articles are in various maintenance categories? (I've heard this feature is broken, and the CSD categories vs. the number on the admin dashboard lead me to belive it.) ~ ONUnicorn(Talk|Contribs)problem solving17:19, 21 November 2018 (UTC)
If I understand the OP's question about category counts, ((clc)) is also an option. You have to purge to update the counts, and the counts are not always correct for categories with counts above 200 (an old bug), but that's probably true of any method. See my list of categories to watch for an example. – Jonesey95 (talk) 18:51, 21 November 2018 (UTC)
They just attached a screenshot of an attempt to reset the password which states they are unable to reset a password because they are blocked globally.
Sphilbrick, global blocks apply only to IPs; accounts get globally locked. I'm therefore guessing their IP is probably the culprit, and why nothing shows in the log for the account. Home Lander (talk) 23:31, 21 November 2018 (UTC)
Home Lander, Fair enough, but then why does the linked page say "Global block log" and have a place to enter "User:username for user"? (I'm not disputing you are right, I'm just suggesting that the page is misleading.) S Philbrick(Talk)23:55, 21 November 2018 (UTC)
Sphilbrick, dunno, but that log seems quite useless, I tried entering a globally locked account, a globally blocked IP, and a globally blocked range. None of them generated anything from that log. Home Lander (talk) 00:03, 22 November 2018 (UTC)
I also saw it for the first time a week or two ago. I clicked the first button to close it and possibly another button to get rid of it altogether and haven't seen it since. I did not enable it to begin with. Killiondude (talk) 23:09, 19 November 2018 (UTC)
Gone now. Found where it was buried, although the wording makes it look like the tool is enabled by default. [59]. Maybe it's ticked on by default though. Headbomb {t · c · p · b}23:12, 19 November 2018 (UTC)
You can enable it by clicking "curate this page" in the toolbar on the side and disable it by clicking the little arrow pointing at the wall thing and then clicking the x. Natureium (talk) 23:15, 19 November 2018 (UTC)
All of my links on the site today are purple for some reason, except those that have a colour added (like redlinks or those coloured by linkclassifier). What up? Ivanvector (Talk/Edits) 14:31, 22 November 2018 (UTC)
@Ivanvector: It might be something as simple as a dodgy cable. I have a monitor where the wire carrying the blue signal develops a break as the monitor warms up - white becomes yellow, magenta becomes red, cyan becomes green and blue becomes black. If I turn it off and let it cool, then turn it back on again, the blue is restored together with full colour. In your case, the green wire might be faulty. --Redrose64 🌹 (talk) 21:17, 22 November 2018 (UTC)
I had an old CRT that used to do that. Good times. This one is HDMI though, I don't think the same thing is the culprit. The monitor has been on its way out for a while anyway. Ivanvector (Talk/Edits) 22:32, 22 November 2018 (UTC)
Special:ProtectedPages - Hide redirects filter
Hello all,
I'm unsure as to whether the following is a bug or a deliberate decision:
Special:ProtectedPages has a filter to hide redirects. However, this filter does not exclude pages which are members of the category: Redirects to Wiktionary. For example, these five pages all show up in a search excluding redirects:[60].
In my opinion redirects to Wikitionary should be excluded, just like redirects within Wikipedia itself, but am happy to be convinced otherwise. Greg (talk) 22:57, 22 November 2018 (UTC)
Wiktionary redirects aren't technically redirects. They're "soft redirects"—instead of redirecting you to the page directly, it presents you a link to the article that you have to click on. From a technical standpoint, the Wiktionary redirects are just like any other article, so filtering them out of Special:ProtectedPages would require nontrivial developer work to implement. It would probably also require an implementation specific to the English Wikipedia (not all projects have Wiktionary redirects), and my understanding is developers don't like writing enwiki-specific code (things should be as globally applicable as possible). Mz7 (talk) 00:08, 23 November 2018 (UTC)
A problem with large tables is that the user, when scrolling, may lose track of what each column represents. This is occasionally solved by adding a duplicate of the table header at the bottom or even multiple in the middle of the table. A better solution would be for the table header to follow along when the user scrolls, so it would be fixed to the top of the screen when the user is scrolled partway down the table. This can be implemented by position: sticky;, which is now widely supported on th elements. However, this would only work when the header is only one row high.
Assuming this is considered a good idea, what would be a good way of implementing it? We can't just add some CSS to common.css to have it apply to all tables, as that would really mess up some tables, and it's only really useful on very large tables. Perhaps a new template, that could be used in particular situations? Any ideas? --Yair rand (talk) 00:04, 22 November 2018 (UTC)
Are you suggesting something like Excel employs, freezing the header rows so the ones below scroll up and behind the header as the user scrolls? — Maile (talk) 00:09, 22 November 2018 (UTC)
Huh. I was not aware of that gadget. However, it only works on Firefox and Safari, because it applies the styles to thead. If it were applied to th, it would also work for Chrome and Edge. --Yair rand (talk) 00:59, 22 November 2018 (UTC)
But then you have the problem you describe above where headers are only one row high. There are sufficient numbers of tables where there are two rows (the one I happened just to be testing on was list of Presidents of the United States, which I guess gets a few views...). There are also skin differences (THANKS TIMELESS <3). We could ostensibly make it so the Javascript applies a different class for IE and Chrome on the THs instead... --Izno (talk) 01:23, 22 November 2018 (UTC)
Honestly the whole sticky thing is terrible. Exactly the thing you'd want to use it for <thead>s are what they don't work for and they have little use outside of that. It's really annoying that after 5 years the browser vendors still haven't figured out that sticky is a broken thing. I've personally sort of given up on it. If you want this, you will need JS, but JS is less performant and harder because our website has lots of OTHER javascripts that interact with tables, which makes it pretty impossible. —TheDJ (talk • contribs) 08:30, 23 November 2018 (UTC)
Are portals being made automatically with an automated system?
Just noticed there's seems to hundred(s)? of new portals on the most odd topics...like the new series below.
I'm not sure I follow you. The topics listed above appear to be pretty normal as far as encyclopedia subjects go. They are covered by articles, navigation templates, and categories. Portals are generally set up to help navigate subjects comprised of multiple articles, like these subjects. — The Transhumanist03:58, 21 November 2018 (UTC)
The portal-spammers seem to be working through this enormous list. Since none of these portals seem to have any particular value we should probably have a mass MFD of them and if necessary send the creators on an enforced wikibreak, as this is about a blatant a WP:POINT exercise as I can imagine. Any discussion is probably better off at WP:ANI regarding the disruption side of things. ‑ Iridescent11:12, 20 November 2018 (UTC)
There was an attempt to mass delete all portals and the portal namespace itself, at Wikipedia:Village pump (proposals)/RfC: Ending the system of portals. In that discussion, it was decided by an overwhelming majority, not only to keep portals, but to automate most of them to minimize the maintenance required. The lack of maintenance was making many portals stale and out-of-date. The new design elements that portals use now overcome the "stale" excerpt problem, and add new material over time by automatically harvesting links from the encyclopedia. Portals are modeled loosely after Wikipedia's Main page, a format many people find highly familiar and useful. Thus, each portal provides a similar layout and functionality to navigate a particular subject's coverage throughout Wikipedia, just like the Main page provides samplings from Wikipedia's coverage of knowledge as a whole. The Portals WikiProject is very careful to follow Wikipedia's policies and guidelines, and portals are covered by specific guidelines as well. You are welcome to come check out the latest design concepts we are working on at WT:WPPORTD. — The Transhumanist04:54, 21 November 2018 (UTC)
Excerpts are all done through wp:selective transclusion. They used to be copied and pasted by hand to a subpage, and then the subpage transcluded. Transcluding excerpts directly onto the page ensures that they always match the source, avoiding content forking issues and growing out of date.
Selected sections, for the most part, are populated from source pages in the encyclopedia. For Selected images (slideshow) sections, source pages are typically articles. For Selected articles (excerpt slideshows) sections, the source pages are usually navigation templates or lists.
Did you know sections are powered by lua search of the did you know archive. The typical search period is the past 3 years. These sections only appear when there are results to display.
In the news sections are powered by lua search of the current events news archive. The typical search period is the past 45 days. These sections only appear when there are results to display.
The Subcategories sections are powered by CatTree.
The Recognized content sections of portals are powered by JL-Bot.
The Associated Wikimedia sections are powered by a template that includes searches based on magic words.
No, the portals are not being created automatically, there's a person on the other end. They are generated using ((Basic portal start page)), a template which does alll the heavy lifting and allows portals to be created in (nearly) a single edit. A lot of work has gone into automating different parts of a portal's structure, which significantly reduces the maintenance burden and manual upkeep costs. This means that you will see fewer old, outdated, or abandoned portals floating around. We are also overhauling the old portals as well, and cleaning up the namespace. You can check out what we've been up to at WP:WPPORT, and for an overview of what everything in the new generation portal does, you can see a demonstration at WP:WPPORT/D. BTW, for those who keep tabs on what happens in the Portal namespace, there is an ongoing discussion on updating the portal guidelines over at WT:PORTG, if you'd like to chime in. — AfroThundr (u · t · c)20:42, 20 November 2018 (UTC)
Should we hAve these?.....look at the pictures at Portal:Prostitution in Canada. ... prime minister Harper and the father's of Confederation... is this a joke.....looks like someone is going out of the way to mock Canada?--Moxy (talk) 22:15, 20 November 2018 (UTC)
So to be clear.....he did not choose the images of Canadian prime minister's on purpose? ...just automated junk that by chance chose the fathers of the confederation?--Moxy (talk) 22:35, 20 November 2018 (UTC)
Moxy not quite. The pictures for that portal are automatically pulled directly out of the eponymous root article, Prostitution in Canada. Slideshow contents can be adjusted to match the portal's context more closely by removing the root article as the sourcepage, finding other more appropriate pictures for the slideshow, or by removing the slideshow altogether. Each portal is a work-in-progress, just like the rest of the pages on Wikipedia. The best places to report issues with a specific portal is the portal's talk page, or WT:WPPORT, from where any problems you have spotted can be most rapidly addressed and fixed. — The Transhumanist02:59, 21 November 2018 (UTC)
Looking more closely, it doesn't appear that any of them are actually linked from anywhere other than from each other (e.g.). As long as this is just a walled garden and they're not trying to link to these things from articles, this is just low-level misuse of Wikipedia as a web host, which is still a violation of policy and a waste of the time both of the people creating these things and of everyone else when they inevitably get MFD'd, but not the kind of major issue that warrants immediate action. I repeat that WP:VPT is not the best place to be having this discussion, as the issue is one of inappropriate content rather than a technical matter. ‑ Iridescent22:27, 20 November 2018 (UTC)
@Iridescent: Portals created through batch creation are not getting linked and this is something as a WikiProject we are trying to do. Apart from some AWB, adding links is not as automated as the creation is, so users such as The Transhumanist have not been adding links when they create portals in batches. Over time links will be added, but at the moment this is a slow process. Dreamy Jazz 🎷 talk to me | my contributions22:36, 20 November 2018 (UTC)
Also these portals are not being generally deleted at Mfd. The current portal guidlinea allow these portals due to the large number of articles that they have, even when the topic is neiche. An automated portal now usually has 30 articles, which is way above the minimum. Dreamy Jazz 🎷 talk to me | my contributions22:43, 20 November 2018 (UTC)
The portals are being created in good faith, though I think their creators could be more selective. The RfC highlighted that, whilst many portals are lovingly curated, others on broad subjects had fallen into disrepair. Tools have been produced to assist with replacing rotting portals by a low maintenance format. (For example, excerpts from articles now update automatically whenever the article is edited, rather than becoming outdated unless an editor interves.) These tools have also been quite properly used to fill in gaps where important topics lacked a portal. They are now being used to create portals on subjects which fall within the guidelines but are much narrower. There has been useful discussion but no consensus as to the merits of such portals. Certes (talk) 01:05, 21 November 2018 (UTC)
It might not hurt to come up with some kind of criteria for a creating a portal (though for the love of [whatever you hold in awe] nothing like the terrible bureaucracy surrounding wikiprojects and stub tags/categories, or even barnstars for that matter – we need to be killing off bureaucracy, not generating more of it). We don't need a portal for every conceivable topic. Being navigable with navboxes is one baseline, but maybe not sufficient to keep everyone happy. I would think that nothing that doesn't have navboxes should be a portal, but it doesn't necessarily imply that everything with a navbox must have one (especially pop-culture topics with navboxes and categories that are never going to grow larger, e.g. bands that put out 3 albums and 10 singles and then broke up, or TV shows that got cancelled). Just being about, say, food doesn't make it automatically non-portal-worthy, however. Now that I know Portal:Sandwiches exists, I'm apt to use it, since it might give me some mealtime ideas (damn am I getting tired of turkey and mustard). People use our site for all sorts of personal reasons, and that's not just perfectly fine, it's part of Wikipedia's intent and design.
That said, with portals now being rather robustly automated (after a lot of impressive technical work by a lot of people), the "redirects are cheap" rationale starts to apply to portals. If they don't cost us much of anything in editorial attention and other maintenance, and someone somewhere, like a potato-specializing horticulturalist, is going to love Portal:Potatoes, and it may inspire them to write some articles on notable cultivars and improve some existing ones, what exactly is the burning issue here? Just to pick at the list presented above, how on earth is anyone going to object to a portal on hunger relief (a major global social issue), beef (a massive industry), herbs and spices (one of the driving factors of international commerce and even wars and worse since prehistory), liquor (a huge industry, a human preoccupation since prehistory, and a complicated regulatory regime and public health issue), vegetarianism (a "big encyclopedia topic", though I would merge the veganism one into it as a minor variation), water (one of the most important and broadest topics we have, from social material like drinking water crises to basic science), and so on? The topic-opener's objection simply isn't sound when presenting such a list.
In the module Module:lang, in line number 1082, should the word language which is present inside single quotes be translated into local language while translating it into malayalam? I am getting around thousand errors as in the page [Hindi].Adithyak1997 (talk) 18:11, 24 November 2018 (UTC)
SECR N class - inconsistency between watchlist and page history
The article SECR N class has been on my watchlist since forever. It's on the main page right now as WP:TFA. The only changes to this page that are showing on my watchlist for the last 24 hours are the last six log entries. Yet when I look at the page history, there have been approximately 30 edits today. Several have been WP:REVDELled; but such edits should appear in my watchlist and either show as struck through or greyed out, in the same manner as the page history. Why are these edits - both good and bad - absent from my watchlist yet present in the page history? --Redrose64 🌹 (talk) 22:09, 24 November 2018 (UTC)
OK, my watchlist now shows edits timed at 22:40, 24 November 2018 (UTC) or later, but it still does not show the earlier ones. Could this be something to do with the deletion of the page by Killiondude (talk·contribs) and its subsequent restoration by Black Kite (talk·contribs) - would those actions have nixed the watchlist? --Redrose64 🌹 (talk) 13:07, 25 November 2018 (UTC)
I have no helpful answer, but I'll note that when I first tried to restore the article, it crashed with a red message that I can't quite recall but was something along the lines of "too busy". I tried it again and it (seems to have) worked. Black Kite (talk)13:50, 25 November 2018 (UTC)
WebCite has been hard-down since last Friday (current 5 days), and the week previous it was flaky.
If WebCite were simply going through growing pains it would be different. But of the 20 or so web archive providers on enwiki, WebCite has been the second-worst I have worked with while checking millions of links with WaybackMedic over the past years. I know User:Cyberpower678 (User:InternetArchiveBot aka IABot) thinks the same. All attempts at communication with them go unanswered. There are regular outages. Bugs in the software, API and data are never fixed. They have more archive link rot (dead archive pages) than any other web archive provider. No new features are added. The documentation and website have not changed in over 5 years. They have had funding problems in the past and almost closed down in 2013.
WebCite accounts for between 5% and 7% of all archives on enwiki. It is the second-largest archive provider on enwiki, behind Wayback. Recent metrics taken by the WMF show significant user click-through to archive links when reading an article, so this is important:
Wikipedia is outsourcing the core pillar of WP:V to a third-party company that is unreliable, unresponsive and out of our control.
IABot and WaybackMedic have the ability to find new archive providers and swap out WebCite links, should that be required. Replacements won't be found for all, but could be for a lot. This is not an RfC to do that, but an initial discussion for feedback and information about WebCite status. -- GreenC16:34, 20 November 2018 (UTC)
GreenC, yes please. IABot has a difficult time working with WebCite, to the point I have to implement aggressive caching to make IABot run more independently of WebCite. —CYBERPOWER(Chat)16:37, 20 November 2018 (UTC)
On the 7th day of WebCite not working, my true love gave to me.. half-a-million 404-errors (the apx number of dead webcite links on enwiki). -- GreenC16:41, 22 November 2018 (UTC)
Zoiks. I was wondering if this day would come. I assume there's some automatable way to replace it? Maybe a bot that checks multiple archivers (favoring Wayback Machine, I would think) and replaces calls to WebCite when a replacement is found, and another bot to keep testing a WebCite-only case until it works, then get Wayback (or whoever) to chain-archive THAT, and replace it, while it's presumably-temporarily accessible at WebCite? — SMcCandlish☏¢ 😼 02:56, 24 November 2018 (UTC)
WaybackMedic already does this for years. When it finds an archive at one provider is no longer working, it searches for another and replaces it. Automatically. But not when there is a total site outage, as currently with WebCite, then it skips. WC came back online today, for the moment. -- GreenC03:31, 24 November 2018 (UTC)
That was a pretty long outage, and potentially an indicator of things to come. Maybe we should start preparing for the worst, with a phase out. Half-a-million dead citation links is pretty serious. — The Transhumanist19:49, 25 November 2018 (UTC)
mw.util.jsMessage
Hi, the mw.util.jsMessage() function was deprecated in 2012, and will soon not be working. According to phab:P7840 there's at least one gadget using this function on your wiki, but it is likely it won't cause much of a problem anyway. We don't see this function being used much and this message is mainly to be on the safe side. There's a migration guide that explains how to use mw.notify instead. See phab:T193901 for more information. /Johan (WMF)
09:39, 26 November 2018 (UTC)
Filter blocked my edit but then it mysteriously happened anyway
This is the mysterious diff. I got a pink message before I submitted and it said I was trying to add something not allowed. What I hadn't noticed was when I copied and pasted the title of the source, the URL and "Read more here" also were copied. I removed those but never submitted them. Yet they somehow ended up being submitted anyway with the edit summary from my previous edit.— Vchimpanzee • talk • contributions • 17:44, 24 November 2018 (UTC)
According to the logs you got the warning at 17:03, saved without "Read more" at 17:04, and then saved the original edit with "Read more" at 17:08. Did you save the edit at 17:04 in the same tab as the warning? If you kept the warning in an open tab while saving in another tab at 17:04 then I guess you accidentally saved in the warning tab at 17:08. Your browser may have remembered the 17:04 edit summary and suggested it in the warning tab. Wikipedia:Edit filter#Basics of usage says: "The next lowest setting is to warn. In this case, the user will see a customisable message (this one by default) that the edit may be problematic. The user then has the option to either proceed with the save or abandon the edit." Special:AbuseFilter/702 displays MediaWiki:Abusefilter-warning-clipboard-hijacking in a pink box. If you saved at 17:04 in the same tab as the warning then the 17:08 edit does look odd to me. PrimeHunter (talk) 00:02, 26 November 2018 (UTC)
I raised a moderately severe security bug on Phabricator on 5 November. Other than someone allocating a project and changing my title, there's been no response.
@Pigsonthewing:: Hi Andy, if you think it is a serious issue and it is being ignored on phab, I suggest you try emailing trustandsafety@wikimedia.org first, especially if it is a bug in mediawiki core or a popular extension. — xaosfluxTalk23:54, 24 November 2018 (UTC)
Since I wrote the above, my ticket has been merged, as a duplicate with another, that was - remarkably - opened in 2008. Since that ticket is both long-lived and public, I'm now prepared to discuss the issue openly, here.
The spam blacklist system, and the abusefilter are heavily outdated. Likely over 10 years of neglect for the former, and a bit less for the latter. --Dirk BeetstraTC09:49, 26 November 2018 (UTC)
According to our article about hostnames, valid top-domains are ASCII only. Is there any reason to not simply reject any edit adding a link whose top domain contains anything non-ASCII? (Which solves the current issue, independently of an overhaul of the spam blacklist system.) TigraanClick here to contact me17:15, 26 November 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Recent changes
On wikis with translatable pages you could create a mess when you moved a page that had translatable subpages. A subpage is when you use / to create a new page: /wiki/Page/Subpage. The subpages would be moved but not the translations. The subpages are no longer automatically be moved. This is to make it safer to move pages. [62]
Changes later this week
The advanced search interface will be available by default on all Wikimedia wikis. It makes it easier to use some of the special search functions that most editors don't know exist. It's already active on German, Farsi, Arabic and Hungarian Wikipedia. [63]
Special:UnusedCategories show empty categories with no files or other categories. You can soon choose to not show soft redirect categories or some maintenance categories there. You can do this with the magic word__EXPECTUNUSEDCATEGORY__. [64]
The new version of MediaWiki will be on test wikis and MediaWiki.org from 27 November. It will be on non-Wikipedia wikis and some Wikipedias from 28 November. It will be on all wikis from 29 November (calendar).
Meetings
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 28 November at 16:00 (UTC). See how to join.
Future changes
The mw.util.jsMessage() function was deprecated in 2012. It will be removed next week. Look for the warningUse of "mw.util.jsMessage" is deprecated in the JavaScript console to know if you use an affected script or gadget. If you are a gadget maintainer you should check if your JavaScript code contains mw.util.jsMessage. There is a migration guide. It explains how to use mw.notify instead. [65]
Next version is 1.33-wmf.6, we're currently on wmf.4, and neither wmf.4 or wmf.5 have release notes (neither does wmf.6, but the latest never does for a few days). Not a huge issue — I presume the US holiday? — just wanted to note it. ~ Amory(u • t • c)22:59, 26 November 2018 (UTC)
Finding your way back from multi-referenced footnotes
When you click on a footnote which is referenced multiple times in an article, it can be hard to get back to your reading position in the text. Soon it will be easier to find your way back up:
You can click on the ^ jump mark at the beginning of the footnote (in other wikis it's an arrow). This wasn’t clickable before. The tooltip now says "Jump back up".
Or you can click on the superscript jump mark (e.g. a ) in the footnote. The one leading you back to your original position is highlighted bold. Please note that this part of the change won't be effective in enwiki, because jump marks here are bold by default. If your wiki wants to have highlights for secondary jump marks, the default style for them would need to be changed to regular.
Template:Cite web objects to lack of URL implied by doi
((Cite Journal)) and most other Cite templates allow omission of the |url= parameter. ((cite web)), ((cite podcast)), and ((cite mailing list)) responds to the lack of a |url= with Missing or empty |url= (help). But |url= is not needed when there is a link caused by parameters such as |doi=, |ssrn=, |pmid= and possibly others. National Association for Gun Rights has such an error erroneously reported for footnote 33, which follows "NAGR was cited by the court prior to the victory for gun owners." —Anomalocaris (talk) 19:17, 26 November 2018 (UTC)
This is by design. If you're using ((cite web)), ((cite podcast)) or ((cite mailing list)) correctly, then there must be url. If there's no url to link, then the templates should not be used. I fixed the one in National Association for Gun Rights just by supplying the url. Template ((Cite Journal)) allows neglecting the |url= because, the journal can well be offline/print only and that's valid scenario. There's no reasonable way to claim that a website or mailinglist is offline only –Ammarpad (talk) 19:48, 26 November 2018 (UTC)
Ammarpad: I agree that ((cite web)) means something is online and a valid online locator must be supplied. But |doi=, |ssrn= and |pmid= are all valid online locators that resolve to URLs. There is no need to supply a URL when the user can click the DOI, SSRN or PMID link. —Anomalocaris (talk) 21:26, 26 November 2018 (UTC)
@Anomalocaris: If you have one or more of those document identifiers such as DOI, PMID etc. the original source is almost certainly going to be an academic journal, so ((cite journal)) is the better choice. The ((cite web)) template is, to quote its own documentation, "for web sources that are not characterized by another CS1 template". In other words, it's a fallback for when none of the others is appropriate, and it should not be treated as a first choice. --Redrose64 🌹 (talk) 10:03, 27 November 2018 (UTC)
The mobile main page has the featured article title awkwardly positioned so that the title is splitting apart. I'm guessing it's something to do with the image. Home Lander (talk) 20:05, 27 November 2018 (UTC)
Home Lander, well sort of. It's just the word 'experienced' not fitting between the end of the caption and the right side of the screen. It's because people are using height autoscaling, which that mainpage image template doesn't support. Specifying the width, makes it work as it is supposed to. —TheDJ (talk • contribs) 20:29, 27 November 2018 (UTC)
What does 'span class="nickname"' actually do, technically?
I searched in the archives of this page, and I poked around all of en.WP and the MW site with various searches, but I came up empty-handed. I am trying to figure out what span class="nickname" does and why it is used to format some parameter values in infoboxes. I don't necessarily need a detailed explanation here if one is provided on another page.
The reason that I am curious is that when a set of <div>...</div> tags, or a template that incorporates them, is used in one of these infobox parameter values, we end up with a div wrapped inside a span, which results in a "div-span-flip" Linter error. Removing these span tags or changing them to div tags (or some other workaround/hack/solution) would fix errors on many thousands of pages. – Jonesey95 (talk) 21:36, 26 November 2018 (UTC)
Templates using that class should include in their documentation ((UF-hcard-geo)) or similar; which answers your question. You can change a <div>...</div> to a <span>...</span> or vice versa, without affecting the microformat (other things may break, of course). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits10:27, 27 November 2018 (UTC)
Super! That is pretty much what I was looking for. I have changed a few spans to divs, and typically, a straight swap works. I have found that in some places I need to use div style="display:inline" in order to prevent unwanted line breaks. (NB: div style="display:inline" seems like a contradiction in terms to me, but it certainly does the trick.) – Jonesey95 (talk) 13:00, 27 November 2018 (UTC)
That is because div's are block-level elements and span's are line-level elements. Semantically, div's contain sections "of the document" while spans contain sections "of content" (generally text). — xaosfluxTalk14:29, 27 November 2018 (UTC)
Re: fixing the right problem: I'm just fixing the problems that the WMF, in their infinite wisdom, has created for me to fix. That's how this works, right? I donate money to the WMF, and they use it to make a nice list of problems for me to fix. In comparison to the size of my donation, I view the WMF as wildly generous with their contribution. Do I wish they would fix some of the multi-year-old really annoying bugs that I have filed on phab over the years? Sure I do, but that's not how this system works. – Jonesey95 (talk) 15:16, 27 November 2018 (UTC)
I have actually resisted changing the span/div flip errors precisely because it's not obvious to me that we are using the parameters as documented correctly in many of the cases where span/div flip errors show up. --Izno (talk) 15:19, 27 November 2018 (UTC)
WP editors are endlessly creative. Take a look at Hospital (TransMilenio), for example, to see a div tag that ends up inside the Infobox's span tag for |name=. 15 Temmuz Kızılay Milli İrade (Ankara Metro) is another beauty. People put all manner of div-containing templates inside of |native_name= (and many other parameters), causing the span that is in a typical infobox's native_name data parameter to wrap a div. Examples include ((plainlist)) and its siblings. It's good fun to track all of these down. – Jonesey95 (talk) 23:01, 27 November 2018 (UTC)
For some reason, any optional fields in the video game infobox left empty are returning error messages regarding Lua, and even hiding the box art. igordebraga≠18:56, 21 November 2018 (UTC)
Another editor overwrote the en.wiki version of Module:I18n/complex date with a copy of same from commons. The commons version require()s a module that doesn't exist on en.wiki.
Hmmm, the <categorytree>...</categorytree> tags will list subcategories for a given category, but not other kinds of members. --Redrose64 🌹 (talk) 00:09, 28 November 2018 (UTC)
Hi. I noticed yesterday that every page I edit has a redlink for "page notice" in the top-right corner. I've tried to hide them following the instructions at WP:EDN, but this doesn't seem to work (I've done all the bypassing of the cache, etc). Any ideas? Thanks. LugnutsFire Walk with Me18:07, 28 November 2018 (UTC)
Looks like the HTML structure has changed; the CSS class "editnotice-area" isn't always present. Try just:
.editnotice-redlink{display:none!important;}
Also, I don't think the literal "skin.css" page does anything; it's supposed to redirect you to "<name of skin>.css", e.g. "monobook.css" if you're using Monobook. Try putting the above CSS snippet in your common.css page instead; then it should work in any skin. Writ Keeper⚇♔18:19, 28 November 2018 (UTC)
Ahhh, that explains it - thanks. Usually little things like this are due to some behind the scenes update that didn't go to plan, so I waited 24hrs incase it was rolled back. Thanks again. LugnutsFire Walk with Me18:55, 28 November 2018 (UTC)
Can't log on to UTRS
I don't look at UTRS very often, but I'm currently getting this error message when I try to log on:-
Fatal error: Uncaught exception 'UTRSDatabaseException' with message '<b>A database error occured when attempting to process your request: </b><br />array ( 0 => '23000', 1 => 1062, 2 => 'Duplicate entry \'<my email address redacted>\' for key \'email\'', )' in /usr/utrs/production/public_html/src/userObject.php:107 Stack trace: #0 /usr/utrs/production/public_html/src/userObject.php(78): UTRSUser->insert() #1 /usr/utrs/production/public_html/login.php(207): UTRSUser->__construct(Array, false, Array) #2 {main} thrown in /usr/utrs/production/public_html/src/userObject.php on line 107
Voice of Clam the tool mentions: "Note: If your Wikipedia account is renamed, you must create a new UTRS account or contact a tool developer to have your Wikipedia username changed in the database" and it's footer links to an email address for the people who maintain the tool. —TheDJ (talk • contribs) 15:43, 26 November 2018 (UTC)
@TheDJ: Sorry for being thick, but can you post a link to the page in question? I can't find what you're referring to, or instructions on how to create a new account. Thanks. 19:57, 28 November 2018 (UTC)
Advanced Search will become a default feature on your wiki on November 28. This new interface allows you to perform specialized searches on the search page, even if you don’t know any search syntax. Advanced Search originates from the German Community’s Technical Wishes project. It's already a default feature on German, Arabic, Farsi and Hungarian Wikipedia. Besides, more than 40.000 users across all wikis have tested the beta version. Feedback is welcome on the central feedback page.
Recently, when using the visual editor, the "Publish changes" button doesn't work. It's just greyed out and disabled even after I make changes. I'm using Chrome. Is this a known problem? Popcornduff (talk) 21:24, 26 November 2018 (UTC)
I don't normally use VE but I enabled it to write this reply. And you can see it worked. Are you still seeing the problem? –Ammarpad (talk) 06:38, 27 November 2018 (UTC)
Firefox only, English Wikipedia only, the second time you edit the same page. A temporary workaround might be opening a different page, and then going back to edit the first one again. The devs are working on it, but they didn't have it solved as of six hours ago. Whatamidoing (WMF) (talk) 00:34, 28 November 2018 (UTC)
Skin variants and themes - user styles, css overrides, etc
Hi, y'all, I'm currently working on an RfC about adding functionality to support built-in skin variants across all skins (so things like dark Vector etc could potentially be chosen as options under the skin in special:preferences). To help with this, though, more information on what sort of things you already have floating around/want to have would be quite helpful, and especially what folks really liked/dislikes/whatever with them. Like I think there was a dark vector gadget at one point? What else is/was there? Is this documented anywhere? -— Isarra༆18:32, 27 November 2018 (UTC)
Reminder: Technical RFCs are very much focused on the "comment" part, not the "voting" parts. They mostly focus on "how" to do something, not whether it would be popular. If you have some potentially useful technical or semi-technical information (no matter how small detail that might seem to you), then please post that information over there. Also, technical RFCs are also long by enwiki standards (some of them have stayed open for years), so if you can't think of anything today, but you think something in a month or two, you will likely still have an opportunity to contribute usefully. Whatamidoing (WMF) (talk) 00:20, 29 November 2018 (UTC)
Custom infobox cell color
Hoping someone can help me... I'm working on converting an old school infobox from table format to using ((Infobox)) and ((Infobox3cols)). Is there a trick to making a data or label a specific color? I'm looking at Template:Infobox_actor_awards/testcases. Notice on the left how "awards", "wins" and "nominations" are a different color then the rest. There is an argument that all these color violates MOS:COLOR and that is a discussion I'm planning to have later, but for now just trying to figure out how to technically achieve this. I know there is |rowstyle#= & |rowcellstyle#= but can't seem to get either of those to work the way I need. Any help appreciated. (Please ((ping|zackmann08)) if possible). --Zackmann (Talk to me/What I been doing) 21:05, 28 November 2018 (UTC)
@Zackmann08: You could try using TemplateStyles CSS, by e.g. putting a CSS class on the collapsible table and using a CSS selector which applies to the first/second/third td element in each tr. Jc86035 (talk) 13:53, 29 November 2018 (UTC)
Contributions list has multiple lines for each contribution