Wikipedia:Village pump (technical)

From Wikipedia, the free encyclopedia
 Policy Technical Proposals Idea lab WMF Miscellaneous 
The technical section of the village pump is used to discuss technical issues about Wikipedia. Bug reports and feature requests should be made in Phabricator (see how to report a bug). Bugs with security implications should be reported differently (see how to report security bugs).

If you want to report a JavaScript error, please follow this guideline. Questions about MediaWiki in general should be posted at the MediaWiki support desk. Discussions are automatically archived after remaining inactive for five days.

Wikimedia\Rdbms\DBQueryError[edit]

I got this error when trying to edit an article. I see this has happened before in 2020. Looks like someone at Wikipedia:Help desk#Database error when trying to edit an article has filed a ticket at Phabricator, tracked at phab:T352628. InfiniteNexus (talk) 07:17, 4 December 2023 (UTC)Reply[reply]

@InfiniteNexus I encountered the same bug about 10 minutes ago. Bug now, it seems that the this DB bug is resolved. Hooman Mallahzadeh (talk) 08:13, 4 December 2023 (UTC)Reply[reply]
I keep getting this error trying to edit Wellington (disambiguation). I'm able to edit other articles. olderwiser 12:05, 4 December 2023 (UTC)Reply[reply]
I just got it, and reproduced, trying to do a small edit on Handball (disambiguation), something's fishy here.
[880bbfb7-540e-49ff-bd80-5d1f0880e872] 2023-12-04 13:11:04: Fatal exception of type "Wikimedia\Rdbms\DBQueryError"
--Joy (talk) 13:12, 4 December 2023 (UTC)Reply[reply]
My new attempts there also get:
To avoid creating high replication lag, this transaction was aborted because the write duration (3.465854883194) exceeded the 3 second limit. If you are changing many items at once, try doing multiple smaller operations instead.
[5b6f2722-25fb-4736-a8a1-061b5bc2d481] 2023-12-04 14:31:18: Fatal exception of type "Wikimedia\Rdbms\DBTransactionSizeError"
--Joy (talk) 14:31, 4 December 2023 (UTC)Reply[reply]
JFTR that edit went through in the meantime. And Phabricator indicates the developers are figuring it out. --Joy (talk) 17:05, 4 December 2023 (UTC)Reply[reply]
I also had errors on two edits in the last couple of hours. I didn't record the first but the second was [29eec93d-299b-4bdf-89d1-88caca3466b5] 2023-12-04 14:13:41: Fatal exception of type "Wikimedia\Rdbms\DBTransactionSizeError" on this edit. Both worked after I used the browser's Back button and clicked Publish again. Certes (talk) 14:17, 4 December 2023 (UTC)Reply[reply]
Still getting an error on this. I was getting something about a transaction being too large and taking too long (>3 sec) for a couple attempts. This attempt I got
A database query error has occurred. This may indicate a bug in the software.
[4d773f04-424e-4f64-af0f-f259aa808ecd] 2023-12-04 14:44:28: Fatal exception of type "Wikimedia\Rdbms\DBQueryError"
Attempts to edit my user sandbox page were successful.
Kimen8 (talk) 14:45, 4 December 2023 (UTC)Reply[reply]
I've been getting this error this morning too: To avoid creating high replication lag, this transaction was aborted because the write duration (15.307519674301) exceeded the 3 second limit. If you are changing many items at once, try doing multiple smaller operations instead. [1cd8ddfe-f560-4300-b162-1d4d0448423c] 2023-12-04 16:19:03: Fatal exception of type "Wikimedia\Rdbms\DBTransactionSizeError" These aren't large changes I'm attempting. I have gotten a few edits through, but it's hit or miss. – Muboshgu (talk) 16:23, 4 December 2023 (UTC)Reply[reply]
It seems to be an issue with DB queries taking too long, see https://phabricator.wikimedia.org/T352628#9379558 for more information NW1223<Howl at meMy hunts> 16:35, 4 December 2023 (UTC)Reply[reply]
Got it when trying to edit Legislative Assembly of Costa Rica, but when I tried a second time, the edit went through fine. Cremastra (talk) 17:44, 4 December 2023 (UTC)Reply[reply]

I got this type of error just now trying to edit Luzerne County, Pennsylvania. At least now I know what the problem is. QuicoleJR (talk) 16:46, 4 December 2023 (UTC)Reply[reply]

Database error[edit]

I'm trying to revert this unsourced/WP:CRYSTAL edit to 2007 Formula One World Championship, but I keep getting error messages of the form:

A database query error has occurred. This may indicate a bug in the software.
[38a2ed1b-239f-4448-841c-e65ec46331ef] 2023-12-04 08:28:23: Fatal exception of type "Wikimedia\Rdbms\DBQueryError"

I've tried a couple of different edit summaries, and I'm able to edit other articles. Could someone else please try reverting the edit? Thanks. DH85868993 (talk) 08:32, 4 December 2023 (UTC)Reply[reply]

I tried to revert it and also got a DBQueryError. It looks like there is already a ticket filed on Phabricator; see Wikipedia:Village_pump_(technical)#Wikimedia\Rdbms\DBQueryError. Malerisch (talk) 08:46, 4 December 2023 (UTC)Reply[reply]
Am having the same problem at First Mithridatic War and another editor has reported similar problems at WP:Teahouse - Arjayay (talk) 11:52, 4 December 2023 (UTC)Reply[reply]
Have now edited First Mithridatic War - problem may be cleared/clearing? - Arjayay (talk) 12:46, 4 December 2023 (UTC)Reply[reply]
Me as well at Vista, California ... [ab2894fb-78ff-4187-a212-713464e91635] 2023-12-04 12:22:58: Fatal exception of type "Wikimedia\Rdbms\DBQueryError". Magnolia677 (talk) 12:25, 4 December 2023 (UTC)Reply[reply]
Same here, at Deer Park, New York: [a4dc9654-fbed-48be-8eb3-a597446272e9] 2023-12-04 12:32:07: Fatal exception of type "Wikimedia\Rdbms\DBQueryError" WikiDan61ChatMe!ReadMe!! 12:33, 4 December 2023 (UTC)Reply[reply]
And the same here (in the UK), from about 2 hours ago, editing Caligula, using MacBook. Sometimes it lets me edit but not save. Other times, it works fine but not for long. There's no pattern to it that I can discern. My Watchlist responds to changes in all other articles. Haploidavey (talk) 12:59, 4 December 2023 (UTC) Just tried a test edit, and was treated to the following:Reply[reply]
I am getting this with many different articles. Mellk (talk) 13:21, 4 December 2023 (UTC)Reply[reply]
Database error

To avoid creating high replication lag, this transaction was aborted because the write duration (5.7858679294586) exceeded the 3 second limit. If you are changing many items at once, try doing multiple smaller operations instead.

[b08b554e-cf70-4cd8-8291-03dd2733d85a] 2023-12-04 13:01:46: Fatal exception of type "Wikimedia\Rdbms\DBTransactionSizeError"

Haploidavey (talk) 13:08, 4 December 2023 (UTC)Reply[reply]

This is the kind I'm getting, and lots of them:

Database error

A database query error has occurred. This may indicate a bug in the software.

[74e1f80b-d290-4d16-836e-3c887005d683] 2023-12-04 14:01:55: Fatal exception of type "Wikimedia\Rdbms\DBQueryError"

 — SMcCandlish ¢ 😼  14:08, 4 December 2023 (UTC)Reply[reply]

Getting the same issue with the article Martin Artyukh. [85a67d40-833b-409f-a662-ba0486de9b09] 2023-12-04 15:51:16: Fatal exception of type "Wikimedia\Rdbms\DBQueryError" --BlameRuiner (talk) 15:52, 4 December 2023 (UTC)Reply[reply]
Yup, I'm getting the same error across multiple articles - sometimes takes multiple attempts to get a change to save. Parsecboy (talk) 16:23, 4 December 2023 (UTC)Reply[reply]

Recovering from it?[edit]

When this error hits me, if I go "back" in the browser to get back to the text I was working on, all my changes are gone. This is actually weird behavior in Chrome. E.g., if I instead get an edit-conflict page, I can "Back" in the browser freely. But when this DB error happens, it's like it somehow messes up the page cache. Does anyone know of a way to recover the text that was being worked on? I have an article that did a whole lot of cleanup work in (probably a good half hour of it) and don't want to lose all that work if I can help it. I'm still sitting on the DB error page on that one.  — SMcCandlish ¢ 😼  14:08, 4 December 2023 (UTC)Reply[reply]

Update: I took a gamble, and it turns out that clicking the Reload button worked (caveats: in Chrome, and without triggering another DB error; I have no idea what another browser would do, or what would happen if the DB error had recurred, and it might, since in trying to make a typo fix at another page I had to try five times, though each time I did it as a manual edit, not a reload).  — SMcCandlish ¢ 😼  14:28, 4 December 2023 (UTC)Reply[reply]
FWIW in my Firefox, the back button got me back into the previous state, with the content and summary fields intact, so I can keep retrying trivially. --Joy (talk) 14:33, 4 December 2023 (UTC)Reply[reply]
I'm in Microsoft Edge, and the back button gets me to the state pre-submission, with edits and summary in place. Though I am now taking a copy of the text before clicking Publish. Tacyarg (talk) 14:55, 4 December 2023 (UTC)Reply[reply]
I'm using Chrome on a chromebook and if I hit the Back button, I get the original article state from when I hit edit (i.e., I lost my changes and edit summary). I'm just going to wait until it seems to be resolved before making any more edits. Kimen8 (talk) 14:58, 4 December 2023 (UTC)Reply[reply]

Fixed?[edit]

phab:T352628 claims the issue is fixed. I went back to retry my edit at Deer Park, New York and was successful. WikiDan61ChatMe!ReadMe!! 18:06, 4 December 2023 (UTC)Reply[reply]

It is indeed fixed. NW1223<Howl at meMy hunts> 18:11, 4 December 2023 (UTC)Reply[reply]
Yep, fixed now. InfiniteNexus (talk) 18:54, 4 December 2023 (UTC)Reply[reply]

Very weird issue with the reply tool[edit]

The reply tool is doing some rather dumb and bad stuff and I do not know how to fix it. I'm using the mobile website (the app, while nice for reading, is more or less unusable for editing and administrative tasks) on Android. The issue is with saved comments. If the reply tool has a saved partial comment, it will disable itself for any other comment on the page. Fine. But it will also gray out the expand button for the section it's in! It's impossible, then, to access the comment I have saved, or even to read the section it's in. It's also impossible to discard the comment so that the page works normally. The only way to fix it is to switch to the desktop display, then edit the URL to be the desktop URL, then scroll down to the section and cancel the reply, then go back to the mobile version and it will work again.

I don't think a noob is gonna figure this out.

Yall heard about this??? jp×g🗯️ 20:41, 30 December 2023 (UTC)Reply[reply]

@JPxG: I’ve also been having this issue, and the only thing that occasionally fixes it (ie. opens the section and allows me to continue typing the reply in mobile mode) is repeatedly refreshing the page until it decides to work. Wonder if it’s worth a phab ticket (unless it’s already got one?). Best, ‍—‍a smart kitten[meow] 21:34, 30 December 2023 (UTC)Reply[reply]
Oh yeah, ReplyTool does this to me too. I think I've been able to open the source editor and publish an edit to the page, which restores my ability to discard the saved comment. It will trigger the inaccessible comment subroutine any time I close a tab with a comment neither published nor discarded. Folly Mox (talk) 01:41, 5 January 2024 (UTC)Reply[reply]
When I ran into this problem, my solution was to click "edit full page" and make a dummy edit under the section header 47.188.8.46 (talk) 21:58, 5 January 2024 (UTC)Reply[reply]
purging did nothing btw 47.188.8.46 (talk) 21:58, 5 January 2024 (UTC)Reply[reply]

Man the reply tool is really "on some other shiznit" as the kids say, I can't reply to this either. jp×g🗯️ 21:42, 30 December 2023 (UTC)Reply[reply]

Is this phab:T338920? I searched Phabricator and found this other ticket, which looks like the issue in question; but that was closed in favour of re-opening the aforementioned task. Best, ‍—‍a smart kitten[meow] 02:08, 3 January 2024 (UTC)Reply[reply]

Pages with most transclusions of a template[edit]

Is there a way to easily identify which page has the most transclusions of a particular template? For example, which article has the most {{citation needed}} tags? Nikkimaria (talk) 00:35, 1 January 2024 (UTC)Reply[reply]

Nikkimaria, I am fairly sure the answer is no, because that would require looking at the source text. — Qwerfjkltalk 09:52, 1 January 2024 (UTC)Reply[reply]
insource:... is a longstanding feature of the search interface. If you want to make the servers fall over, you could craft a regex (see mw:Help:CirrusSearch#Regular_expression_searches) to look for "at least 100" of a template. Then depending on how many hits you get, look for "at least 200" or "at least 50", gradually narrowing down the largest number of uses. Maybe the WP:Quarry magicians can help? DMacks (talk) 10:13, 1 January 2024 (UTC)Reply[reply]
Quarry won't be able to help with this, because it has no access to the wikitext. The nearest it could manage is to list pages with one or more tags, without saying how many tags each page has, which is insufficient and more easily done by other means. Cirrus search has its own flavour of regular expression which lacks many features found in most flavours. In particular, Cirrus lacks a syntax such as (Foo){100} meaning 100 occurrences of Foo. It would have to be written out in full, grossly exceeding the 300-character limit for search expressions. It's a job for a (simple) custom program running on a database dump. Certes (talk) 12:32, 1 January 2024 (UTC)Reply[reply]
I agree this would be easy on a local dump:) But the CirrusSearch docs say grouping and repetition-count are supported. And I just verified that my test page containing the string:
foo wocka foo fiddle foo banana foo more
was found by
insource:/(foo.*){XXX}/
when XXX was any single number 1–4 (but not any larger) or any range of numbers (comma-separated) that included at least any subset of 1–4 (but not if it only included 5+). Not surprisingly,
insource:/(\{\{citation needed.*){10,25}/
timed out. But before doing so it did find >20 results and spot-checking they do have at least 10 CNs. DMacks (talk) 13:23, 1 January 2024 (UTC)Reply[reply]
Hey, let's play kill-the-wiki!
insource:/(\{\{citation needed.*){100}/
limited to mainspace timed out after finding 8 results:
so there are at least these articles with 100+ literal CN tags in their wikisource. In Kazuhiko Inoue, they are all in a commented-out table. DMacks (talk) 13:46, 1 January 2024 (UTC)Reply[reply]
Thanks, I never knew Cirrus supported {n} – I always wondered why { and } were special characters! A shorter version of kill-the-wiki
citation insource:/(\{\{[Cc]itation needed.*){100}/ prefix:A [1]
produces one result and might find most cases if repeated for B–Z (and prefix:2, which has several). Certes (talk) 14:24, 1 January 2024 (UTC)Reply[reply]
Now just need to find out what < and > do.— Qwerfjkltalk 16:31, 1 January 2024 (UTC)Reply[reply]
< and > are for number ranges, e.g. <1998-2002> finds any of those five years. Certes (talk) 20:05, 3 January 2024 (UTC)Reply[reply]
I suppose one added problem here is that 550k+ pages transclude {{citation needed}} but another 35k have {{fact}} & 97k have {{cn}} (plus a couple of thousand in total using one of the more obscure redirects). I guess about 20% of uses are using a variant form, and presumably some fair chunk of articles that have built up tags over time will use a mixture of formats? Andrew Gray (talk) 18:32, 3 January 2024 (UTC)Reply[reply]
There are 70 redirects from {{Are you sure?}} to {{Uncited}}. In theory, we could check them all. In practice, only the two you mention are probably worth the bother.
hastemplate:"citation needed" insource:/(\{\{([Cc]itation needed|[Cc][Nn]|[Ff]act)[^a-z].*){100}/ prefix:A [2]
traps a couple more, such as Akira Ishida which has 80 {{citation needed}} and 83 {{cn}}. Certes (talk) 20:13, 3 January 2024 (UTC)Reply[reply]

Dumb author name scraping[edit]

Somehow, in Integrated Software for Imagers and Spectrometers, this (basically, "Alessandro Frigeria,d,*, Trent Hareb, Markus Netelerc, Angioletta Coradinia, Costanzo Federico d, Roberto Oroseia"), got scraped into Wikipedia as "Alessandro Frigeria,d,n, Trent Hareb, Markus Netelerc, Angioletta Coradinia, Costanzo Federicod, Roberto Oroseia". This sort of thing could potentially leave readers searching for the "Trent Hareb" or "Costanzo Federicod" erroneously listed here. Any thoughts on how to find and fix instances of this? BD2412 T 14:45, 1 January 2024 (UTC)Reply[reply]

It happened in [3]. The link syntax was completely wrong and clearly not made by a tool so I assume the user manually copy-pasted from the PDF. I don't see a practical way to discover similar human errors. This user hasn't edited since 2020. I don't know whether any of our tools can make similar errors. PrimeHunter (talk) 22:50, 1 January 2024 (UTC)Reply[reply]
Thanks, I am astounded. BD2412 T 03:49, 2 January 2024 (UTC)Reply[reply]
Not surprising. Entry of long lists of journal author names is error prone for this reason, cut and paste then fail on manual cleanup. I've had this problem personally (though not this severe). It should be possible for a bot to retrieve the names from a source (DOI?) and compare it with the names in the citation. It could verify not only misspellings but also normalize abbreviations "James A. P. Smith" vs "James AP Smith", check for missing names, and for long lists with et al. create something like Footnote #5 in Rising Star Cave. It would have be a tool or usertool with diff preview. We have all sorts of issues like this and a lack of programmers to make the tools. Another one is a tool to generate |author-link=. A tool to convert |author= to |first= / |last=. It goes on. -- GreenC 04:20, 2 January 2024 (UTC)Reply[reply]
Formatting citations is one of the most irritating parts of Wikipedia editing. I suspect that one's eyes glaze over after only a little time and then such errors happen. Jo-Jo Eumerus (talk) 08:39, 2 January 2024 (UTC)Reply[reply]
Those who are focused on this kind of work, automated citation maintenance, could get masters degrees. This is no joke. For example the history of {{cite encyclopedia}} and its polysemic |title= field could easily fill at least a class or two. A masters thesis on how to rewrite this template and how to implement those changes, technically and politically, would be most helpful. We have thousands of other citation template varieties to consider. -- GreenC 18:02, 3 January 2024 (UTC)Reply[reply]
PrimeHunter, I suppose you could look for a, b, c etc. at the end of author names. — Qwerfjkltalk 10:15, 2 January 2024 (UTC)Reply[reply]

Massviews[edit]

I would like to run off a list of the most popular articles under Category:WikiProject Military history articles in 2023 but massviews is limited to 20,000 "for performance reasons" (getting the wrong answer more quickly). Is there a way to run this query? Hawkeye7 (discuss) 02:41, 4 January 2024 (UTC)Reply[reply]

You can derive it from the revision history of Wikipedia:WikiProject Military history/Popular pages, at least when the December 2023 page view stats are out ... which will probably be in a few days. Many other WikiProjects have such bot-generated lists, collated at Category:Lists of popular pages by WikiProject. Graham87 (talk) 16:59, 4 January 2024 (UTC)Reply[reply]

Lua errors[edit]

Lua errors for not enough memory seem to be popping up in articles - Google Chrome is an example. I expect this may quickly become an UBN in Phabricator but I wanted to drop a note about it in VPT. Best, ‍—‍a smart kitten[meow] 00:05, 5 January 2024 (UTC)Reply[reply]

Okay, on further inspection it might just be the Google Chrome article, and this thankfully might not be as much of a problem as immediately thought. ‍—‍a smart kitten[meow] 00:14, 5 January 2024 (UTC)Reply[reply]
I previewed {{Infobox software}} alone with no parameters on the article and "Parser profiling data" at the bottom said Lua memory usage 49,651,978/52,428,800 bytes. That's 95% of the limit. PrimeHunter (talk) 00:20, 5 January 2024 (UTC)Reply[reply]
The problem is d:Q777. Infobox software is running out of memory because it's trying to get data from a 2MB wikidata page for the version numbers. Fixed here. – Hilst [talk] 00:36, 5 January 2024 (UTC)Reply[reply]
In case anyone's interested, this is a screenshot of what the article looked like with the Lua errors. Best, ‍—‍a smart kitten[meow] 00:39, 5 January 2024 (UTC)Reply[reply]
Sidenote -- in my opinion, even though the immediate error has been resolved, it's still worth this being looked into if anyone has the time (and/or patience), as I don't know if it should be possible to break an article on enwiki through good faith additions to its item on Wikidata. I'll put a link to this section at Template talk:Infobox software as that seems to be the template that temporarily broke the article. Best, ‍—‍a smart kitten[meow] 00:58, 5 January 2024 (UTC)Reply[reply]

Vector (2022) still has limited width with the skin setting turned off[edit]

The thing is, if you click the preview link in Preferences, it displays correctly as if the setting is in effect, but otherwise it is not. KPu3uC B Poccuu (talk) 04:58, 5 January 2024 (UTC)Reply[reply]

Somebody? KPu3uC B Poccuu (talk) 08:54, 7 January 2024 (UTC)Reply[reply]
I'm a bit confused about what you are reporting.
  1. Are you using the desktop interface?
  2. Do you have a global skin defined here: Special:GlobalPreferences#mw-prefsection-rendering? If so what is that settings?
  3. What are your settings in: Special:Preferences#mw-prefsection-rendering
    1. What do you have "Skin" set to?
    2. If set local exception is available, is it set?
    3. What is Enable limited width mode set to?
    4. Have you saved your preferences?
  4. If set above, and saved - what are you seeing that is different from what you expect to see?
. — xaosflux Talk 12:25, 7 January 2024 (UTC)Reply[reply]

Vector legacy (2010) drawing infoboxes very poorly[edit]

This is how infoboxes look like in legacy since yesterday. All other skins show infoboxes ok.   ~ Tom.Reding (talkdgaf)  09:28, 5 January 2024 (UTC)Reply[reply]

Have you tried to Bypass your cache ? It looks like one of the css files is missing, maybe it was cached incorrectly in your browser. —TheDJ (talkcontribs) 12:56, 5 January 2024 (UTC)Reply[reply]
(edit conflict) @Tom.Reding: That's the safemode look of Wikipedia when some CSS isn't loaded. It works for me in Firefox. Make sure "Always enable safe mode" is disabled at Special:Preferences#mw-prefsection-rendering (it probably already is when other skins work). Try to bypass your cache. Use Ctrl+F5 in Windows browsers, not F5 or the reload button alone. If it still happens then does it go away if you log out? What is your browser? What did you do to get black background? PrimeHunter (talk) 12:58, 5 January 2024 (UTC)Reply[reply]
@TheDJ and PrimeHunter: wonderful, thank you! It was a cache issue in Chrome. I use the Dark Reader browser extension to get a nice, comfy, black background, and User:Tom.Reding/Night-mode style.css in AWB.   ~ Tom.Reding (talkdgaf)  13:08, 5 January 2024 (UTC)Reply[reply]

Make use mdy dates have effect on the as of template[edit]

I can't figure out how {{use mdy dates}} works, and could somebody make {{as of}} automatically change date formats based on it? Aaron Liu (talk) 22:26, 5 January 2024 (UTC)Reply[reply]

The primary function of that template is to assign a tracking category. Secondarily, when Citation Style 1 templates (like {{cite web}}) are used, date formats in those citation templates are rendered to match the {{use mdy dates}} or {{use dmy dates}} template, as described in the template's documentation. Making {{as of}} behave like {{cite web}} automatically is probably not trivial, but there are some good coders around here. Since {{use mdy dates}} or its sibling template should be stable on a given page, adding |df= to a given {{as of}} template transclusion shouldn't be that much work. – Jonesey95 (talk) 23:48, 5 January 2024 (UTC)Reply[reply]
How do the citation modules detect the dateformat templates? I couldn't find anything in the source code. Do they just scan the page content for the template syntax? Aaron Liu (talk) 00:27, 6 January 2024 (UTC)Reply[reply]
See Module:Citation/CS1/Configuration for detection and Module:Citation/CS1/Date validation for reformatting. As I said, not trivial. – Jonesey95 (talk) 01:16, 6 January 2024 (UTC)Reply[reply]
Not so difficult really. Because the date-format detection code is in a module that is loaded using mw.loadData ('Module:Citation/CS1/Configuration'), it's relatively inexpensive to write a small module that will use the global_df value that ~/Configuration creates. See Module:Sandbox/Trappist the monk/as of.
My sandbox has {{use mdy dates}} and live and sandbox versions of {{as of}}. {{as of/sandbox}} auto-formats to the mdy date format.
Just because this is relatively easy to do does not necessarily mean that we should be doing it. That is a question for a different venue.
Trappist the monk (talk) 16:11, 6 January 2024 (UTC)Reply[reply]

Vector 2022 Sidebar problem[edit]

The way Vector 2022 displays seems to have altered on my PC screen, though not on my Android device. On my PC I use the Vector 2022 Skin on Chrome, on Windows 10; on my Android I also edit using the Chrome browser rather than an app, using Desktop view with the Vector 2022 skin. Since yesterday, my PC has displayed pages differently, with a large left sidebar rather than a drop-down toolbar under the search bar. Pages still display correctly on my Android, and I have not altered my preferences. I am not aware of any change I have made on my PC which could cause this behaviour. Can anyone suggest what might have happened to cause this and how I can recover my preferred display without the obtrusive sidebar? RolandR (talk) 22:30, 5 January 2024 (UTC)Reply[reply]

Did you try clicking the "hide" link at the top of the sidebar? – Jonesey95 (talk) 23:49, 5 January 2024 (UTC)Reply[reply]
There doesn't seem to be one. RolandR (talk) 03:50, 6 January 2024 (UTC)Reply[reply]
I have discovered that if I log out of Wikipedia, the Hide link appears and I can close the sidebar. But as soon as I log in again, the sidebar reappears, and the Hide does not stick. There seems to be something in my settings which blocks this, but I can't figure out what, or why it should suddenly have started. RolandR (talk) 23:28, 7 January 2024 (UTC)Reply[reply]
Sounds like it is maybe a gadget or a user script. You have a "User:PleaseStand/hide-vector-sidebar.js" in your common.js, maybe that's it ? —TheDJ (talkcontribs) 09:28, 9 January 2024 (UTC)Reply[reply]

android app login[edit]

How do I log into the Android app? The only settings I see are "Customize toolbar" Uhoj (talk) 01:45, 6 January 2024 (UTC)Reply[reply]

Uhoj, it's a bit hidden when you've got an article open. First, find and tap the "Explore" button in either the bottom bar or the top right three-dot menu (screenshot). That should bring you to a page with a "More" button in the bottom right corner of the screen (screenshot). Tap that and you should see a menu with a login button. Rummskartoffel 11:12, 6 January 2024 (UTC)Reply[reply]
I really appreciate your help Rummskartoffel! All logged in now :) — Preceding unsigned comment added by Uhoj (talkcontribs) 16:07, 6 January 2024 (UTC)Reply[reply]

Italicized titles cut off by the hamburger menu[edit]

Are the serifs of the first letter of these titles being cut off by the hamburger menu in Vector 2023 for anyone else? An American Journey, Jane's Defence Weekly, Zhou Enlai: The Last Perfect Revolutionary. Only affects the letters A, J and Z from what I can tell. (Edge, Windows 11) Schierbecker (talk) 02:29, 7 January 2024 (UTC)Reply[reply]

Not for me. —TheDJ (talkcontribs) 09:30, 9 January 2024 (UTC)Reply[reply]

DOI links give generic citations in VE citation tool[edit]

Hi all

I'm working on a partnership with FAO (the part of the UN that works on Food and Agriculture). I have a question about the VE citation tool, when I try to cite their links which all use a DOI eg https://doi.org/10.4060/cc7937en the citation tool doesn't give the name of the publication, just 'publication preview page'. Can someone tell me why this is happening so that I can ask them to change it so their DOI links work with the VE citation tool?

[1]

Thanks very much

John Cummings (talk) 06:19, 7 January 2024 (UTC) John Cummings (talk) 06:19, 7 January 2024 (UTC)Reply[reply]

References

  1. ^ "Publication preview page | FAO | Food and Agriculture Organization of the United Nations". FAODocuments. doi:10.4060/cc7937en. Retrieved January 7, 2024.
Might be an issue with the information passed on to Zotero, my understanding is that Citoid draws its information from there. Jo-Jo Eumerus (talk) 07:18, 7 January 2024 (UTC)Reply[reply]
Thank you Jo-Jo Eumerus is there anywhere I can ask or anyone I can ask who would know about this and specifically what to change at their end to make this work properly? John Cummings (talk) 08:53, 7 January 2024 (UTC)Reply[reply]
Just a quick question, are you using "10.4060/cc7937en" or "https://doi.org/10.4060/cc7937en" when you automatically generate the cite?
If you include the https:// part it generates a cite web:
"Publication preview page | FAO | Food and Agriculture Organization of the United Nations". FAODocuments. doi:10.4060/cc7937en. Retrieved January 7, 2024.
If you just use the doi (10.4060/cc7937en) it realises you want a cite report and generates:
In Brief to The State of Food and Agriculture 2023 (Report). FAO. November 6, 2023. doi:10.4060/cc7937en.
I'd guess citoid automatically thinks you want cite web if you use a URL. -- LCU ActivelyDisinterested «@» °∆t° 14:17, 7 January 2024 (UTC)Reply[reply]
Thanks very much ActivelyDisinterested, this is a great workaround :) John Cummings (talk) 05:44, 9 January 2024 (UTC)Reply[reply]
And it looks like it is gathering information from the metadata of the target page, e.g. <meta property="og:title" content="Publication preview page | FAO | Food and Agriculture Organization of the United Nations" />. —  Jts1882 | talk  14:25, 7 January 2024 (UTC)Reply[reply]
Thaks very much Jts1882, really helpful. John Cummings (talk) 05:44, 9 January 2024 (UTC)Reply[reply]

Thanks very much Jts1882, ActivelyDisinterested and Jo-Jo Eumerus I've created a Phab task to request that urls to DOI.org are treated as DOIs, this should hopefully produce much higher quality references on Wikipedia for people using DOIs in this way. John Cummings (talk) 06:08, 9 January 2024 (UTC)Reply[reply]

Edit summary memory[edit]

Hey, everyone. I did a disk clean-up the other day and appear to have checked too many boxes. My memorized edit summaries are gone and, worst of all, they do not seem to be being recorded again. It means that I have to type out all of my detailed edit summaries every time I edit instead of using autocomplete. What do I do? Surtsicna (talk) 13:09, 7 January 2024 (UTC)Reply[reply]

On my computer, those edit summaries are remembered by my web browser (Firefox), not by Wikipedia. See this help page for instructions on how to tell a browser to remember form field entries. – Jonesey95 (talk) 14:40, 7 January 2024 (UTC)Reply[reply]
Well, you could get a list of your edit summaries and put them into your own Default Summaries gadget (paste it in lines 20-24 and lines 27-33), at least until your browser remembers them again. Do you use all of your edit summaries, or is there an delimiter, like last x days or top x used ? Snævar (talk) 16:12, 7 January 2024 (UTC)Reply[reply]

spacing between non-indented paragraphes and indented lines[edit]

What I'm seeing

As an example: Talk:Jess Bush

Under the headers, the spacing between the first paragraph and the following colon-indented line is larger than the spacing between subsequent colon-indented lines. There is no excess space in the code. I'm using Safari 17.2.1 on macOS Sonoma 14.2.1. — Fourthords | =Λ= | 17:29, 7 January 2024 (UTC)Reply[reply]

I can replicate this, at least in Vector 2022 (logged in or logged out, in Firefox and Brave for Mac OS). The colon-indented paragraphs are in one big block whose CSS includes
.ns-talk .mw-body-content dd {  margin-top: 0.4em;  margin-bottom: 0.4em; }
The opening (non-colon-indented) paragraph's CSS includes
.vector-body p { margin: 0.5em 0 1em 0; }
the relevant portion of which translates to
margin-bottom: 1em;
Someone more versed in site-wide CSS will have to take it from here, or debunk my research with a better explanation. – Jonesey95 (talk) 18:29, 7 January 2024 (UTC)Reply[reply]
Thanks for getting much further than I did! Is this the/a correct venue for this, or should I have posted it somewhere else? — Fourthords | =Λ= | 15:07, 8 January 2024 (UTC)Reply[reply]
It makes general sense -- a top-level block of text is a <p>, but the : syntax creates (nested) definition-lists and the text within them will just be contained in <dd>s. CSS that applies spacing to paragraphs won't automatically also apply to list items.
It's plausible that we should have Vector 2022 incorporate its own version of that override to the .ns-talk list item spacing. DLynch (WMF) (talk) 23:52, 8 January 2024 (UTC)Reply[reply]
(Or amend the site CSS to match the presumably-new paragraph spacing in Vector 2022.) DLynch (WMF) (talk) 23:57, 8 January 2024 (UTC)Reply[reply]

Map in infobox on Bowery isn't displaying[edit]

Anyone know why this might be happening? Jake Wartenberg (talk) 19:11, 7 January 2024 (UTC)Reply[reply]

This problem also affects Park Row (Manhattan), Worth Street, and Mott Street and probably many others. Worth Street is worst: I suspect a problem with the Wikidata that is read by {{infobox street}}. --Redrose64 🌹 (talk) 19:46, 7 January 2024 (UTC)Reply[reply]

Wikipedia Mobile website problem[edit]

The left sidebar buttons on Wikipedia mobile don't work, when I push on them the list simply hides again/exits. I've tried this on multiple Android phones all running the latest Android Google Chrome version and have ran into the same problem everytime. I can't even log in from the mobile site, I have to first scroll down to the bottom to force desktop then I can log in and then go back to mobile after. I have no problems editing articles using the mobile version however. Bzik2324 (talk) 19:55, 7 January 2024 (UTC)Reply[reply]

@Bzik2324: I can reproduce this on both Firefox and Chrome on Android 14. And yet, you've asked this question twice without a single "me too". So there must be something unusual about our settings. Do you have animations turned off under Android Settings -> Accessibility -> Visual Enhancements -> Remove animations? Does the boat move at WP:Articles for deletion/Ever Given? If you turn animations back on, does the problem go away? Suffusion of Yellow (talk) 22:15, 8 January 2024 (UTC)Reply[reply]

 You are invited to join the discussion at Template talk:Contains special characters § Displaying only when needed. {{u|Sdkb}}talk 04:59, 8 January 2024 (UTC)Reply[reply]

Does anyone know how to change the RefToolbar so that it is compatible with sfn?[edit]

Currently, the Wikipedia:RefToolbar/2.0 outputs a template within ref tags. Is there anyone who knows how to write an userscript or gadget with similar functionality that instead outputs a template w/o ref tags, but with a pre-filled {{sfn}} output? I don't know anything about how to write JS. Jo-Jo Eumerus (talk) 07:55, 8 January 2024 (UTC)Reply[reply]

I don't think VE has anything in the cite-making for making sfn:s either. Gråbergs Gråa Sång (talk) 10:30, 8 January 2024 (UTC)Reply[reply]
I'll ask a few technically versed editors if they know of any such script, or how to make one. Jo-Jo Eumerus (talk) 06:51, 9 January 2024 (UTC)Reply[reply]

Adaptive multiple tables[edit]

If an article has several small tables, it is often "nicer" to have them displayed in a row rather than in a narrow vertical column. This fine for large screens, but on a mobile or other small screen it is better to have the tables float into several rows to match the available display. To do this, I have used an inline-table style like this:

<div><ul><li style="display: inline-table; margin-right:1em;"> ..TABLE.. </li><li style="display: inline-table; margin-right:1em;"> ..TABLE.. </li></ul></div>

This does indeed work. However, is there a better way? Is there a more "official" way? — GhostInTheMachine talk to me 14:38, 8 January 2024 (UTC)Reply[reply]

Can't you just use {| style="float:left" on each table? —  Jts1882 | talk  15:01, 8 January 2024 (UTC)Reply[reply]
I can! I used float:left; margin-right:1em; That works too and is a lot simpler. Thanks. Any view about an "official" way? — GhostInTheMachine talk to me 17:06, 8 January 2024 (UTC)Reply[reply]
See Help:Table/Advanced#Side by side tables. PrimeHunter (talk) 17:13, 8 January 2024 (UTC)Reply[reply]

Google curiously refusing to index a page[edit]

I've noticed this curious case regarding Google's behaviour in indexing Wikipedia. The page Elephant duel was created as a redirect to War Elephant in 2009. I converted it into an article on 3 October 2023, but so far Google has refused to show it in its search results, not even when specifying the Wikipedia domain in the query. However, it is shown as a knowledge panel. Other search engines behave normally, so the problem lies with Google, but I'm curious if anyone has an idea of what might be going on? The article isn't a new page, so it shouldn't have anything to do with the 90-day noindexing. --Paul_012 (talk) 16:32, 8 January 2024 (UTC)Reply[reply]

Pages which are newly converted from redirects must go through NPP the same as others. Izno (talk) 19:35, 8 January 2024 (UTC)Reply[reply]
Paul_012 is autopatrolled since 2017 so an article creation should allow indexing right away. Does that not apply when autopatrollers convert a redirect? PrimeHunter (talk) 21:06, 8 January 2024 (UTC)Reply[reply]
Also it's already beyond 90 days, and the knowledge panel shows up long before then (as well as the Bing results). --Paul_012 (talk) 04:51, 9 January 2024 (UTC)Reply[reply]

Searching for the source of the user-facing "email this user" guidance[edit]

Where is the text in the box at the top of the Email this user special page coming from?

Searching mw:Codesearch for "confidential subject" finds nothing. Searching for "private log" finds several uses, but "private log will record" finds nothing.

Presumably SpecialEmailUser.php is the code that runs the special page for sending email, but if the boxed message at the top of that page isn't part of the code base, where does that code get it from? This is not easy to figure out. How does public function sendEmailForm() generate the "send email form"? Is the box above the form part of the form, or separate from it? wbm1058 (talk) 18:39, 8 January 2024 (UTC)Reply[reply]

A test of that page using ?uselang=qqx here shows that the box at the top comes from the interface message MediaWiki:Emailpagetext Aidan9382 (talk) 18:49, 8 January 2024 (UTC)Reply[reply]
Yes, and the message is customized here at the English Wikipedia. The default message can be seen at MediaWiki:Emailpagetext/qqx. See WP:QQX for tips to find system messages. PrimeHunter (talk) 21:13, 8 January 2024 (UTC)Reply[reply]
At least it's an intuitive shortcut. — Qwerfjkltalk 21:17, 8 January 2024 (UTC)Reply[reply]
Right. QQX. I have no idea what that acronym stands for. Thanks for your help. I recall being shown this trick before, but had forgotten the specific syntax that was used.
I guess this is the line of code that findsMediaWiki:Emailpagetext, since it's the only place I found 'emailpagetext' in the code:
->addPreHtml( $this->msg( 'emailpagetext', $this->mTarget )->parse() )
Should add Special:AllMessages to the code base that mw:Codesearch searches, so that it finds the system messages that are searched for. – wbm1058 (talk) 22:59, 8 January 2024 (UTC)Reply[reply]
qqx is a special code which can be deployed for internal purposes. MediaWiki uses it to request debugging information. Certes (talk) 23:08, 8 January 2024 (UTC)Reply[reply]
I made the redirect WP:QQX for users who already know there is something called qqx and search for it. Somebody else added {{Shortcut|WP:QQX}} at the target section. PrimeHunter (talk) 23:13, 8 January 2024 (UTC)Reply[reply]
@Wbm1058: qqx isn't an acronym, it doesn't stand for anything. See ISO 639-2#Reserved for local use, which has: The interval from qaa to qtz is "reserved for local use" and is not used in ISO 639-2 nor in ISO 639-3. Therefore, we can use qqx as an unofficial pseudo-language code, safe in the knowledge that the ISO will never allocate it to a real-world language. --Redrose64 🌹 (talk) 00:17, 9 January 2024 (UTC)Reply[reply]

Tech News: 2024-02[edit]

MediaWiki message delivery 01:17, 9 January 2024 (UTC)Reply[reply]