This page contains discussions that have been archived from Village pump (technical).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.
Hi. I didn't find anything in the hundreds of archives here and I feel a bit displaced. So I just ask:
Most times I edit on the mobile app on a tablet. I always search on the browser, also in wikipedia, eg differently named lemmas and such. When I click on a wi:link I get the pop-up, don't you want to open it in the app, and everytime I have to say no. If a link opens in the app, while I am editing in it, the edit is lost. (I just lost about six hours of work on a lemma.) I know, stupid enough, I could prevent at least some of it, if I would edit incrementally and save my edits once in a while.
But, isn't there a way to (optionally) suppress the pop-up in the browser altogether or have another way to NOT loose my edits? The app doesn't allow other tabs, while editing, so the app seems not to be designed for editing anyway. Would you advice not to edit in the app as a principle rather than to think about any kind of solution to fix the unsecure state of an open editing page?
Another question: is there any way I can restore my (last) lost edit via a log or protocoll in the tabletvor the app?
Thanks in advance, MenkinAlRire23:30, 10 November 2023 (UTC)
@MenkinAlRire we can see if there is a known bug open. Please let us know which version of the Wikipedia App you have installed. To verify, this only breaks if you navigate away from the edit in the app (via the browser link) while you are editing correct? And a fix would be if you could unregister the link handler from your browser? — xaosfluxTalk10:33, 11 November 2023 (UTC)
@User:Xaosflux Hi. Yes, I am not totally sure, but the pop-up comes up specifically, if I click a link of a search result (different from the edited page; although then it might break with the edited page as well, I don't know). If you are not fully concentrated, you are surprised, how it happened; maybe the default is 'resolve link in app', so that pushing enter on the key pad would suffice. You need to be able to search things while editing (verify links...), and I really try to pay attention, but it happens. So, as you said, to be able to unregister the link handler would do the trick. My version is 2-7-50455-r-2023-10-10, should be the latest, and I use firefox (def. newest v). Nice to see you interested, thanks, MenkinAlRire13:14, 11 November 2023 (UTC)
You found answers outside wiki. Obviously you have to understand the problem to ask the right questions. I knew that I have the option to suppress/asked/allow to handle links, downloads etc. But it didn't came to my mind in this case, since it is a universal rule for links, dwnld... so usually I wanna be asked - as a result I have the problem with wiki. Let's see if it'll bother me, when ff does everything on its own; on the PC its the default (unfortunately the tablet has no right-click menu). The other link proposes tools to manage behaviour. My experience with xy-manager is that they don't provide anything, that is not already there (on windows at least). So I doubt, that they could allow me in firefox to just suppress this one pop-up on this one site.
Hey, anyway, thank you very much for this simple answer, I hope, it will suit my behaviour - and also, that I wasn't at the wrong place and bothered you with it. Problem (dis)solved. MenkinAlRire15:10, 12 November 2023 (UTC)
@MenkinAlRire was trying to find you a "quick fix" while a longer solution could be looked at; as far as a feature request could go would your user story be something like: "If another program opens a link with the Wikipedia App, it should not cause the app to abandon any edits in progress in the app"? — xaosfluxTalk15:46, 12 November 2023 (UTC)
Yeah, exactly, simple as this (sounds like). How and where would I make such a request? Or are you already about to do it for me? MenkinAlRire16:02, 12 November 2023 (UTC)
@MenkinAlRire: I've opened a feature request to require a confirmation before abandoning the edit in this manner that I think would solve your concern. It is tracked as phab:T351037, you may log on there using your Wikipedia account and subscribe for updates. Please keep in mind that not all feature requests will be selected and there is no SLO for implementation. — xaosfluxTalk16:51, 12 November 2023 (UTC)
“just lost about six hours of work on a lemma” The simplest lesson of course, is to save more often, regardless of any changes made to the software. —TheDJ (talk • contribs) 19:42, 12 November 2023 (UTC)
If I may quote myself: "I know, stupid enough, I could prevent at least some of it, if I would edit incrementally and save my edits once in a while." So, yes, I already try to get used to it. (I probably overcompensate my discontent with edit sessions of others, who save after every point and leave a half page of entries in the history. I worked offline, too, which bears other risks.) MenkinAlRire22:17, 12 November 2023 (UTC)
On Google Chrome with this page, the combination of U+033A◌̺COMBINING INVERTED BRIDGE BELOW with certain characters, such as U+0282ʂLATIN SMALL LETTER S WITH HOOK, causes the latter to become a box: ʂ̺. Тɦ!§ ο∏|ץ ♯ⴷp̪Ᵽ3クΣ f̃ʘᒋ 𖼺𖽙𑣣ع 🐍♈️▛▛▌ß◑λⓈ. Т̺ɦ̺!̺§̺ ̺ο̺∏̺|̺ץ̺ ̺♯̺ⴷ̺p̺̪̺Ᵽ̺3̺ク̺Σ̺ ̺f̺̺̃ʘ̺ᒋ̺ ̺𖼺̺𖽙̺𑣣̺ع̺ ̺🐍̺♈̺️̺▛̺▛̺▌̺ß̺◑̺λ̺Ⓢ̺.̺ –LaundryPizza03 (dc̄) 23:53, 12 November 2023 (UTC)
LaundryPizza03, could you add a translation for the last two sentences of your original post? I think I understood them, but writing something that isn't English, and which neither Google Translate nor a LLM can translate, isn't a good way to communicate with your fellow editors. (Perplexity.ai: 'I'm sorry, but the text you provided cannot be translated into plain English as it contains a mix of different characters and symbols that do not form a coherent sentence or phrase. It is possible that the text is in a different language or encoding that cannot be recognized by the system. If you have a specific text or phrase that you would like to translate, please provide it in a format that can be recognized by the system.') BlackcurrantTea (talk) 07:58, 13 November 2023 (UTC)
"This only happens for some symbols." (I didn't even realize it was meant to resemble a sentence until BlackcurrantTea pointed it out.) Nardog (talk) 09:03, 13 November 2023 (UTC)
The inverted-bridged s-hook works for me both in Chrome on Windows desktop and in my usual extreme-minority browser, both here and in Voiceless retroflex fricative#Occurrence. Both give me tofu for 𖼺𖽙𑣣, whether or not they're bridged; Chrome also tofus ⴷ and ᒋ when bridged. —Cryptic10:34, 13 November 2023 (UTC)
Would appreciate it muchly if some of the experts around here would take a look at User talk:Shearonink and respond here telling me whyyyyy. When you figure out what is wrong post here and just let me fix it myself please, that way maybe I'll learn.
Here's the code:
{{User:MiszaBot/config
|archiveheader = {{talkarchivenav}}
|maxarchivesize = 100K
|counter = 11
|minthreadsleft = 1
|minthreadstoarchive = 1
|algo = old(30d)
|archive = User talk:Shearonink/Archive %(counter)d
}}
Thanks, Shearonink (talk) 15:01, 9 November 2023 (UTC)
You can always copy-paste to commons, in the summary link to the original, then drop the import request; it should work while you are waiting. — xaosfluxTalk18:42, 13 November 2023 (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.
Last week, users who previously visited Meta-Wiki or Wikimedia Commons and then became logged out on those wikis could not log in again. The problem is now resolved. [5]
Last week, some pop-up dialogs and menus were shown with the wrong font size. The problem is now resolved. [6]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 14 November. It will be on non-Wikipedia wikis and some Wikipedias from 15 November. It will be on all wikis from 16 November (calendar).
Future changes
Reference Previews are coming to many wikis as a default feature. They are popups for references, similar to the PagePreviews feature. You can opt out of seeing them. If you are using the gadgets Reference Tooltips or Navigation Popups, you won’t see Reference Previews. Deployment is planned for November 22, 2023.
Canary (also known as heartbeat) events will be produced into Wikimedia event streams from December 11. Streams users are advised to filter out these events, by discarding all events where meta.domain == "canary". Updates to Pywikibot oder wikimedia-streams will discard these events by default. [7]
Why can't I stop IPs using the ping feature to abuse me?
IPs can ping me, and I don't see any way I can disable that in my preferences. It should at least be disable-able, as it enables abuse. I have a recent example of how that can work, which I don't want to describe here per WP:BEANS, but I can e-mail it if you ask. Bishonen | tålk13:28, 12 November 2023 (UTC).
Thank you, PrimeHunter. That's disappointing. I'm not much good at talking to Phabricator, and have trouble understanding most of the comments there. But I guess the upshot is the LTA in question can simply go on sending me disgusting messages with pings from random IPv6's? Because if I was able to mute them, that would be a "privacy violation"...? :-( Bishonen | tålk14:26, 12 November 2023 (UTC).
The privacy violation mentioned at phab:T207659 is only if you also got the ability to mute logged in users based on their current IP address which would otherwise be unknown to you. PrimeHunter (talk) 14:37, 12 November 2023 (UTC)
Yes, that's what it looked like. It may have been unfortunate that that obviously much more controversial request got mixed up with the simple (and surely uncontroversial) one. So why wouldn't they make it possible simply to mute all IPs? Just, there was no interest? I suppose you wouldn't like to re-awaken the 2018 request and emphasize that it's the simple muting of IPs that's wanted, PrimeHunter? Or start a new one, if that's the appropriate way of doing it. Bishonen | tålk17:40, 12 November 2023 (UTC).
In 2017, I corresponded with Trevor Bolliger who was Product Manager, Anti-Harassment Tools (User:TBolliger (WMF)). He was very responsive and I was going to suggest emailing Trust and Safety, if they are still active, but I see that TBolliger authored phab:T207659 which quickly faded to nothing on the premise that muting an IP range would also mute any users in that range which might allow determining information about a user's IP. That is clearly not what is needed. If someone is being harassed by a troll using IPs, the target needs a way to mute all IPs or a range of IPs on the understanding that the mute would not apply to registered users. My guess is that the WMF would be unable to identify an obviously required measure and would just suggest adding to a page of feature requests that will be ignored. I wouldn't think that posting to the moribund phab:T207659 would achieve much particularly with the upcoming hiding of IPs. Johnuniq (talk) 01:50, 13 November 2023 (UTC)
There is related discussion about the upcoming IP masking at phab:T331058 and phab:T344647. Somebody apparently thought you could mute IP's now and requested it should remain possible after IP masking. It was noted that you cannot do it now and the request was therefore declined as based on an incorrect assumption. That doesn't sound satisfactory to me. Even if you cannot do it now, it could be added as a new feature. It looks like phab:T344647 should be reopened or a new request created if you want it after IP masking. It may not be worth the effort to try to get it added to IP's now when the whole IP system is going to change. PrimeHunter (talk) 02:06, 13 November 2023 (UTC)
Emailing Trust & Safety would make it clear to them that there's a need for tools to deal with this. Even if (as Johnuniq suggested, realistically) it's one more item on a long list, it will make the point that this isn't a 'hypothetical use case': it's real. BlackcurrantTea (talk) 07:22, 13 November 2023 (UTC)
I have raised a patch for phab:T207659 yesterday, but no guarantees on when it would be reviewed or merged as the Echo extension does not receive much of maintenance nowadays. – SD0001 (talk) 18:59, 14 November 2023 (UTC)
Coming soon: Reference Previews
A new feature is coming to your wiki soon: Reference Previews are popups for references. Such popups have existed on wikis as local gadgets for many years. Now there is a central solution, available on all wikis, and consistent with the PagePreviews feature.
Reference Previews will be visible to everyone, including readers. If you don’t want to see them, you can opt out. If you are using the gadgets Reference Tooltips or Navigation Popups, you won’t see Reference Previews unless you disable the gadget.
Reference Previews have been a beta feature on many wikis since 2019, and a default feature on some since 2021. Deployment is planned for November 22.
Searching the archive, the last time I could find any update was about 6 months ago. Do we have any action we should take to fix pages that use the disabled Graph template? Or are we still in a holding pattern waiting for a fix? Thank you 87Fan (talk) 16:29, 15 November 2023 (UTC)
What is needed here is assistance in implementing this template. You can see Template talk:Refideas editnotice for full discussion, and the original discussion happened here.
I did a limited test on a small selection of articles for 30 days, but they were not high traffic articles so while most of the users who edited the articles advised me they could see the template, none of them made actual use of the sources on the talk page. If you believe that more testing is required before implementation, then we can try again with a selection from this list of talk pages with Refideas and work with some high-traffic articles such as Jazz and Puerto Rico and Star Wars (film) to see if that will lead people to use any of the sources they find on the talk page.
What is needed specifically is to make sure that the edit notice can be made to:
detect if there is a Refideas template at the associated talk page
be blank/invisible if no such template is present
if there is such a template: display the editnotice and include a link to the article's talk page
The goal is that we want everyone who clicks "edit" on any article with the Refideas template on its talk page to be able to see this editnotice. BOZ (talk) 21:50, 8 November 2023 (UTC)
BOZ, based on Template talk:Refideas editnotice, this has gone way beyond my technical competence. It seems like there are a fair number of articles that would need this editnotice, and it would have to be configured to appear on articles whenever someone transcludes {{refideas}} onto its talk page. Both of those sound like tasks for a bot, and since no one has responded to your request here, maybe your next port of call might be WP:BOTREQ? Sorry I'm not able to be of more assistance. Folly Mox (talk) 05:26, 14 November 2023 (UTC)
So here is a question. When I click "edit" on a BLP article, I get a red box that says "Notice about sources This article must adhere to the biographies of living persons policy..." which I assume appears because the talk page has the WikiProject Biography tag with the "living=yes" parameter. If that is correct, then would it be a matter of replicating that for articles with the Refideas template? Just hoping to not reinvent the wheel. BOZ (talk) 06:11, 14 November 2023 (UTC)
I'm not particularly technically oriented, but from a glance that looks like it could potentially be similar to my request? BOZ (talk) 14:09, 14 November 2023 (UTC)
@Redrose64, I thought of this; what if Refideas generated a category? Would we then be able to implement the editnotice on any article that had that category on its talk page? BOZ (talk) 14:43, 15 November 2023 (UTC)
Yes, because Common.js looks at the categories that the page is actually in; it doesn't examine the wikimarkup for the page source. --Redrose64 🌹 (talk) 16:35, 15 November 2023 (UTC)
@Izno oh OK, is there a better method you can think of to achieve the same goal? I'm not technically oriented so I am not aware of what the possible issues might be. BOZ (talk) 23:10, 15 November 2023 (UTC)
The fundamental issue is not one of better or worse, it's that everything has a cost. Adding such a module to Common.js (or any reader-facing JavaScript solution) carries a cost. In this case, the cost is much more than any benefit than I can see.
Someone could probably trivially put together a script for you to import if you wish to know if a specific page has refideas.... Izno (talk) 23:15, 15 November 2023 (UTC)
Currently, Wikipedia articles are designed to be read mainly on screens. There's a mobile view and a short description but their format doesn't seem well-suited to a coming class of devices. Smart speakers are already familiar and now there's the ai pin which is a wearable which looks promising. The technical issues which will arise with these may be similar to those arising from the use of screen readers by the blind or visually disabled. And there are users like car drivers and runners who want to keep looking where they are going. Or the elderly who are not comfortable with sophisticated smartphones. So what's the road map or strategy for this class of devices and user? And is someone working on a voice interface for editing so that there can be a two-way traffic? Andrew🐉(talk) 23:46, 15 November 2023 (UTC)
The new Vector theme appears to have changed the wording of the "New section" tab to "Add topic". This is causing some confusion, as seen here but also in general as help pages such as WP:TPNEW still use "New section". I hate trying to remember how to dig for and find the specific MW config pages that deal with this, but would it be possible to change the wording of the new theme back to "New section"? (please do notping on reply) Primefac (talk) 11:31, 15 November 2023 (UTC)
With the shortcut being QQX, you can see why I have trouble remembering it... I honestly don't care which we use I mainly prefer "New section" just because it's what I've always seen, but I don't think we should be splitting uses across flavours (and I won't lose sleep if I'm in the minority). Primefac (talk) 12:59, 15 November 2023 (UTC)Okay, so I do have a small preference
Yes. It was actually me who made the redirect WP:QQX. The idea was to help users who already know there is something called qqx and try to search it. I didn't think it was worth a {{shortcut}} at the target section. PrimeHunter (talk) 18:29, 15 November 2023 (UTC)
I bring up an issue, and the only responses are about an irrelevant point I made about forgetting a shortcut.... goodness. Primefac (talk) 20:35, 15 November 2023 (UTC)
Which is the MediaWiki default in MediaWiki:Tooltip-ca-addsection. Maybe the default tooltip deliberately uses different wording from the default label "Add topic". A tooltip isn't very useful if it just repeats the text you are already hovering over. PrimeHunter (talk) 23:08, 15 November 2023 (UTC)
I prefer add topic as it describes what it does to someone who has never used wiki software before, much less talk pages. New section doesn't. Izno (talk) 22:37, 15 November 2023 (UTC)
Also, users of other MediaWiki wikis may be familiar with "Add topic" which has been the default since Vector became the default skin in 2010. And Special:Preferences#mw-prefsection-editing says "topic" seven times about recent features. I think a consensus discussion would be required before changing to "New section" in Vector 2022. PrimeHunter (talk) 23:08, 15 November 2023 (UTC)
"Add topic" seems rather odd to me. If I am on the talk page for Band X, the "topic" is Band X. Adding a topic implies that I would now be talking about Band Y. Or dinosaurs. Or the Pythagorean theorem. "New section" indicates quite clearly that I am adding a new section to the same page and will still be talking about Band X. --User:Khajidha (talk) (contributions) 06:18, 16 November 2023 (UTC)
I would also support keeping "Add topic", and updating the wording in documentation pages to match that. It seems that the customization to "New section" in 2009 was mostly for consistency with MonoBook [8], which is no longer the default for like 15 years now (also also, it seems to have changed since – the current label there is just "+"). The customization also means that all normal MediaWiki documentation doesn't match Wikipedia, and has to be copied and customized here – for example, someone had to rewrite [9] as [10]. Matma Rextalk12:34, 16 November 2023 (UTC)
My script used to be able to load the 2010 Wikitext editor with CodeMirror. (when configured to do so) Now it only loads the 2010 Wikitext editor and shows buttons for bold, italic, links and images. But the next buttons should be "insert a template" (puzzle piece icon), "reference" (book icon) and (most important) syntax highlighting (marker icon), which activates CodeMirror. The "Cite" link at the end is also missing. I tried loading some older versions of my script, but it doesn't seem to make a difference. I'm unsure when it broke, but I suspect something changed in the Wikitext 2010 editor. Does anyone have a working example of how to load CodeMirror? Like, any other script that loads an instance of the 2010 Wikitext editor with CodeMirror? — Alexis Jazz (talk or ping me) 18:05, 14 November 2023 (UTC)
@Alexis Jazz How are you enabling CodeMirror? I was expecting to see a CodeMirror.fromTextArea() in your source, but I don't. That would be the normal way to (currently) enable CodeMirror on an arbitrary textarea.
I'll take this opportunity to inform you and others here that CodeMirror is currently being upgraded to the latest major version, and there will be no backwards compatibility. I aim to have this finished sometime early next year. You can follow phab:T259059 for progress. Fixing your script shouldn't be hard, and I'll find it via toolforge:global-search, so you can expect communication with a migration guide well in advance. I also plan to replace the Ace syntax highlighter (and probably the CodeEditor extension entirely) with CodeMirror, so in your script you won't have to check for Ace anymore. — MusikAnimaltalk20:53, 15 November 2023 (UTC)
MusikAnimal, that's quite a good question. I'm not sure if mw.user.options.set('usecodemirror',1); was required for it. I think I changed the ID of my textarea to wpTextbox1, mw.loader.using(['ext.CodeMirror'], changed the ID back and click()ed the button in the 2010 Wikitext editor to enable it as needed. I think CodeMirror kind of automatically attached itself to WikiEditor instances. I found the fromTextArea thing in phab:T214989#7287480 now too, I've been testing and it loads CM but .textSelection('getContents') isn't working on it. (just returns an empty string) — Alexis Jazz (talk or ping me) 00:05, 16 November 2023 (UTC)
@Alexis Jazz Yeah, the jQuery.textSelection() integration is inconveniently baked into the WikiEditor integration. For now, you can just copy the desired code, or even interact with CodeMirror directly. When code CodeMirror 6 is rolled out, adding a full-featured CM instance to any arbitrary textarea will be much easier. — MusikAnimaltalk02:12, 16 November 2023 (UTC)
MusikAnimal, there's codeMirror.Doc (capital D, a function) but codeMirror.doc doesn't seem to exist? (I'm probably just being an idiot for not seeing something obvious here) Sorry to keep bugging you, personally I wouldn't mind waiting for CM6, but I know some of my users are kind of addicted to CodeMirror. Edit: I am in fact an idiot, but I still don't get it and the longer I look at it the more confused I get. — Alexis Jazz (talk or ping me) 02:47, 16 November 2023 (UTC)
And you wouldn't want it to in either case - it would allow anybody who can edit the redirect change which (other) page is protected or blocked. Same reason why cascading semiprotection isn't workable.Something that might serve the same purpose would be to warn if you're partially blocking from a redirect or non-existing page. —Cryptic11:50, 16 November 2023 (UTC)
A useful feature might be to check that the account has edited the page. I can see circumstances when you might want to block someone who hasn't – perhaps they've been hammering an edit filter without finding the magic formula to get their dross through – but it must be rare. Certes (talk) 19:04, 16 November 2023 (UTC)
GeoHack error
If anyone, like me, is perplexed why the National Map viewer in GeoHack always takes one to Newton County, Missouri:
It is because there's an error in the URL construction where , is used instead of &.
T346920 is now marked as resolved. VisualEditor's Add a link should suggest a redirect with exact case match, thus no longer encouraging editors to add sub-optimal links through redirects from incorrect capitalisation, etc. Certes (talk) 11:44, 17 November 2023 (UTC)
Wierd contribs list.
My contributions list looks like this all of a sudden, over the last couple of days. It's gone a bit funny: whatever page I click on '''bolds'' the text, gives it a gives it a yellow background but worst of all, doesn't open. The only way of opening a page is right-click/open. Which is mildly inconvenient. I don't think I've changed anything in my settings for ages, so I'm having difficulty tracking the cause down. Any suggestions? Cheers folks! ——Serial18:02, 16 November 2023 (UTC)
Also, it acts as if it's an editable page: when I click away from it, the box pops up saying 'This page is asking you to confirm that you want to leave — information you’ve entered may not be saved.' ——Serial18:41, 16 November 2023 (UTC)
I don't see CSS in those scripts to cause this... coloring. All that does is make the diffs open/close and provide the appropriate buttons (User:Bradv/Scripts/ExpandDiffs.js is an alternative on this point). I think figuring out where the coloring is coming from is probably the first thing to do. Izno (talk) 19:00, 16 November 2023 (UTC)
Agreed. I was looking at a talk page thread that mentioned "inspect diff" but not the colors. It may be User:Ais523/topcontrib.js. When I load that in my vector.js, I see all sorts of colored backgrounds in my Contributions page. – Jonesey95 (talk) 19:16, 16 November 2023 (UTC)
Yes, this script would cause display like this. I don't personally know why things would have changed like they have. SN, try removing the relevant line and see if it restores the expansion. Izno (talk) 21:38, 16 November 2023 (UTC)
@Serial Number 54129 The #1 thing to do when you encounter weirdness like this is to test that it is in fact a script that is at fault. You can do this by viewing the same page while logged out, or add safemode=1 to the URL (example). If the issue goes away, you know it's one of your scripts.
You'll likely have to remove scripts one by one to debug what is causing this. I can say with moderate confidence that what you're experiencing is probably not due to a recent change in MediaWiki. The yellow background is a rather glaring issue; if it were at all widespread, I feel like we would have heard about it by now. — MusikAnimaltalk21:41, 16 November 2023 (UTC)
Believe me, MusikAnimal, I wasn't blaming MW; I wouldn't know how to! But it would be cool if I could get a developer to do it, what with me being sausage fingers/muttton-brained in this kind of thing :) it would also be cool to have a script that identified bad scripts. But that rabbit hole might lead us to needing a script that identifies when the bad script identifier itself goes bad, ad infinitum :) ——Serial14:37, 17 November 2023 (UTC)
Mobile web DiscussionTools: Wikipedia:* vs. *talk:
I have recently noticed that there are differences in the UI of DiscussionTools on pages of the Wikipedia: namespace and the talk namespaces on the mobile web. I am not sure if they are intentional or not, or reproducible, so I am asking here. They are:
There is no subscribe button on WP: discussions.
The reply button is different – on WP: discussions, the reply button looks like the desktop one, while on the latter type of pages, the button shows in a mobile-specific design.
A related matter is that collapse/expand icons (caron/caret-like symbols) are centered next to the section title in mainspace sections or WP: discussions but are displayed next to the first line of title in *talk: discussions. This is related to T344629.
Janhrach (talk) 11:24, 17 November 2023 (UTC)
1 & 2 are both expected, as is the collapse icon position -- they're symptoms of the DiscussionTools visual enhancement features being restricted to Talk namespaces. We've got a pending deployment (T331635) for "turn on visual enhancements on any page that has __NEWSECTIONLINK__", but I think the current plan is for that to reach enwiki last because we want to make sure any potential problems are ironed out before it goes to the highest-traffic wiki. Though (speaking as someone who's not in charge of the deployment schedule) I'd imagine that we might be responsive to community requests for faster deployment if it's causing problems. (If you're curious, you can see what would happen when it's turned on for a given page by using the dtenable URL parameter, for example on this page.)
3 is a bug, though. Looks like a CSS specificity issue on our end -- we hide reply links inside .mw-archivedtalk, but one of the mobile CSS rules is overriding that. I've made T351542 for this and it'll get fixed. DLynch (WMF) (talk) 15:30, 17 November 2023 (UTC)
Okay, T351542 is fixed. It'll wend its way through the standard deployment process, and make its way to enwiki by Thursday-after-next. (US Thanksgiving is causing disruptions.) DLynch (WMF) (talk) 16:33, 17 November 2023 (UTC)
My guess is that removing the flags would fix the problem but doing that would be tricky because several different templates are used to include them. Johnuniq (talk) 06:48, 17 November 2023 (UTC)
Agreed. I tried {{Flagg}}, which claims to reduce post-expand include size, but it is slower and takes the page over the 10-second rendering limit. Certes (talk) 10:12, 17 November 2023 (UTC)
Yep, ditch the flags. Those templates are tranclusions within transclusions, and we discovered last decade those quickly engage the tranclusions limits. SandyGeorgia (Talk) 17:52, 17 November 2023 (UTC)
A template with arguments like {{flagu|United States}} makes a full recursive expansion at each of the 145 identical calls with that argument. Each template call in the process contributes to the limit. A template without arguments like {{USA}} only makes the full expansion once and then copies the result the next 144 times where they only count with the size of the final output. I have used this to make a fix [11] which removes the category for now without removing flags. The page went from well above the limit to 94% of the limit. That's still close and the problem may return later. PrimeHunter (talk) 01:49, 18 November 2023 (UTC)
Are my [expand / collapse][chevrons / carets] supposed to be misaligned with the section headings now on Minerva? (Android + Firefox, duplicated in Chrome). Just wondering if this is intentional. Folly Mox (talk) 13:24, 9 November 2023 (UTC)
I hesitantly suspect this might be related to phab:T344629 (just based on the timing and keywords), so I've added a note there with some additional details. Will await confirmation before adding a {{tracked}} template here though. Quiddity (WMF) (talk) 21:24, 9 November 2023 (UTC)
I was indeed related to T344629 -- we didn't think about how the change would interact with pages that do get DiscussionTools processing them but don't (yet?) get the visual enhancements applied. It's fixed in T351044, and the patch will be rolling out on the train this week. Expect it to look correct again on enwiki on Thursday. DLynch (WMF) (talk) 16:53, 13 November 2023 (UTC)
H:INCAT has an example of how to search incategory that does not work. See the search for "ammonia" in Category:German chemists and it ends up searching the main namespace, not the category namespace and returns 0 results. If someone can help me fix this, it will also solve a problem that I'm trying to solve over at species:. Merci. ―Justin (koavf)❤T☮C☺M☯09:41, 18 November 2023 (UTC)
It's not supposed to search in the category namespace, it returns no results simply because there are in fact no articles that contain the word "ammonia" in that category. Nardog (talk) 10:04, 18 November 2023 (UTC)
Okay, that is pretty confusing... Maybe a more intelligible example is better? To be fair, I'm also sleep deprived, with just three hours in the past 36, so maybe that's it. Thanks. ―Justin (koavf)❤T☮C☺M☯10:20, 18 November 2023 (UTC)
I guess I was just fundamentally confused and it seems like the example is not optimal by providing an example search that has no results... ―Justin (koavf)❤T☮C☺M☯21:53, 18 November 2023 (UTC)
Maybe it had results when the example was added, or maybe somebody thought it would be interesting to give an example which only produces results with deepcat in the following section. I think it's best to have an example which gives results in both cases but a lot more with deepcat. I have changed it to "Berlin" [12] which currently gives 20 results for Berlin incategory:German_chemists and 493 for Berlin deepcat:German_chemists. PrimeHunter (talk) 22:35, 18 November 2023 (UTC)
To butt in, mine also won't load and it's c. 25,000 items. I get: "[characters] 2023-11-18 21:54:38: Fatal exception of type "Wikimedia\RequestTimeout\RequestTimeoutException""―Justin (koavf)❤T☮C☺M☯21:55, 18 November 2023 (UTC)
No. You can probably use some combination of PagePile and Petscan to do what you might want (you should explain what that is). Izno (talk) 21:42, 18 November 2023 (UTC)
@Izno: See the example here: I thought Wikipedia would support haswbstatement as a search parameter to filter search results using their Wikidata properties. I would use these queries to find pages on the English Wikipedia that aren't linked to articles in other languages, and update the links accordingly. Jarble (talk) 22:40, 18 November 2023 (UTC)
It would be nice if, in the VE interface for editing templates, url values had a button that linked to the website, to avoid copy/pasting. This is especially useful when editing citations in the VE. For example, it currently (very, very roughly) looks like this:
Venus was recently edited to remove an image which has been deleted from Commons. The image was added some months ago to replace a similar image, which was presumably considered inferior but better than nothing. The article is now left with no image. Everyone clearly acted in good faith here, and each action considered individually was an improvement, but the net result makes the article worse. This isn't a one-off incident: I've seen similar sequences elsewhere. Do we have any process for detecting, and ideally rectifying, such problems? Certes (talk) 16:49, 19 November 2023 (UTC)
What I don't understand is why the template is trying to use 2022 when the parameters fed to it specify |year=2021. I tried to change 2021 to 2020 or 2019, but no matter what year I give it, it runs off to make the "2022 error". The template coding is very complicated and I don't care to spend another hour trying to solve it. wbm1058 (talk) 13:19, 19 November 2023 (UTC)
There are many interacting templates and I don't know the system but I previewed {{AUDConvert|{{From USD|1290000000|AUS|year=2021|round=yes}}|year=2021}} and hovered over the list of used templates with Navigation popups to find a recent change. {{Inflation/year}} was edited today by Pppery who set AU = 2022.[13] It worked before that. PrimeHunter (talk) 13:49, 19 November 2023 (UTC)
What's the right way to display a transparent PNG containing white text on a black (#000000) background? Is there some image code or infobox parameter that can be used (with {{Infobox company}}) or should a derivative image with a black background be made (and how so)?
The article is GirlsDoPorn (warning: graphic descriptions of rape) and myself and Vanjagenije are unsure. While we're here I wonder if the PNG should be converted to an SVG but don't know how to do this. — Bilorv (talk) 15:06, 19 November 2023 (UTC)
Dated maintenance categories "as of" the future or the past
There are recurring problems at Special:WantedCategories that happen entirely as a result of misuse of dated maintenance templates, which I wanted to ask if there's any way to prevent.
Somebody postdates {{as of}} to a future date whose "current" conditions cannot possibly be known yet (this time out I came across examples for both January and October of 2024), thus autogenerating the dated monthly category for "Articles containing potentially dated statements from Month YYYY". AnomieBOT then comes along and automatically creates that category since it's a non-empty dated monthly maintenance category, but then leaves the container "Clean-up categories from Month YYYY" as a redlink that whoever's dealing with redlinked category cleanup has to manually clean up by correcting the misdated template and then deleting the premature dated-statements category in order to resolve the clean-up redlink. Category:Articles containing potentially dated statements from January 2024, for example, has already had to be deleted five times as of today (three of them in the past week alone), with its first creation taking place in May 2023. But, of course, we cannot already know as of today what will become true months or years into the future, so there's no call for postdating an "as of" template that way and no reason why any categories beyond the current month would need to already exist now. Is there any way that this template can be made to do an ifexist check, and use one "as-of templates with dating problems" catchall category if the template is dated to a future month whose category doesn't and rightly can't exist yet, instead of autogenerating a redlink that cascades into a two-step problem?
Somebody who doesn't understand how {{citation needed}} works backdates the dating on that template to either the month and year when the uncited statement was originally added to the article in the first place, or the month and year in which the event that the statement is describing took place, thus repeating the same "AnomieBOT creates the dated monthly and then humans have to deal with a redlink for the clean-up container parent" cycle. But, of course, what we're tracking here is when was the template added to the article, not how long ago did the things the article is about originally happen or how long the article has existed, so there's never any reason to add a citation needed template dated any earlier than the current month. So, again, is there any way that this template could be coded to do an ifexist check, and use a "citation needed templates with dating problems" catchall instead of autogenerating a dated monthly that doesn't already exist?
The dated subcats of Category:Clean-up categories don't get created when non-empty because the cats with redlinks to those don't get added to Category:Articles with invalid date parameter in template. I can have AnomieBOT specially look at every category listed starting with "Clean-up categories from" to look for missing "month year" ones. Hopefully that's the only one, having to list all the cats and pick out the very occasional missing ones is a bit wasteful of resources.AnomieBOT mostly won't create categories for dates more than 1 day in the future even if non-empty. There are currently two exceptions: non-empty subcats of Category:Articles containing potentially dated statements will be created up to 366 days in advance because there are potentially some cases where things are reasonably expected in some advance (and getting everyone to always use {{update after}} seems unlikely), and non-empty subcats of Category:Articles that include images for deletion will be created up to 8 days in advance due to how that one is added by deletion templates.Similarly, AnomieBOT re-creates non-empty cats for past dates in case of reverts that restore a past-dated template. It won't go earlier than 2004 though (Category:Wikipedia articles in need of updating is an exception), since no revert could produce an earlier date than that. But it can't easily tell the difference between a revert and a human doing the wrong thing. Anomie⚔23:48, 19 November 2023 (UTC)
Visibility of changes on mobile
When editing on my (Android) phone in mobile mode (not Desktop), there is not a Changes page or display, only a Preview of the rendered edit. Does functionality exist to show Changes in mobile mode before publishing, or can that function be added? Haven't found discussion of this issue. DonFB (talk) 00:04, 20 November 2023 (UTC)
Bot authorization question
I just fixed a problem with ChristieBot but I don't understand what happened and would be glad of an explanation. I normally use Chrome but yesterday had a reason to log in to Wikipedia on a separate browser, so I started Edge and discovered that I had logged in as ChristieBot there. I logged out, logged in as myself, and then found a few hours later that ChristieBot was failing. I logged back in as ChristieBot on Edge and when that didn't fix it I ran the bot interactively on the toolserver. It prompted me for ChristieBot's password, and once I put that in everything started working again. What happened? Why would logging out of a web browser cause the toolserver to require a fresh login? Mike Christie (talk - contribs - library) 11:48, 20 November 2023 (UTC)
When you logout, you invalidate your login session, which logs out your account on all your devices. What kind of authentication are you using for the bot account ? user/pw, botpasswords or oauth ? —TheDJ (talk • contribs) 11:59, 20 November 2023 (UTC)
Thanks for the explanation; that makes sense. I'm not at that computer and can't easily check (it was a year ago and I don't remember) but I think it was botpasswords; I think I tried OAuth and couldn't get it to work. Mike Christie (talk - contribs - library) — Preceding undated comment added 12:20, 20 November 2023 (UTC)
Book Citation - Ref Tool Bar
Not new, but am wondering if anyone else experiences this. Specific to the Book Citation - Ref Tool Bar. Used to be that if I entered a ISBN and clicked on the Autofill button, it populated the template. Hasn't worked for a few months now. And it's only that one item on the tool bar. The rest work as they always did. Has anyone else experienced this glitch? — Maile (talk) — Preceding undated comment added 04:49, 20 November 2023 (UTC)
@Maile66: Always give an example when you report an issue. It works for me on tested ISBN's with Firefox and Vector legacy or Vector 2022. I clicked "Cite" above the edit area in the source editor, selected "cite book" and entered 0-596-51516-2 in the ISBN field. PrimeHunter (talk) 12:51, 20 November 2023 (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
There is no new MediaWiki version this week. [15][16]
Starting on Wednesday, a new set of Wikipedias will get "Add a link" (Quechua Wikipedia, Romansh Wikipedia, Romani Wikipedia, Rundi Wikipedia, Aromanian Wikipedia, Tarandíne Wikipedia, Rusyn Wikipedia, Kinyarwanda Wikipedia, Sanskrit Wikipedia, Sakha Wikipedia, Santali Wikipedia, Sardinian Wikipedia, Sicilian Wikipedia, Scots Wikipedia, Sindhi Wikipedia, Northern Sami Wikipedia, Sango Wikipedia, Serbo-Croatian Wikipedia, Sinhala Wikipedia, Slovak Wikipedia, Slovenian Wikipedia, Samoan Wikipedia, Somali Wikipedia, Albanian Wikipedia, Serbian Wikipedia, Sranan Tongo Wikipedia, Swati Wikipedia, Southern Sotho Wikipedia, Saterland Frisian Wikipedia, Sundanese Wikipedia, Silesian Wikipedia, Tamil Wikipedia, Tulu Wikipedia, Telugu Wikipedia, Tetum Wikipedia, Tajik Wikipedia, Thai Wikipedia, Turkmen Wikipedia, Tagalog Wikipedia, Tswana Wikipedia, Tongan Wikipedia, Tok Pisin Wikipedia, Turkish Wikipedia, Tsonga Wikipedia, Tatar Wikipedia, Twi Wikipedia, Tahitian Wikipedia, Tuvinian Wikipedia, Udmurt Wikipedia, Uyghur Wikipedia, Uzbek Wikipedia, Venda Wikipedia, Venetian Wikipedia, Veps Wikipedia, West Flemish Wikipedia, Volapük Wikipedia). This is part of the progressive deployment of this tool to more Wikipedias. The communities can configure how this feature works locally. [17][18][19]
The Vector 2022 skin will have some minor visual changes to drop-down menus, column widths, and more. These changes were added to four Wikipedias last week. If no issues are found, these changes will proceed to all wikis this week. These changes will make it possible to add new menus for readability and dark mode. Learn more. [20]
Thanks for the explanation. I don't know why userspace is treated differently to other non-content pages. — Martin (MSGJ · talk) 07:28, 21 November 2023 (UTC)
Because user space can also have user scripts (which in turn can use json). But if suddenly .json pages started showing up in article namespace, the community would flip out. —TheDJ (talk • contribs) 09:56, 21 November 2023 (UTC)
Images purge issue
Hi all. Updated a couple Commons images some days back:
The new version is not using xlink, which i assume is the problem
These are not actually SVGs. They are PNGs wrapped in an SVG. If you don't have the original, then you can't modify them properly. This is not how SVGs are supposed to used.
I'm attempting to correct an issue with the map on this page. As the article lead suggests, the map shown in the infobox - File:Inner Loop Rochester map.svg - is no longer accurate and the map that should be appearing instead is File:Inner Loop Rochester 2014 map.svg. However, simply replacing the file name is causing the map to not render correctly. Can someone with better skills than me possibly get the file replaced? Home Lander (talk) 20:43, 19 November 2023 (UTC)
The infobox is actually currently showing me neither - it's a wildly-inaccurate openstreetmap render which shows the east section intact and none of the southwest third (concurrent with 490) in red. —Cryptic23:50, 19 November 2023 (UTC)
Hello, all, I'm looking for some opinions. Wikipedia:Village pump (policy) tends to have size issues. Yesterday, it was at 970K – nearly a million bytes long, or approximately the length of any single volume in The Lord of the Rings. For comparison, this page is presently 70K, VPPR is 263K, VPIL is 100K, and VPM is 20K.
WP:SIZESPLIT suggests 30K to 50K, as encyclopedia articles aren't supposed to be book-length. IMO village pumps don't need the same restriction, because they aren't normally read straight through, but there are other considerations, namely that we still need people to be able to read them, including reading/scrolling through them on a smartphone and to reply to the comments there.
My question for you all is: How much text should be considered unwieldy or inaccessible (in a technological or practical sense) for people who are using a smartphone? WhatamIdoing (talk) 20:35, 20 November 2023 (UTC)
I know that VPP now loads a lot better, it had started to breakdown. 970K is way over whatever the limit should be. VPPR is still loading without issue though. -- LCU ActivelyDisinterested«@» °∆t° 20:19, 21 November 2023 (UTC)
I can't give exact figures without a lot of testing, but I'd guess the figure is higher. Maybe encouragement at 250k, with strong advice to do so at 500k. It would also depend on how much more discussion is likely, if the discussion has died down splitting it may not be worthwhile. -- LCU ActivelyDisinterested«@» °∆t° 21:13, 21 November 2023 (UTC)
100-150K seems like a reasonable upper bound in the general as we've generally held to even for our longest pages (e.g. FAs), both in the discussion context and outside it. Tenders of VPPOL itself should do more to encourage splitting threads to subpages of VPPOL or RFC or a topic-specific proposal page, as much of the size is caused by novels of discussions that would be good to preserve in a specific location anyway. If a 1-line advertisement of that discussion needs to be pinned on VPPOL until the conclusion, that seems like a reasonable tradeoff for purpose.
(And it's usually pretty easy to know which discussions will be large and which won't.)
Much of what makes these pages difficult to load is the fact we use the discussion system we do, rather than one suited to a "forum" (for lack of a better word). Opening a new list and list item and then closing it every time someone makes a comment isn't efficient.
The only particular pain I have with such pages on mobile is when discussions become too indented. But I have 5G networks. You should consider how big a page should be when dealing with 3G networks, which I think is what WMF still tests with. You can simulate network speed in browser console. Izno (talk) 21:29, 21 November 2023 (UTC)
Readability improvements: prototype testing results and next steps
Hi everyone,
In the past few months, the Web team from the Wikimedia Foundation has started working on the project Accessibility for reading. Our goal is to improve the typography on the desktop and mobile sites, and build an easy way for logged-in and logged-out users to set their preferred typography.
Cohorts of designs sent by editors
In October and November, we shared a prototype with 13 Wikipedia language communities and received feedback from more than 600 editors and other logged-in users. The prototype allowed them to experiment with different font sizes, line height, and paragraph spacing, and report which configuration was most comfortable for them.
The majority preferred a slightly larger font size and line height than the current default (around 15 – 17px). A large group showed a strong preference for the current default (14 px), while a smaller group preferred significantly larger font sizes (20–24px). See the full results of the testing.
First version of the new Theme menu when the Tools menu is hidden
We learned that editors who decided to take part in the test are interested in changes. We hope that small, medium, and large options to choose between will be received positively.
In the first half of December, on all wikis, we will launch a new beta feature which allows to select the preferred typography. The feature will be opt-in and available in your user preferences. Turning it on will display a new collapsible menu in the right sidebar, which will appear under the Tools menu.
We invite you to look out for it, try it out when it becomes available, and tell us what you think. In the meantime, if you have time - read over the report linked above and take a look at the mock.
(If you're curious about tiny details, there's a small error on the mock, the Tools menu should be next to the "View history" link and the watch star. Also, we may choose the wording "Night mode" instead of "Dark mode". The "hide" button doesn't have brackets and the word "Theme" is bolded out because this mock is built upon visual changes which we're rolling out next week, and which we mentioned in Tech News: 2023-47 - learn more about that here).
Greetings, Hoping the following is not "off-topic", as I thought to share my Monobook skin Font and Color changes for readability.
Recently, I downloaded and installed Font Inter from Github here. Inter a trademark of Rasmus Andersson
(DBA: RSMS) This font IMO is very readable and "easy-on-the-eyes"; and contains many foreign language glyphs.
Within my fav Skin (monobook.css), I made these changes:
Content (font-family) = "Inter Medium" - very readable
Content color = 800000 (Maroon) - a dark font color choice
Content background = ffe4e1 (Misty Rose) - contrast color for maroon
Wikilink font = "Inter" - slightly lighter than Inter Medium
Wikilink color = 348781 (Aquamarine Stone) - tones down Blue default
Since computer displays show colors and contrasts differently, these settings can be changed for individual needs. I am certain these choices may not be for everyone but I am hoping this example is helpful, especially for editors with visual difficulty , and those with many hours of Wikipedia updates. Regards, JoeNMLC (talk) 16:00, 24 November 2023 (UTC)
Hello! Is there a way for mass-reverting moves? In my homewiki we're in a situation where we need to mass-revert some hundred pages moved recently (together with their talk pages). Any way to achieve that without needing to manually move each one? I thought I'd ask here for help, maybe you can point me into a tool or a native way to achieve that. - Klein Muçi (talk) 11:57, 24 November 2023 (UTC)
So, I've been using one-click archiving on my talk page for years. I realized it was a little cluttered today so I went to archive some stuff, only to find the links on the section headers are gone. Checked a few other pages that I know use one-click and I'm not seeing them there either. Beeblebrox (talk) 23:56, 22 November 2023 (UTC)
Oddly, it tried to stop from saving the code, saying it contains errors, but I just confirmed it is working just fine. Thanks! Beeblebrox (talk) 00:22, 23 November 2023 (UTC)
This question is about the ins and /ins tags, and presumably the del and /del tags. If I omit the / in the closing tag, and so have an extra ins, and then have another insertion, it appears that its end tag does not turn off the underscoring. My inference is that the extra insert tag not only fails to turn off the insertion (of course), but specifies two levels of insertion, so that multiple turn-off tags are also required.
You can use nowiki tags to disable tags. For a shorthand you can also use < as the opening bracket and that will display the tag. Or you can use {{tag}}.
Your understanding is correct. Open a tag and forget to close it, you now simply have two open tags which both need closing. Izno (talk) 23:16, 25 November 2023 (UTC)
I don't really understand why that number is there. There is no particular limit in either section or full page editing (well, there is for full pages, but it takes a lot of text to get there and the engine will tell you it's at the limit).
I would recommend changing it to something along the lines of "if the section also isn't large" or similar. Izno (talk) 23:19, 25 November 2023 (UTC)
This probably got broken in the figure element rewrite. You can also use the thumbtime parameter to get the thumb of an offset in the video. Ill file a ticket later tonight. —TheDJ (talk • contribs) 17:13, 24 November 2023 (UTC)
TheDJ, thanks, that's exactly what I was looking for! Somehow thumbtime wasn't documented on mw:Help:Images. I had already filed phab:T351937, I changed it from a feature request to a bug report. — Alexis Jazz (talk or ping me) 18:45, 24 November 2023 (UTC)
Pages using infobox boxer (amateur) with conflicting parameters
The latest run of Special:WantedCategories features a redlink for an infobox-generated (hence unremovable) maintenance category that I don't know what to do with, Category:Pages using infobox boxer (amateur) with conflicting parameters. I don't know if this is a category we want, or if it's one that should just be banished into the ether — and even if we want it I have absolutely no idea where it would belong in the category tree, while if we don't want it then I have no idea how to make it go away.
So could somebody with more knowledge about infobox maintenance categories either create the category if it's wanted, or kill it off so that it's never generated at all anymore if it isn't? Thanks. Bearcat (talk) 15:21, 25 November 2023 (UTC)
When there are pages listed in that cat, the thing to do is to visit the page (e.g. Luke Jackson (boxer)), and edit the lead section. Then, without altering anything, click Show preview. This will yield a message like this:
Preview warning: Using more than one of the following parameters in infobox boxer (amateur): medals, medaltemplates.
Return to the point where you were editing the lead section, locate the parameters concerned - in this case |medals=2006 Commonwealth Games and |medaltemplates= {{MedalCompetition|[[Boxing at the Commonwealth Games|Commonwealth Games]]}} {{MedalBronze| [[2006 Commonwealth Games|2006 Melbourne]] | [[Boxing at the 2006 Commonwealth Games|Featherweight]]}}. Decide whether they should be combined, or one of them simply removed. Amend and save, the page should drop out of the category. --Redrose64 🌹 (talk) 16:31, 26 November 2023 (UTC)
Definition of Local Project
Hi,
I read here that Wiktionary is not considered local to the Wikipedia project as far as Interwiki prefixes are concerned, yet a call to mw.site.interwikiMap("local") includes it. Should it be there?
I added a file to Fire alarm notification appliance with a recording of the standard pattern used for fire alarms in North America, and while the file shows fine when in landscape, when viewing in portrait (first on my phone but later by using the F12 developer tools to test the page), it turns out that the "height: auto" might be interfering with the proper display of the player. AwesomeAasim22:12, 26 November 2023 (UTC)
So while browing Wikipedia as a reader, I noted that Template:State historic sites of Louisiana map is being used at List of Louisiana state historic sites. If I'm reading the code right, this is created by taking an underlying map, and then overlaying the pins by hard-coding a pixel location in the template. This works at a standard presentation on a laptop, but at least on my mobile the way this code is written creates a warping of pin locations - everything is shifted to the west to the extent that Mansfield and Fort Jesup are showing up in Texas. Hardcoding pixel locations like this can't be a best practice. Is there any way to fix this within the current template, or does this just need replaced with something derived from Template:Location map? Hog FarmTalk00:43, 26 November 2023 (UTC)
I'd start by replacing it with one of the few standard templates, yes. If that doesn't fix it, that's a bigger problem. (Memory says those aren't perfect but also aren't as broken at low resolution.) Izno (talk) 01:47, 26 November 2023 (UTC)
With text-align:center the map was centered above the caption so if the caption has a longer line than the map width then the map is moved to the right. The pins were not moved so that's bad. In my desktop view the first caption line is slightly longer than the map so the pins also became slightly displaced for me in desktop. It was much more in mobile. PrimeHunter (talk) 00:22, 27 November 2023 (UTC)
Emboldening a boundary in a table
Hi,
Will someone who knows table formatting please have a look at the table in this page. I wonder about making the association of the four source language headings with system variants more evident. Ie. "Source Texts in Oberon" should associate with the variants preceding "Source Texts in Oberon-2" and "Source Texts in Oberon-2" should associate with variants between that and "Source Texts in Active Oberon". Might be clear enough already but I imagine emboldening the line above each "Source Texts" heading. Thanks, ... PeterEasthope (talk) 14:50, 26 November 2023 (UTC)
Please don't Apply background colour and bold to make it look like a heading, it will create an accessibility problem. If it's supposed to be a header for the row, mark it up as a header cell (by using ! at the start of the line instead of |) and give it the attribute scope=row. --Redrose64 🌹 (talk) 16:50, 26 November 2023 (UTC)
Broken into multiple tables again, to avoid violation of MOS:COLHEAD. With multiple tables, the emboldened line is not required. I find the uniform column widths aesthetically preferable to distinct widths in each table.
New Lua module replacement for {{Name in official languages}}
Hello, I've decided to make a Lua module to replace {{Name in official languages}}. I'd appreciate it if anyone could check whether the code has any issues.
At the moment I'm trying to figure out how to create wikilinks for the languages, which I'm having trouble with, due to Lua error at line 22: attempt to concatenate local 'display' (a nil value).
Not at all clear to me what it is that you are trying to do. It appears that you are attempting to use Module:Lang/data to get language names from language tags. Your life might be easier if you use the functions built into Module:Lang that already do that. For example, your test template has 'en = English' so I presume that you want to fetch the language associated with en from Module:lang/data. You can do that with a template call that looks like this:
{{lang|fn=name_from_tag|en}} → English
You can require() Module:lang into your module to do the same thing:
Or, because I suspect that the variety of languages needed is limited, you are probably better off using the MediaWiki magic work {{#language}} to get the name associated with a language tag:
Changing my mind. Better to require('Module:Lang') because you will have to anyway to properly render the official name associated with the language tag.
Also, no need to construct language name links (as you are attempting in line 22. The call to Module:lang to get the language name can be tweaked to get a linked version:
So, basically you can't concatenate nil, yet display is indeed nil because you're feeding in values that aren't in language_names (Module:Language/data/iana_languages). This is because you're still iterating over the frame arguments including width, font_size & name as the frame object exposes its parameters via a metatable-provided interface, so your attempt to set these to nil has no effect. Because you intend to support multiple languages, and so multiple parameters, you'll need to filter out these other parameters in your iteration. However, you can't just exclude these three (width, font_size & name) because users may stuff in other random ones too, so you'll need to check the item exists within your language_names first. Due to the structure of your code, checking in ::resolve_language is too late, because you'll have already set up table lines by that point, so you'll need to check before constructing the row, e.g. in p.main.
Note too that Module:Language/data/iana_languages has entries with multiple ISO 639-1 names, e.g. "cu", so you may want to consider handling these. You should also consider users entering no languages at all, i.e. your template should not even begin to construct a table.
make_language_row can't expand the lang template directly in the Module. You should use the frame object's expandTemplate function to handle this.
I've put in some changes to your template as a demo. I'll let you sort out the styling & article links etc..
Thanks! Well, I was under the impression that my module wouldn't be able to call local functions in other modules.. but again I'm quite a newcomer to Lua, and Wikipedia's particular flavor of it. A diehard editor (talk | edits) 03:07, 27 November 2023 (UTC)
I recently updated my phone from iOS 16 to iOS 17.1.1. When I access Wikipedia’s desktop site on my phone, every page I load starts out zoomed in about 200%. I can pinch to fully zoom out, but it’s annoying that I have to do that on every page. Also the width of the search bar is messed up. Are these known bugs, or is this because of some sort of browser setting on my end? —pythoncoder (talk | contribs) 20:16, 21 November 2023 (UTC)
Might be a setting, because i'd expect a lot of people to report about such a problem if it were widespread. Have you checked the fontsize option (under the Aa button) in the browser ? It's persisted per website, so if you accidentally change it, you are stuck with it until you reset it to 100%. —TheDJ (talk • contribs) 23:10, 21 November 2023 (UTC)
I'm trying to pass a parameter to #if which begins with a space. Essentially, I want it to output nothing if quiet=1, ELSE output a space followed by some more output.
But the template engine trims the preceding space and I can't figure out how to escape it. I can use Template:Spaces, but that produces and requires the inclusion of a template which surely shouldn't be required. I'm sure this must be an easy fix, but I can't for the life of me find the relevant documentation pages. Can anyone help?
If you want to see my code, I'm using the sandbox at Template:X28.
Can you clarify what is still broken for you? The bug and the patch you linked are for the Pywikibot framework, but later you talk about MediaWiki, so I am a bit confused. The previous thread you linked is pretty long and also confusing. Matma Rextalk13:44, 20 November 2023 (UTC)
Just tried purging a page with forcelinkupdate=1 via ApiSandbox. I can see the page_links_updated value in database is updated. I don't think anything is wrong in MediaWiki; it could be an issue with the client code used. – SD0001 (talk) 13:52, 20 November 2023 (UTC)
THIS and THIS is the code for my BRFA-approved bot which is running on the Toolforge. It was working fine in August, when I pretty-much wrapped up my project to install this automated process. I've not made any (breaking) code changes since then, as I thought I was done tweaking the code to the point where it just ran itself and no longer needed close supervision. The first page it is attempting to refresh links for is Soygu (which is a redirect). The bot is stuck repeatedly attempting to purge this page, and no error code is returned to indicate there was any problem. Quarry query shows that the date is not being updated. Likely it was last updated by my bot on August 17, 2023. – wbm1058 (talk) 16:48, 20 November 2023 (UTC)
It seems to me that your bot has successfully purged that page with recursive links update about 50 times per day for the last couple of days (and completed 750,000 other purges every day – I'm surprised we have the logs for this, but we do: [23]). It's interesting that page_links_updated is not being updated, I will try to find out if that's expected or when it changed. It seems like a different issue than the last time (and it seems like the purges are working). Matma Rextalk18:26, 20 November 2023 (UTC)
When viewing a Gerrit patch, if you click on the three dots at the top-right and select "Included in", it tells which branches the patch has been merged into. If it were a MediaWiki patch, you can then cross-reference with toolforge:versions to determine if it has been deployed yet. Here however, the patch is to Pywikibot. I see it is included in the 8.5.0 and 8.5.1 releases, so you'll want to make sure your Pywikibot is updated to one of those versions in order for the change to take effect. — MusikAnimaltalk18:46, 20 November 2023 (UTC)
Thanks for the helpful information. I realize now that I incorrectly assumed it was a MediaWiki patch. I don't think I have access to the Logstash, which would have helped me verify my bot's actions. I guess I could apply for nda group access for researchers and volunteers, if I sign an NDA for access to confidential data?
page_links_updated was definitely getting updated until sometime after August, so answering the question "why did it stop getting updated?" should hopefully lead to the solution. wbm1058 (talk) 19:50, 20 November 2023 (UTC)
(Logstash is indeed not public, and you could apply to get access, but you're not missing much in this case – I just linked to log messages that indicate that your API requests reach this point in the code: [24]) Matma Rextalk20:41, 20 November 2023 (UTC)
I've looked for any recent changes to MediaWiki that would affect this, and I didn't find any. Then I tried doing a similar request using Special:ApiSandbox, and it worked for me – the page_links_updated field got bumped. So I tried running your code, and… it worked too. I'm definitely confused.
My current theory is that your bot is successfully scheduling the links update (as indicated by the log messages), but they are not actually happening for some reason.
In order to test that theory, I'd like to ask you to do two things, and run the script just once with these changes:
Ändern Sie curl_setopt($this->ch,CURLOPT_HTTPHEADER,array('Expect:')); to curl_setopt($this->ch,CURLOPT_HTTPHEADER,array('Expect:','X-Wikimedia-Debug: backend=mwdebug2001.codfw.wmnet; log')); in botclasses.php line 109
The first will show you the API result, just in case it contains some interesting warnings. The second will enable debug logging on the WMF servers for these requests (wikitech:WikimediaDebug), I'll be able to view the results in Logstash and perhaps we'll find out what is happening.
OK, I've shut down my bots while we debug this, since they weren't working right anyway... toolforge jobs delete refreshlinks and toolforge jobs delete refreshmainlinks
I started testing this by doing the purges at a slower rate, and then ramped up the speed, figuring I would refresh-links as fast as I could, until the system couldn't keep up... I think the problem may be that the system doesn't have a good way of telling me that it's not keeping up. Now I'll make the code changes and run the test that you requested. – wbm1058 (talk) 21:41, 20 November 2023 (UTC)
I also commented out the last line of refreshmainlinks.php so that it wouldn't run forever: #goto doit_again;
OK, it's refreshed 10,000 links in 34 1⁄2 minutes, so you should have plenty of X-Wikimedia-Debug stuff to examine in the logs now. – wbm1058 (talk) 22:47, 20 November 2023 (UTC)
Well, I don't know why this is happening. I filed a task about it, so that someone else from my team might have a look with fresh eyes, feel free to follow along: T351729. It might take us a few days to get around to this (it's the holiday and flu season), I hope the problem is not urgent. Matma Rextalk15:24, 21 November 2023 (UTC)
I picked two random pages with August "links-updated" dates (verified by a Quarry query), then logged into the Bot1058 account (using the normal password, not the BotPasswords password) and performed the API sandbox purge. It worked!
Noting one difference, in the API sandbox results, "purged": true, and "linkupdate": true which in the console log they are null: [purged] => and [linkupdate] =>
Right, thanks. I added &formatversion=2 to the API call and ran the test again. Returned "1" rather than "true" but I suppose that's just a matter of variable-type declarations.
Finished another 10K edits run with X-Wikimedia-Debug stuff, under a different bot account of mine. Checking with Quarry for a couple of the pages that it was supposed to have purged, they still show page_links_updated August 17. So it still fails when running under a different account. Hmm, wbm1058 (talk) 22:43, 21 November 2023 (UTC)
The template adds an extra line break (?) during render and I don't know why. Example Thridrangaviti_Lighthouse#Technical_features where the last bullet uses the template, there is an extra line after it.
I am unable to replicate the problem in an example like this:
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 28 November. It will be on non-Wikipedia wikis and some Wikipedias from 29 November. It will be on all wikis from 30 November (calendar). There is no new MediaWiki version next week. [25][26]
MediaWiki's JavaScript system will now allow async/await syntax in gadgets and user scripts. Gadget authors should remember that users' browsers may not support it, so it should be used appropriately. [27]
The deployment of "Add a link" announced last week was postponed. It will resume this week.
CodeMirror button for wikitext syntax highlighting is gone
I'm using source editor with editor toolbar ("2010 wikitext editor"). But the button for syntax highlighting is gone. It's called "Activate highlighting in wikitext" at Help:Edit toolbar. It is supposed to be between dropdown "Advanced" and puzzle-piece icon button . Instead, I see button called "Reference" .
@Andrybak: You need to disable the syntax highlighter gadget; this is unrelated to the CodeMirror extension. If you're wondering how it could possibly interfere while in safe mode, CodeMirror apparently checksfor for this gadget (and also WikEd), and won't load if it's enabled. Suffusion of Yellow (talk) 18:54, 26 November 2023 (UTC)
This is a recent change, introduced in commit c0b0100 (Add $wgCodeMirrorConflictingGadgets instead of checking wikEd directly, 2023-10-10). The change seems to come from phab:T178348, quote Investigation: CodeMirror and WikEd gadget (and maybe other syntax highlighting gadgets) don't get along well
I couldn't find any mention of DotsSyntaxHighlighter in any related tickets or commits. So it seems, that the combination CodeMirror+DotsSyntaxHighlighter is only hypothetically causing problems? The combination works fine for me. I used DotsSyntaxHighlighter most of the time, and enabled CodeMirror on a case-by-case basis for complicated template edits.
MusikAnimal, in my experience DotsSyntaxHighlighter loads just the tiniest bit snappier than CodeMirror. CodeMirror adds just enough friction for my gnomish edits, so it is turned off most of the time.
Could someone please go to Cupcake#Themes and click on one of the images? The gallery higher up on the page lets me click through to the image (licensing/copyright implications here), but this gallery doesn't let me do that. WhatamIdoing (talk) 20:12, 27 November 2023 (UTC)
I have a patch up to fix it now. After it merges, galleries with the issue will unfortunately need to be fixed up manually in source mode. DLynch (WMF) (talk) 17:55, 28 November 2023 (UTC)
This search currently finds 428 possible mainspace cases. That's doable with AWB if somebody (not me) wants to work on these which break the image license in most cases by not linking to the file page with attribution and license. PrimeHunter (talk) 19:12, 28 November 2023 (UTC)
Hello! Recently, I sent emails directly to an @wikipedia.org address and emailed a user via Special:EmailUser founction. However, both got no response. Is there any technique issues or am I blacklisted? How do I fix it? Thanks.--Whisper of the heart03:03, 22 November 2023 (UTC)
@Whisper of the heart: Your Wikipedia mail should be working. Have anyone said they don't receive mails from you or are you merely wondering why they haven't replied? That's not a technical issue but you can mail yourself or me as a test if you want. PrimeHunter (talk) 20:20, 22 November 2023 (UTC)
@Whisper of the heart: If you email somebody via Special:EmailUser, the recipient is not obliged to respond. Indeed, over the last fourteen years about twenty people have sent emails via Special:EmailUser/Redrose64 and IIRC I only replied to one of them, and that was concerning when this RfA should be filed. You might wonder why I have email set up if I have no intention of responding: it's in case I forget my password (this has actually happened). --Redrose64 🌹 (talk) 22:37, 22 November 2023 (UTC)
@Redrose64: Make that at least two; you responded to mine :-) ... I do intend on following up on that one properly (by making my way to your neck of the woods) within the next year ... this is not a threat. :-) Graham87 (talk) 10:11, 23 November 2023 (UTC)
The boilerplate appended to mail sent through Special:EmailUser actually warns against responding, at some length. —Cryptic22:46, 22 November 2023 (UTC)
You also mentioned a direct email to a @wikipedia.org address. I don't know about other addresses, but if that was emergencywikipedia.org, you should definitely have received a reply, probably within about 10 minutes or so. If you didn't, don't assume your message has been read. Suffusion of Yellow (talk) 23:41, 23 November 2023 (UTC)
I'll add that the emergency email seems to have been having problems with emails being directed to the spam folder. This may or may not be the same with whichever wikipedia email address Whisper of the heart was directing their email towards. EggRoll97(talk) 01:31, 25 November 2023 (UTC)
@Suffusion of Yellow and EggRoll97: The groups I emailed was the checkuser team, in fact I still get no reply so far so I believe my mail was lost or spamed, but luckly I got in touch with them by other way... I think maybe their mailbox have the same problem as the emergency email's.--Whisper of the heart18:39, 29 November 2023 (UTC)
Error message
Yesterday I moved a page and today I get the following error message when I try to view the page view: Error querying Pageviews API - Not Found
"I moved a page" i takes at least a day before pageview information becomes available. So if you moved the page yesterday, it cannot have collected pageviews for the new title yet. —TheDJ (talk • contribs) 21:49, 29 November 2023 (UTC)
Template sidebar design alternative?
Based on my reading of Template:sidebar, 65% of our readers do not see the sidebars (they are mobile) and no technical fix is likely or even planned AFAICT.
In Wikipedia:WikiProject_Physics the template sidebar is used for contextual navigation in many articles. Unlike the "See also" or wikilinks in the page, the sidebar acts as a stable, well-ordered, annotated, consensus global view of sub-topics in physics.
Thank you for raising this. I came across this exact problem when adding {{LGBT sidebar}} to GaLTaS. No-one can see it on mobile devices unless they switch to Desktop view, and they don’t know what they’re missing unless they do. Chrisdevelop (talk) 08:49, 30 November 2023 (UTC)
This query shows the largest page in bytes at time of promotion to FA. Getting the number of prose words would be difficult, and impossible with the Quarry tool that WP:QUERY uses. Certes (talk) 19:51, 29 November 2023 (UTC)
@Certes, thanks v much! i've rerun the query going back as far as possible i.e. 2014-2023 [30] I can manually retrieve the prosable word size for the articles you've helpfully found, thanks again! Tom B (talk) 11:10, 30 November 2023 (UTC)
@PrimeHunter: Thank you for your reply. I reverted back to the default Vector 2022 skin, and i see no such icon. But ah, if i hide the main menu from the left sidebar, then i get the menu icon at the top left. Regardless: lo and behold, the TOC appears now. —Jerome Potts (talk) 12:25, 30 November 2023 (UTC)
I'm trying to use Template:Scalable image on some help pages and my user page which will appear on Wikipedia and Meta. However the template doesn't exist on Meta yet. It uses some other templates and I don't want to try to do it and make a mess I have to ask someone else to clean up, can I request someone else please copy it over for me? It will be really useful for creating documentation. I know this is a request for Meta fom Wikipedia but there are very few people who look at the notice boards there.
For 2 days now, Quarry has automatically stopped any SQL that runs for longer than 60 seconds.
Unfortunately, this means that certain queries, e.g. searching for syntax errors in URLs, are no longer possible. (e.g. Quarry:query/77794 for dewiki expected result 0-20 recordsets within 120-180 seconds, for enwiki correspondent Quarry:query/78127]] needed 2543 seconds with ~10k recordsets)
I noticed this too, but assumed it was just me accidentally clicking the Stop button. The limit used to be 30 minutes. (Perhaps someone tried to raise that to 60 minutes and forgot it was in seconds.) Certes (talk) 12:39, 30 November 2023 (UTC)
It seems like I missed setting an explicit timeout when moving the wiki replicas to a different set of load balancers, and the defaults set were set to much lower limits that would be more appropriate for HTTP connections. I deployed a fix at around 11:00 UTC today after someone pointed out the same issue on IRC. Taavi-WMF (talk) 12:57, 30 November 2023 (UTC)
I actually thought the limit was longer than 30 minutes? When trying to identify candidates for mass draftification I’ve sometimes needed to run scripts for times considerably in excess of that - over an hour in some cases, I believe. BilledMammal (talk) 13:40, 30 November 2023 (UTC)
On Wikipedia:Vital articles we have a number of headers of level 1, 2, 3 and 4. The heading "General" appears nine different times on the page, under different parent sections. Is there any way to reliably link to the "General" level 3 section which comes under "Mathematics" level 2 section? Or should we use {{anchor}} for this? I notice that using an underscore and number, e.g. Wikipedia:Vital articles#General_9, works, but would presumably break if one of the other General sections was removed. — Martin (MSGJ · talk) 13:10, 1 December 2023 (UTC)
"General" tends to be the first subheading within each section, so you could almost link to the higher level heading instead. I say "almost" because, sadly, the actual heading is "Mathematics (45 articles)" so you'd need to change the link every time the number of articles changes. This looks like a job for anchors. Certes (talk) 15:26, 1 December 2023 (UTC)
The count should probably not be in a section header at all. Section headers should be stable (they also make them a bit longer). The count can be plain text right under the section headers. Gonnym (talk) 15:44, 1 December 2023 (UTC)
Why not rename the "General" header to "<type> general"? So "History general" or some other word with the topic name. Gonnym (talk) 15:43, 1 December 2023 (UTC)
@Aram we offload that gadget to mediawikiwiki, so you will want to engage someone on that project to make changes to it. The only thing we could really do would be to just disable it project wide. As that "show a prompt" preference doesn't appear to be on by default this doesn't seem like an urgent issue? I've added a warning on the description to our gadget here. — xaosfluxTalk21:38, 30 November 2023 (UTC)
@Xaosflux Thanks, that’s not bad for now. I think enough people are aware of this problem now that there is no need for me to tell anyone else. ⇒ AramTalk16:25, 1 December 2023 (UTC)
I just got it working using Special:Preferences. It was giving me the error when I loaded it from metawiki global.js using mw.loader.load('https://en.wikipedia.org/w/load.php?modules=ext.gadget.Navigation_popups');. This used to work, but stopped working for me within the last day or two. Interesting. CORS maybe? –Novem Linguae (talk) 09:40, 1 December 2023 (UTC)
We're experiencing this issue on Wiktionary, too (wikt:Wiktionary:Grease_pit/2023/December#is_popups_disabled?), and it seems to disable(??) other gadgets and js when it fails, like the js Wiktionary uses to expand collapsed tables. Loading navigation popups a different way reportedly fixes not only the navigation popups but also the other, seemingly unrelated gadgets that broke. It's odd. -sche (talk) 01:46, 2 December 2023 (UTC)
IMDb
In recent days, an editor has been going around changing the IDs in films which were using {{IMDb title}} from the actual IMDb ID to a qid= link to the film's Wikidata page. The template certainly features that option, so it's not a problem per se — but it's having undesirable side effects in certain cases, that need to be addressed: specifically, if it's done in any template that features extra clarifying "description text", such as {{IMDb title|0928413|Naked Boys Singing!|(2007 film)}} being switched to {{IMDb title|qid=Q6960446|title=Naked Boys Singing!|(2007 film)}} at Naked Boys Singing!, then using the Wikidata qid instead of the IMDB ID breaks the template display from the proper "Naked Boys Singing! (2007 film) at IMDb" to the improper "film)/ Naked Boys Singing! at IMDb", and files the article in Category:IMDb template with invalid id set.
Interestingly, by far the majority of the couple of dozen cases that were in that tracking category today were plays where a film adaptation already had its own separate Wikipedia article about the film, and thus the film adaptation's IMDb profile wasn't even needed on the play's article in the first place — so in those cases I just removed the template outright, but in the case of Naked Boys Singing the film doesn't have its own separate article yet and so I had to revert the qid change. I also question whether there's any reason why using the Wikidata qid instead of the direct IMDB id would be critically important enough to justify doing an active switchover on thousands and thousands of film articles like this — it's fine to have as an option, but is there any reason why it should be the gold standard? — but at the very least the template needs to be modified to ensure that using the qid doesn't break it.
So, could somebody with more experience in template coding than I've got fix this so that the template displays properly if the Wikidata qid is used? Bearcat (talk) 22:30, 29 November 2023 (UTC)
Since this is about me, I thought I would appear and explain what I am doing... To start off, the problem listed above was fixed 2 days after I made that edit and isn't an issue anymore. I am using Category:IMDb ID not in Wikidata to find pages with invalid imdb links, update outdated ids (example) and also create wikidata items for episodes/films/series where they don't exist. The simplest way to do this is to just set a wikidata item instead of an id when it is checked, so that it is removed from that category. Considering the fact that the imdb id is a pretty static value (Set it and forget it) it shouldn't really matter if it is set on wikidata or the wikipedia page. I don't think the template needs updating this was just a case of me being careless... I do manually check the edits I make but sometimes you miss things when you make a lot of edits at once. Any questions / concerns please ping me. Thanks Terasail[✉️]09:33, 30 November 2023 (UTC)
To fix the issue at Naked Boys Singing!, the fix would be: {{IMDb title|qid=Q6960446|title=Naked Boys Singing!|description=(2007 film)}} (though because the title is the same as the film, just do {{IMDb title|qid=Q6960446|description=(2007 film)}}) Gonnym (talk) 09:47, 30 November 2023 (UTC)
There's a general presumption against using Wikidata, partly because vandalism there is less likely to be detected. Certes (talk) 22:57, 1 December 2023 (UTC)
I have recently noticed a bug – in DYK nomination archives, reply buttons are shown (see, e.g., Talk:Australiformis). I am almost sure that this is caused by the fact that the archival div (in which the nomination is enclosed) is not of the mw-archivedtalk class. This seems to affect all archived DYK nominations. I have two questions: Could this be fixed? Why is the faulty archive top code repeated over the archives rather than transcluded from a single template (e.g. something like {{atop}})? Janhrach (talk) 11:08, 2 December 2023 (UTC)
@Janhrach: The archive (which was also where the discussion took place) is Template:Did you know nominations/Australiformis. Talk:Australiformis transcludes it. Template space isn't meant for discussions and doesn't have reply links so the actual archive (also applies to open nominations) doesn't have them but the transclusion to a talk page does. I have added class="mw-archivedtalk" to {{DYK top}}[31] per mw:Help:DiscussionTools/Magic words and markup to omit reply links in archived discussions. The template is always substituted so it will only affect pages archived after now. The organization of DYK pages can be discussed at Wikipedia talk:Did you know. A bigger issue is why nominations are in template space (they think it would be too much work to move them). PrimeHunter (talk) 12:15, 2 December 2023 (UTC)
It seems to that the number of pages went down during the last 24 hours from 6.752xxx million articles to 6.750xxx million articles. I have never seen this happen and wonder what could have caused this. It usually goes up by a few hundred per day.-TonyTheTiger (T / C / WP:FOUR / WP:CHICAGO / WP:WAWARD) 12:23, 2 December 2023 (UTC)
I didn't find anything relevant in the deletion log or move log but mw-new-redirect at Special:Tags showed that NmWTfs85lXusaybq has been mass-redirecting disambiguation pages with only one target article like [32]. Disambiguation pages count as articles in statistics. Redirects don't. PrimeHunter (talk) 14:58, 2 December 2023 (UTC)
It sounds like a noble task, but the example you pointed to was two articles not one. I don't understand how you can redirect to two different pages? Should we redirect 2 subject dab pages?-TonyTheTiger (T / C / WP:FOUR / WP:CHICAGO / WP:WAWARD) 17:21, 2 December 2023 (UTC)
Does any of this have to do with the current Admin afdstats down? I'm a little curious, as the afdstats also went down on Oct 12 and Oct 24. — Maile (talk) 03:59, 2 December 2023 (UTC)
I see "Keep" and "Not closed yet" on the stats grid for your Ross Monroe Winter AFD. That looks correct to me. It's currently about the 25th row down. –Novem Linguae (talk) 15:53, 2 December 2023 (UTC)
Thank you! It's kind of out of chronological order, but it's there. Looks like it picked up a date from where I added the signature of the nominator, who had forgotten to sign. — Maile (talk) 18:59, 2 December 2023 (UTC)
Article contents navigation menu has disappeared
There used to be a article table of contents navigation menu on the left hand sidebar of the latest skin. It seems to have disappeared. I was please with the choice of moving the TOC to the sidebar to the article body; but now, without the TOC in the sidebar, we have no TOC at all.
I finally found the icon next to the article title. I think this must have been switched on quite recently. I see I'm not the only person who's been bitten by this, see #Invisible_table_of_contents. Nor is the only sidebar bug that's been reported recently.
This is absoulutely terrible UI with no discoverability: who thought this was a good idea, and how can we get report this to the relevant people who can get this fixed? — The Anome (talk) 14:22, 3 December 2023 (UTC)
For me I have to explicitly hit the "hide" button on the TOC to move the TOC to the heading. Is your screen particularly narrow - it seems that if your screen is super narrow than the TOC automatically moves there which is bad UX. Galobtter (talk) 16:37, 3 December 2023 (UTC)
One man's bad UX is another man's "almost ready for use on mobile". I dream of the day we only have one default skin. Izno (talk) 18:02, 3 December 2023 (UTC)
A single skin, yes, but not everything should be the same for all devices. Tiny devices are not the same as desktop/laptop screens, no matter how high the DPI. Breaking the desktop to make mobile better is not cool, nor vice versa. That's what CSS media selectors are for, and I note Vector 2022 has a fair few of them in its CSS already. They need to be used better, and more carefully. — The Anome (talk) 21:14, 3 December 2023 (UTC)
The irony is that they already use media queries - that is fundamentally how JavaScript decides to be wrapped up in a little button at narrow widths. Izno (talk) 21:41, 3 December 2023 (UTC)
defined? It's not in MediaWiki:Common.css. The problem: at Wikipedia:WikiProject Doctor Who/Article alerts under MonoBook skin and Firefox 120.0.1, the box at the top (with the TARDIS/police box image at each end) overflows the right-hand edge by a tiny bit, causing a horizontal scrollbar to appear. The page source has <div class="center" ...> and I have found that suppressing the width: 100% declaration stops the overflow from occurring. --Redrose64 🌹 (talk) 21:00, 3 December 2023 (UTC)
The class is defined in core MediaWiki and was originally intended primarily for centering images (well, to remove what was then use of <center>...</center> -- which was in the image centering code). It spread both onwiki and code-side. Wiki editors now use it for other reasons when they perhaps should not. (See also phab:T282063#7064571.)
As to the specific motivation, I think the appropriate fix is simply to remove the class. Divs are already 100% by default allowing some leeway for internal content, so this necessitates adding the text-alignment. The other fix would be adding box-sizing: border-box if for some reason you think the class merits keeping. I favor removal in this case. Izno (talk) 21:25, 3 December 2023 (UTC)
Because WantedCategories has a hard limit to how many categories it can detect, those of us who work with it can never leave any redlinks unaddressed: because they'll just pile up and push the report ever closer to that limit if there are redlinks left unaddressed by the time a new run happens 72 hours later, we always have to ensure that every single redlink in that list is always resolved before the next run happens. But since I don't ordinarily work with Wikiproject templates, I can't just edit them to make the redlinks disappear without running the risk of breaking other stuff, and I can't just wrap the template in {{Suppress categories}} because that will suppress legitimate bluelinked categories too — so with no way to just make the redlinks go away, I'm left with no choice but to invest significant amounts of time into creating the relevant categories so that they'll be resolved from the report.
Sometime after that, somebody who's actually associated with the Wikiproject in question deems those categories as unwanted, and wipes the code back out of the template so that the categories are emptied back out. But, of course, if they were genuinely not wanted, then they never should have been generated in the first place.
As a result, I get carpet-bombed with an unending barrage of creator-notifications that the categories are up for speedy deletion as empty categories — I've gotten at least 35 of these just in the past month alone.
I'm bloody sick and effing tired of this, and it needs to stop. Templates aren't supposed to be generating redlinked categories in the first place, so the templates shouldn't be generating or populating redlinked categories in the first place, and I'm tired of taking the shrapnel from a cycle of unwanted categories getting generated and then ungenerated by a lack of coordination around wikiproject template edits.
It's empty now precisely because it's gone through the exact "populated as a redlink, created accordingly, and then depopulated and deleted" cycle I was describing (which is why I chose it as an example, because it's evidence of the cycle I was talking about.) If steps were already taken to reduce this, they certainly haven't reduced it much, because new template-generated wikiproject assessment categories keep turning up at WantedCategories — I had to create another passel of them literally just yesterday. Bearcat (talk) 22:41, 2 December 2023 (UTC)
Category:Redirect-Class Indian history articles of Unknown-importance was created 4 October 2023 and nominated for deletion as empty 3 November 2023 where it had probably been empty for a while. It's hard to track categorization more than 30 days ago (a MediaWiki time limit) but I trust it had content when it was created. Thanks for all your category work. I hope you don't feel blamed when you say "taking the shrapnel". It's just standard notifications and you can remove them per WP:REMOVED (may be an exception). 35 in a month may be annoying but you make thousands of monthly edits so I guess they take relatively little of your total time here. We don't like red categories on articles but are less concerned with having them temporarily on talk pages. PrimeHunter (talk) 22:51, 2 December 2023 (UTC)
WP:REMOVED says the user may not remove speedy deletion tags. I think that only means when a userspace page itself is tagged for deletion, and does not apply to notifications that other pages are tagged. PrimeHunter (talk) 23:38, 2 December 2023 (UTC)
This is MediaWiki:Enotif body if we want to change it locally, but I don't recommend creating a customized version for such a small change to a long message. The MediaWiki default might get more significant changes later, e.g. to describe a future feature or behaviour. PrimeHunter (talk) 14:36, 3 December 2023 (UTC)
Getting in a mess with templates trying to make a nice looking help page
Hi all
I'm working on a help page and the best way to present the information is having text on the left hand side with a title and subtitle and an image on the right which covers about 50% of the width of the page, the table should consistently cover the width of the page. Basically the structure like this
Titel
Untertitel
Body of text
Image
Image description
I know this is quite elaborate but its extensive documentation which needs illustration and needs to be attractive for an external audience of professionals.
I've worked out I probably need to use the flex columns template, and also the scalable image template. These two work independently of each other and look good but when I try and glue them together all the text just disspears even though its in the template in the source editor...
Could someone who knows about templates help me understand what I'm doing wrong? I'm not sure if I'm making a small formatting error or if I'm trying to use two things together that don't like each other...
You appear to be running into the problem where an = cannot appear inside an unnamed template parameter, so your first parameter is being interpreted as being named <h3 style with its value beginning "height:1%; font:25px/1.2 .... Try using explicit parameter names |1= and |2= in the call to {{flex columns}}. Anomie⚔12:58, 4 December 2023 (UTC)
Hi Anomie thanks very much for your explanation, I'm not sure I understand completely, I've tried adding adding numbers in front of the pipes which has brought back the body text but not the heading text in the h3 code... I've removed the h3, but honestly the documentation won't work without it because I need to allow people to use an index jump to different sections. I think this is a bit beyond my ability, if its easier please feel free to just change it or point me to a page where I can ask or work it out. Its pretty important I have these headings in the text. Thanks again, John Cummings (talk) 17:41, 4 December 2023 (UTC)
Hi Anomie thanks so much, I changed it to 100% and I made the text appear! :) (hope you don't mind me editing your sandbox) but now there is a gap of the whole of the height of the image between the heading and the body text, I have tried removing any kind of padding between them but it doesn't work at all... :( Two steps forward, 1.99999999 steps back, John Cummings (talk) 19:32, 4 December 2023 (UTC)
You can remove the height attribute entirely. I'm not sure what you're trying to achieve by setting the height of the heading element? isaacl (talk) 19:49, 4 December 2023 (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
The spacing between paragraphs on Vector 2022 has been changed from 7px to 14px to match the size of the text. This will make it easier to distinguish paragraphs from sentences. [33]
The "Sort this page by default as" feature in VisualEditor is working again. You no longer need to switch to source editing to edit {{DEFAULTSORT:...}} keywords. [34]
Changes later this week
There is no new MediaWiki version this week. [35][36]
Registration & Scholarship applications are now open for the Wikimedia Hackathon 2024 that will take place from 3–5 May in Tallinn, Estonia. Scholarship applications are open until 5 January 2024.
I'd like to DOUBLE point out to people the Grid Engine shutdown process that is coming up. This has been planned for over a year, but 450 tools have not switched their code yet, so those are bots and tools that might start to go missing in December if people do not take action. Please check the list for the tools that you use and see if you can contact the maintainer who might have missed all the earlier announcements of this. If you have some time available, maybe offer to help out your fellow tool maintainers, or if a tool is completely abandoned, you might want to adopt it. —TheDJ (talk • contribs) 09:31, 5 December 2023 (UTC)
(i) Does anyone know how you introduce colour to a table background, as here?
(ii) Do people think it's effective? (I do—perhaps a slightly paler yellow, though.) Tony(talk)03:12, 6 December 2023 (UTC)
The count (as well as most wiki content) is cached for performance reasons. You can refresh it by purging. Articles are created pretty quickly over here on enwiki, so the counts will likely always be a little bit off depending on when the page cache was last purged. Special:Statistics or using mw:API:Siteinfo should be more exact. — MusikAnimaltalk20:31, 5 December 2023 (UTC)
Several times, I have accidentally logged out while trying to view my contributions page. I think that when you press Log Out, there should be a pop-up where you confirm that you actually want to log out and did not just press the button accidentally. QuicoleJR (talk) 14:10, 6 December 2023 (UTC)
@QuicoleJR: Fred Gandt's script didn't work for me either, but Guywan's did. Paste mw.loader.getScript("/w/index.php?title=User:Guywan/Scripts/ConfirmLogout.js&action=raw&ctype=text/javascript"); // Backlink: [[User:Guywan/Scripts/ConfirmLogout.js]] into your common.js, and it should work. Cremastra (talk) 22:00, 6 December 2023 (UTC)
You can try accessing your My Contributions page using a hotkey. In Chrome you can use Alt-Y or Alt-Shift-Y, I think. hotkeys vary by browser and operating system, but the one for My Contributions will contain the letter Y. –Novem Linguae (talk) 14:27, 6 December 2023 (UTC)
Edit conflict-like situation that recreated a draftified article
The other day in working on some articles in CAT:UNCAT, I came upon Mulashidi. I added a category and removed the uncatted tag, and there my interaction with the article ended, or so I thought. Then Liz came to my talk page to ask why I had recreated an article that had been draftified to Draft:Mulashidi. Huh? I looked at the article history and saw to my surprise that I had "created" this article. Extra huh??? Then I looked at the logs, and saw that SunDawn's draftification and my "creation" of the article happened in the same minute. Apparently I was finding a category at the same time another editor was moving it. I was not, to my recollection, asked if I wanted to create an article that didn't exist. This may be an oddball fluke, but since it happened to me, it could happen to others as well. The system needs some sort of way to handle such things that aren't checking the history to make sure you didn't just accidentally recreate an article that existed before you started editing. Thanks to Mandarax for looking at the situation and suggesting I bring it here. LadyofShalott13:04, 1 December 2023 (UTC)
Not just HotCat; something similar has happened to me a few weeks ago when the article was deleted at the same time as I moved it to draft. Espresso Addict (talk) 23:15, 6 December 2023 (UTC)
Hidden ToC dis-appeares on it's own during scrolling
do click into this little field to the left of my user name,
the ToC does open downwards.
Then I move my mouse pointer into this ToC and
then I start scrolling down ( with a side sweep on my touch pad ) within this ToC, which means: the box with the ToC ( including the whole page i.e. what is out-side of the ToC ) is moving up-wards. - So far, so good.
But when a few upper parts of my 4th main section have appeared,
the whole ToC disappears - and this is not what I expect and want.
The result is:
The only way I can reach the lower sections of my ToC is by moving the ToC to the left---besides scrolling down the actual content.
What I want:
Once the Table of Content is dropped down, it should be scrollable down to the last line, and not dis-appear mid-way of scrolling.
@Steue tried to duplicate, here is what I saw - is this what you see? If you scroll down enough so that the top of the collapsed TOC scrolls off the page, it moves (collapsed) to the header, to the left the of page title which also moves there. — xaosfluxTalk21:20, 6 December 2023 (UTC)
@Steue: I've opened bug phab:T352930 on this. Please feel free to add additional information there, or subscribe to it for updates. This is not a condition that is happening only here on the English Wikipedia, but one that would need to be resolved in the software. — xaosfluxTalk23:44, 6 December 2023 (UTC)
@xaosflux: Thanks for the info and especially the task to Phab, because I don't have an e-mail address, so I can't register to Phab.
Removal of any one of them seems to remove this, perhaps there is an extra line break in one of them? There is also a hidden section name before the lead paragraph starts if that might complicate things. Could also be related to paragraph breaks seemingly having gotten bigger sometime last month? 93 (talk) 04:11, 7 December 2023 (UTC)
There is nothing to be done about this, one too many templates in a row, even empty will do this. The best you can do is pick one and move it up a line.
Pale Moon is not a supported browser and the 11 year old Mac OS 10.8 is not easy to test against either, so it is really difficult to assist with debugging a problem like this. Have you tried the Pale Moon support forums ? These are probably more likely to have experts who can figure out why specific pages won't work in your combination. —TheDJ (talk • contribs) 09:22, 7 December 2023 (UTC)
Redlinked assessment categories, again
There's been a recurring issue at Special:WantedCategories with the "vital=" flag in a template autogenerating assessment categories that not only don't exist, but don't even seem to correspond to any established assessment queues at all.
But try as I might, I can't find hide nor hair of a reason why the vital= flag is filtering some articles into different categories than it filters other equivalent articles. Could somebody with more experience working with WikiProject templates look into this? Bearcat (talk) 17:02, 7 December 2023 (UTC)
I think this is because the topic is different from the section title, please refer to User:Cewbot/log/20200122/configuration#Topics. This issue is expected to be solved by the following steps:
I’ve noticed that, on the mobile website - for articles that transclude {{Infobox gender and sexual identity}} - the notice that states ‹ The template [X] is being considered for deletion › is displayed at the top of the page (and clashes with the opening sentence), rather than on top of the infobox itself (as I assume it should be doing). As an example, see this screenshot of the Transgender article.
I’m assuming that this is something that may (hopefully!) be able to be fixed by tweaking {{Template for discussion/dated}}, however, I’d have no idea where to start figuring it out myself! Is anyone aware if there’s any way to force the deletion discussion notice to stick to the top of the infobox on mobile web?
@Ahecht One problem with adding nomobile to Template:Template for discussion/dated would be that it would mean people on mobile viewing pages that transclude templates at TfD wouldn’t be notified of the ongoing deletion discussion - and thus some of those who may be affected by a template’s deletion wouldn’t know to take part. Best, user:A smart kittenmeow10:11, 7 December 2023 (UTC)
Ahecht only added nomobile to the deletion tag on {{Infobox gender and sexual identity}}. I think infobox templates are rarely nominated. We could make a new class which tries to counteract the unwanted sideeffects of infobox but maybe somebody can find a cleaner way to move a tag along with an infobox in mobile. PrimeHunter (talk) 16:41, 8 December 2023 (UTC)
This is the first time in ever that I can think of where a TFD/M notice has been complained about in the context of this problem (and no, it has been more rather than less common to nominate infoboxes for TFM). I think documenting the weird placement on mobile on the relevant template page is sufficient. Izno (talk) 18:32, 8 December 2023 (UTC)
Triangles in Special:Watchlist too big and pointing wrong way
WP:ITSTHURSDAY I imagine. If you set your watchlist settings just right, each page's revisions will be collapsed until you click on a triangle. At that point the triangle will go from facing down to facing right, and the page's revisions will go from hidden to shown.
Anyway, this triangle recently doubled in size, and now always points down (won't point to the right when you click on it). This is on Vector and Vector 2022. Was not able to reproduce for Monobook.
The problem goes away in safemode, so the bug fix may need to be made in a local gadget related rather than MediaWiki core.
I suspect Vector changed the name or nesting of one of their elements, breaking our local gadget.
Files that might need repair (search for mw-enhancedchanges-arrow):
Please don't. phab:T352456 fixed the watchlist display for me. There is an image in that task that shows how the arrows in my watchlist looked before the fix. Now uncollapsed watched items are marked by right-pointing arrows (much the same as collapsed categories on category pages). When uncollapsed, the watchlist arrows point down, again just like uncollapsed categories on category pages.
For me: chrome current, win10; at Special:Preferences#mw-prefsection-watchlist: Expand watchlist to show all changes, not just the most recent, Use non-JavaScript interface; at Special:Preferences#mw-prefsection-gadgets: Display green collapsible arrows and green bullets for changed pages in your watchlist, page history and recent changes (unavailable with the improved Watchlist user interface); all are checked.
Yeah, I noticed this last Thursday, but I sometimes I tire of reporting bugs that never get fixed and have to go lie on my fainting couch for a while. Both the arrow direction and the uneven vertical alignment are fixed today. Thanks for reporting it, Trappist. Happy Thursday! – Jonesey95 (talk) 02:06, 8 December 2023 (UTC)
When the arrows were fixed for all of you above who enabled "Watchlist → Use non-JavaScript interface" in preferences, the same change also broke them for anyone who has it disabled. They should work correctly for everyone now. Matma Rextalk15:25, 8 December 2023 (UTC)
Not primordial data corruption - if you type "*" as your edit summary today (as I did in this edit, and my previous one to the sandbox), it appears as no summary. * Pppery *it has begun...05:11, 9 December 2023 (UTC)
Interestingly, the * edit summary appears when I view the diff on mobile web, but not when I look in Special:Contribs (or when I switch to desktop mode). Best, user:A smart kittenmeow07:10, 9 December 2023 (UTC)
As noted above in #Toolforge Grid Engine shutdown, some old infrastructure which supports a lot of tools is getting shut down soon. Mostly this is of interest to people who maintain those tools, but it affects users too. Please take a look at the list of tools still running on Grid Engine. If you see any tools you depend on, this might be a good time to start bugging whoever maintains that tool to make sure it's going to get migrated to the new infrastructure called Kubernetes.
If you're interested in the technical details, you can read more on wikitech. For those of you who aren't into the technical details, all you really need to know is that if the people who maintain the tool you use aren't working on migrating it, you should be working on finding a different tool to use before the one you're using now goes away. The most recent report says there's 445 tools yet to be migrated. With 6 migrations in the previous week and about 10 weeks to go before the toolpocolypse, the math doesn't look good. RoySmith(talk)17:45, 7 December 2023 (UTC)
How are regular editors supposed to read that report and know whether a tool they rely on will break? It might be helpful for the Grid Engine administrators to schedule a planned multi-day or one-week outage between now and the drop-dead date, so that when tools break during that planned outage, we (1) know what still needs to be migrated and (2) still have access to the tools for a few weeks before they go away. – Jonesey95 (talk) 19:10, 7 December 2023 (UTC)
It looks like a serious step backwards in terms of tool availability. A wikiproject I work on has already been seriously hit by unmaintained tools stopping working due to internal database changes, and we'll be losing more soon when the useful columns are removed from the pagelinks table. Certes (talk) 19:19, 7 December 2023 (UTC)
Unfortunately the wiki sites grow a lot and as such, things have to change every now and then, to keep everything working and maintainable. The tools are something with a lot of access to the internals of the software stack and are thus pretty vulnerable to breakage of these specific mentioned changes, especially if they are effectively orphaned tools. But that's not really what we have here, this is more a case of how tools are run, which for most tools shouldn't be too difficult to fix (the biggest problem is likely with systems that were already shaky or that use a lot of external dependencies). If everyone pitches in, it's probably not going to be as bad as it looks right now. For instance, i've done a casual check of about 15 tools on that list, and about half aren't even working to begin with, so... —TheDJ (talk • contribs) 20:05, 7 December 2023 (UTC)
It's the 15 tools that we really rely on that I worry about, and I have no idea how to pick them out of the giant list. – Jonesey95 (talk) 20:19, 7 December 2023 (UTC)
You really rely on 15 tools (!). I started trying to compile a list of tools I recognized that would be affected, and gave up at around the letter "G" after realizing there would be too many to properly list. And the apocalypse is in one week (when "Any maintainer who has not responded on phabricator will have tools shutdown and crontabs commented out"), not 10 weeks. And it will be bad. — Preceding unsigned comment added by Pppery (talk • contribs) 20:53, 7 December 2023 (UTC)
If I'm reading it right, then the standouts to me in a quick scan are AfDStats, Yapperbot, Musikbot, and wugbot. Those would all go down if their maintainers don't do something in the next few months? Firefangledfeathers (talk / contribs) 05:31, 10 December 2023 (UTC)
Could someone please provide the user script code to hide my admin buttons (blocking users, protecting pages), except those related to files? Thank you. --Leyo09:22, 11 December 2023 (UTC)
Here's a quick script that will hide block, protect, and unprotect buttons. Add to User:Leyo/common.css, not User:Leyo/common.js. May need to do a hard refresh on each page for awhile (Ctrl-F5) after installing, since CSS files are usually cached.
I can do more work on this if you let me know exactly what you need. Is your end goal to hide every single admin button (delete, undelete, change usergroups, etc.) except in the file namespace? –Novem Linguae (talk) 09:49, 11 December 2023 (UTC)
Thank you, Novem Linguae. When inserting this line to my .css (in preview mode), I get the following warning: Error: Expected RPAREN at line 2, col 38.
It's mainly about the buttons where there's a danger that I might forget my promise not to use them in relation to users, e.g. in case of vandalism, because I'm still using these buttons in two other projects. --Leyo10:51, 11 December 2023 (UTC)
I've now saved it, where I got an error message. On one PC that I regularly use, the FF version is 115.4.0esr, i.e. too old. Is there an option without the has() function? --Leyo13:24, 11 December 2023 (UTC)
Just change span:has(> .mw-usertoollinks-block) to .mw-usertoollinks-block, you get extra brackets but it's not a big deal. Nardog (talk) 13:37, 11 December 2023 (UTC)
Thank you, it works now. It would be great, if the block button in Special:Contributions as well as the deletion button for pages (i.e. except files) could be hidden, too. --Leyo13:51, 11 December 2023 (UTC)
To do the file thing, we need to switch to JavaScript. So go ahead and delete the code from your common.css file, then add this to common.js:
File pages have body.ns-6 and file talk has body.ns-7, which should be enough to do this without resorting to javascript (and the attendant FOUC). —Cryptic14:39, 11 December 2023 (UTC)
I've just implemented Novem Linguae's JS code, but haven't yet deleted the CSS code. You are the experts. I'll just implement what you recommend. --Leyo14:56, 11 December 2023 (UTC)
Thanks, the first-mentioned button is gone now. For the latter, these is actually no danger that I could forget about my promise. I don't think that I have ever used this button. --Leyo22:04, 11 December 2023 (UTC)
Sidebar is hiding when a Wikipedia page was last edited and links below it
The Wikipedia Sidebar is now hiding when a Wikipedia page was last edited and the writing and links below it when you scroll down to the bottom of a page with many sidebar links. For example, the "Privacy policy", "About Wikipedia" and "Disclaimers" are also now hidden from view. Someone has recently edited the page layout on Wikipedia for the worse as this was not the case 5 days ago. 92.24.237.212 (talk) 05:36, 3 December 2023 (UTC)
I can reproduce this with Vector-2022 in Firefox, Edge and Chrome, logged in or out, with or without safemode. Pick a page with a long TOC, e.g. this page. Make sure the TOC is shown in the sidebar and not hidden. The TOC extends to the bottom of the window and covers the left part of the footer. "23, at 06:36." is the only visible part of "This page was last edited on 3 December 2023, at 06:36." The next lines become visible at "Attribution-ShareAlike", "a non-profit organization" and "Contact Wikipedia". Either the sidebar should stop before the footer or the whole footer should move to the right of the sidebar like the content of the wiki page. PrimeHunter (talk) 11:28, 3 December 2023 (UTC)
For articles/discussion pages with very long tables of contents, such as this one, in vector2022 on Firefox on a Mac, I am getting the table of contents truncated after a certain length. For instance, on this page, as I type this, the table of contents that I see (when I scroll down to the bottom of either the ToC or the actual page) shows the top few pixels of #Rotten tomatoes prose template (or rather "Rotten tomatoes prose" as the "template" wraps to the next line that I cannot see), and the next heading, #The center class, is completely omitted. I suspect some length calculation gone awry, as it is always near the end of what should be the full table of contents regardless of the number of entries. Is anyone else seeing this? Any ideas what to do to get it fixed? —David Eppstein (talk) 22:21, 3 December 2023 (UTC)
I doubt it. It has nothing to do with how much of the window the TOC rectangle covers or what might be hidden under that rectangle. It is about the objects within the TOC rectangle and how much of them can be seen by scrolling within the rectangle. To put it another way: once it needs to scroll, the TOC rectangle is always the same size (the height of the window). The *contents* of the TOC rectangle (the stuff you can scroll) may be a much larger size. This bug involves some miscalculation of the size of the contents of the TOC rectangle, causing the TOC entries beyond the miscalculated size to be unviewable. —David Eppstein (talk) 01:23, 4 December 2023 (UTC)
Right, but the reason you can't see the last item is because it's overflowing the bounds it's supposed to and so under the typically-present bar at the bottom of the page that goes with the expander square on the right hand side. I am pretty sure it's the same issue. Izno (talk) 01:28, 4 December 2023 (UTC)
Maybe. On closer view, the problem is that the TOC rectangle is extending past some bottom margin on the screen. You can tell that it's the TOC rectangle being cut off rather than the contents of the TOC rectangle being cut off because the rounded bottom of the scroll bar is also cut off. Only the left sidebar has this margin; the main window and right sidebar contents extend past it. —David Eppstein (talk) 01:32, 4 December 2023 (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 Wikimedia Commons, there are some minor user-interface improvements for the "choosing own vs not own work" step in the UploadWizard. This is part of the Structured Content team's project of improving UploadWizard on Commons. [39][40]
Problems
There was a problem showing the Newcomer homepage feature with the "impact module" and their page-view graphs, for a few days in early December. This has now been fixed. [41][42]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 12 December. It will be on non-Wikipedia wikis and some Wikipedias from 13 December. It will be on all wikis from 14 December (calendar). [43][44]
Future changes
The 2023 Developer Satisfaction Survey is seeking the opinions of the Wikimedia developer community. Please take the survey if you have any role in developing software for the Wikimedia ecosystem. The survey is open until 5 January 2024, and has an associated privacy statement.
$2 is already the file name with that part stripped. My suggestion is simply to turn it into a wikilink, not add extra text. Nardog (talk) 07:25, 6 December 2023 (UTC)
Or you just ctrl click the video and the browser opens the file page in a new tab (just like any image). Or right click and choose "open in new window/tab". And if there is no video found, it should not display a video in the page and you would get "There is no video associated with the current subtitle page." next to the subtitle content. —TheDJ (talk • contribs) 09:43, 5 December 2023 (UTC)
Does it also display "There is no video associated with the current subtitle page." when the video exists as a redirect, or is on Commons? Because in these cases measures other than deletion would be needed. Jo-Jo Eumerus (talk) 09:57, 5 December 2023 (UTC)
For TT pages that refer to Commons repos, It hides their contents and says: "You can view the description page for this file on Wikimedia Commons (production)" (example). For redirects it says "There is no video associated with the current subtitle page" (example). —TheDJ (talk • contribs) 13:42, 5 December 2023 (UTC)
Btw. @Jo-Jo Eumerus there is Special:OrphanedTimedText, which I think does exactly the same as that db report. I fixed it last week to exclude redirects and this should get deployed somewhere this week. Hope that helps. I'm also working on the other issue, but it has a lot of edge cases that I have to verify. —TheDJ (talk • contribs) 12:32, 12 December 2023 (UTC)
You could grab the TimedText page title and extract the expected File: by Template:str rep pattern.
If desired, an {{#ifexist:}} for the related File: could issue a related warning for everybody, otherwise provide a link to the expected File: (perhaps always as requested by initiating this section).
Yeah, that last point is the problem - I need a code that can strip the text after the second-to-last dot. Str rep does not do that, it needs to be told "pt" or whatever. Jo-Jo Eumerus (talk) 15:49, 5 December 2023 (UTC)
For example, "{{#invoke:String|replace|TimedText:Dua Lipa Blow Your Mind (Mwah) sample.ogg.en.srt|^TimedText:(.*)%.%w+%.%w+$|File:%1|1|plain=false}}" produces "File:Dua Lipa Blow Your Mind (Mwah) sample.ogg". Certes (talk) 16:22, 5 December 2023 (UTC)
OK, that works better. To clarify, I am thinking to put {{#ifexist:{{#invoke:String|replace|{{FULLPAGENAME}}|^TimedText:(.*)%.%w+%.%w+$|File:%1|1|plain=false}}|This timed text is attached to a redirect.|There is no video associated with the current subtitle page.}} on MediaWiki:Timedmedia-subtitle-no-video so that we can tell the difference. Or would a software patch to create a new message for TimedText attached to redirects make more sense? Jo-Jo Eumerus (talk) 16:46, 5 December 2023 (UTC)
Yes, in the long run the TimedText:*.srt pages should be equipped with something like MediaWiki:timedmedia-timedtext-intro or whatever (empty by default).
I did suggest MediaWiki:editnotice-710 as first aid for now, since no other system message is transcluded today.
A phabricator proposal might work in some weeks or years.
If the phabricator proposal is simple enough that it doesn't take me hours to verify that we won't break anything, wrt to performance or 3rd party users, i can easily fix them. I'm essentially the only person maintaining TMH next to brion so. —TheDJ (talk • contribs) 08:33, 6 December 2023 (UTC)
That's weird - when I make a test redirect on File:Test.mp3 and then create TimedText:Test.mp3.en.srt, the latter simply attaches the TimedText to the redirect target. I don't think it did the latter yesterday. Once I break the redirect however in this edit it displays "There is no video associated with the current subtitle page." Now I dunno anymore if splitting the MW messages is warranted or trivial (also, would that be one or two Phab tasks?) Jo-Jo Eumerus (talk) 08:51, 6 December 2023 (UTC)
Fixing Citoid created references for Youtube
Hi all
Youtube is the second most popular website in the world, with a huge number of reliable news sources using it as a platform for sharing video. Unfortunately Citoid doesn't work properly for creating refs for Youtube currently. This leads to poor quality labelling of references being made, and while there are some templates to specifically cite video content, they aren't user friendly at all. I started a Phabricator ticket in 2021 to try and address this issue, however its not been worked on. Can I request anyone interested in this:
Subscribe to the Phabricator task so its made clear this is an issue many people would like to be fixed.
Try citing Youtube videos in articles and give feedback in the Phab task if you notice anything I haven't mentioned.
Also just to say I've seen a couple of people say "Youtube is an unreliable source and shouldn't be used", I think this is a missunderstanding of what the platform is, its the channels that should be assessed for reliability rather than the publishing platform as a whole. E.g the BBC News channel is reliable (its listed under Perennialy reliable sources on en.wiki), where as My Toy Reviews or DailyWire or whatever are obviously not reliable.
The annoying thing is that this probably isn't actually that much work to fix - it just requires specific technical skills to work on, and different specific technical skills than most other tasks so even techies like myself aren't willing to step up to the plate. * Pppery *it has begun...20:38, 12 December 2023 (UTC)
essentially, a user record for a remote user, that was imported locally. As you can see, this page was imported in 2020, but the edit is from somewhere outside of en.wp in 2019 —TheDJ (talk • contribs) 14:38, 12 December 2023 (UTC)
In this case Meta, and the import was accidental. Because the edit was imported, MusikAnimal didn't actually protect the template despite appearing to do so. I'm inclined to abuse Special:MergeHistory to delete that edit (and its brethren elsewhere) from the history entirely. This is generally much more of a problem on other wikis where people tend to import stuff without thinking it through than on the English Wikipedia, and before I did a history cleanup in 2021 mw:Template:Documentation had history imported from (at least) 4 different wikis making it impossible to follow. * Pppery *it has begun...21:10, 12 December 2023 (UTC)
Yes, I was exporting/importing revisions between wikis, and simply got them them mixed up. It was a huge boo-boo on my part and I'm lucky nothing bad came of it (Module:Category handler has 4.5+ million transclusions!). I actually banned myself from using Special:Import in the future :)
If someone wants to remove that dummy revision, go for it, but I'm completely fine with it staying if not solely for the accountability of my mistake. Also, each change to that module will fire off millions of jobs, so maybe it isn't really worth it. WP:DWAP applies here but we're only talking about aesthetics of the revision history. If we want to avoid confusion with regards to the misleading edit summary, we can just hide that? — MusikAnimaltalk23:12, 12 December 2023 (UTC)
category for companies with no logo
i want to sort a list of companies that have no logo, so i can search and add logos where appropriate. i asked about this at teahouse and it was suggested that i ask here at village pump. my first ever village pump question, so forgive me if i have no idea where to ask or how to format my request. Iljhgtn (talk) 00:15, 13 December 2023 (UTC)
(New) Feature on Kartographer: Adding geopoints via QID
Since September 2022, it is possible to create geopoints using a QID. Many wiki contributors have asked for this feature, but it is not being used much. Therefore, we would like to remind you about it. More information can be found on the project page. If you have any comments, please let us know on the talk page. – Best regards, the team of Technical Wishes at Wikimedia Deutschland. Thereza Mengs (WMDE)12:31, 13 December 2023 (UTC)
How to get top 100 categories available in most Wikipedias, but not created in a specific Wikipedia?
Hi, I wonder if we can get top 100 categories available on most Wikipedias, but still not created in a specific Wikipedia. Use Quarry (which I want this way if you can), API, SPARQL or any other way you prefer. If you can, add an optional condition to exclude those have __HIDDENCAT__ property, tracking and maintenance categories. Thanks! ⇒ AramTalk05:55, 11 December 2023 (UTC)
I am not sure where else to ask about or report this, so sorry if this is the wrong place for this kind of post. I have encountered a strange issue with the main page on Wikipedia, the page that you would first encounter when simply typing in [wikipedia.org] into the search bar of a browser, the page where you can select your language. Basically when I click on the [Read Wikipedia in your language] button, nothing happens, so I am unable to view the page with the list of languages. Also, when I'm typing in the search bar, I can't see any results until I strike the Enter key. Oikoopa (talk) 23:24, 12 December 2023 (UTC)
I am on Firefox 120.0.1, although the issue happens across multiple browsers across multiple devices, even when going into a Private Browsing tab or similar. I did indeed get a red error message when looking at the logs (image). I don't know how it could be a browser-specific issue, this bug happens on Firefox, Chrome, and a few Chromium-based browsers, and like I said, across multiple devices. And yet, I haven't seen anybody else report this issue (although I haven't looked particularly hard).
It's also worth noting that I have experienced this issue before, but then it just kind of disappeared. Now it's back again. Quite strange. Oikoopa (talk) 00:25, 13 December 2023 (UTC)
Further to the above, is it possible to transclude a previous version of a template, in this case {{ITN}}? I suspect the answer is no, in which case I will have to copy-paste the relevant details from the template's history, but I'm asking on the offchance there's some technical way I'm not aware of. Thanks. — Voice of Clam (talk)16:35, 13 December 2023 (UTC)
I'm trying to build the main page archives for missing dates, as the bot that produces them automatically wasn't working for a couple of weeks. I've built a framework to construct the details for a given date, but I'm having problems with WP:DYK and WP:ITN. For DYK, I could transclude the relevant section of Wikipedia:Recent additions using (for example) {{Excerpt|Wikipedia:Recent additions/2023/November|21 November 2023}}, however this also includes a hatnote and extra time stamp that wasn't on the original template. The above code gives
Adult Balkan terrapin*00:00, 21 November 2023 (UTC)
... that hatchling Balkan terrapins are only 3 to 4 centimetres (1.2 to 1.6 in) in length, while adults (example pictured) can grow as long as 25 cm (9.8 in)?
... that Rush Munro's, New Zealand's oldest ice creamery, has used the same recipes since 1926?
Typography and text improvements now available as a beta feature
The new menu in the un-pinned version, next to the user name
Hi everyone. The Web team from the Wikimedia Foundation has launched a new beta feature to improve readability and accessibility. The feature is only available to logged-in editors using the Vector 2022 skin. To try it out, go to the "beta" option in the user menu and select "Accessibility for Reading (Vector 2022)". You may also enable it on all wikis using the global preferences. To learn more about the Accessibility for Reading project, check out the project page.
In the new menu, there are three text settings – small, standard, and large. Small is the current default. Large is for users who need additional increase in size. The standard setting may later become the new default. This was recommended by both the literature research and prototype testing. To this menu, we have also added the page width setting. Before, it was available in the bottom corner of the screen, and it's easier to find now. The menu is pin-able in a similar way to the Tools and Main menus, both placed in the side columns of the desktop interface. When it's not pinned, it's displayed next to the user name.
We welcome you all to try the new menu. We will collect your ideas and comments to make this feature better. Please, if you have the time, check it out and give us your thoughts! Thanks. SGrabarczuk (WMF) (talk) 21:43, 14 December 2023 (UTC)
Gigantic section titles on talk pages?
What's up with the gigantic section titles on talk pages suddenly? Different font than other level 2 headings as well (e.g. on articles). A Thursday "improvement"? Fram (talk) 20:51, 14 December 2023 (UTC)
Thanks, but I don't have DiscussionTools enabled (Beta Features), and in the "editing/discussion pages" options, I don't see "show discussion activity" (neither do I see it in "preferences/appearance", so no idea where this could be found). Oh well, I'll try logging out and so on to see what that gives, but at least this is only a minor annoyance: when using my smartphone (mobile view), since yesterday evening everything is extremely small (as in, unreadable), and enlarging my screen means that lots of things are no longer visible. I haven't changed my preferences or installed gadgets or so recently, so no idea why I have these two issues which no one else seems to have; I guess some strange combination of preferences which no one else uses? Fram (talk) 08:31, 15 December 2023 (UTC)
If I log out I automatically get new Vector, the issue is in old vector, so that doesn't help me. I'll try safemode now. Fram (talk) 08:33, 15 December 2023 (UTC)
Found it: the gadget "Vector classic typography (use only sans-serif in Vector skin)" was enabled, and this one suddenly causes these issues (for me at least). Disabling this solved it, I now get uglier but normal-sized section headings. Fram (talk) 08:37, 15 December 2023 (UTC)
I don't know if this is the right place to ask this question but, an IP address pointed out at the Teahouse that the OpenStreetMap on the article Istanbul, is not showing the capital, and the location for some reason is at the Gulf of Guinea. [45] Could this issue be fixed? 🛧Midori No Sora♪🛪 (☁=☁=✈) 12:10, 14 December 2023 (UTC)
It works for me: I see a map of Istanbul. 0°N 0°E lies in the Gulf of Guinea, and is a popular destination for broken map links. Certes (talk) 12:40, 14 December 2023 (UTC)
There are three OpenStreetMap in Istanbul. Two on a small globe at coordinates 41°00′49″N 28°57′18″E are working for me. One is hidden under "OpenStreetMap" in the infobox and made with {{Infobox mapframe|frame-width=250|zoom=8}}. That's broken for me. PrimeHunter (talk) 16:07, 14 December 2023 (UTC)
I tried {{Infobox mapframe}} on Special:ExpandTemplates with different cities in the "Context title" field. ExpandTemplates doesn't show the maps at first but I copy-pasted the below to a normal wiki page here.
<mapframe zoom="10" align="center" frameless="1" height="200" width="270">[
{"properties":{"title":"Odense","fill-opacity":0.5,"stroke":"#FF0000","stroke-width":3,"fill":"#606060"},"type":"ExternalData","service":"geoshape","ids":"Q25331"},
{"properties":{"stroke-width":5,"stroke":"#FF0000","title":"Odense"},"type":"ExternalData","service":"geoline","ids":"Q25331"},
{"type":"Feature","geometry":{"coordinates":[10.383333333333,55.4],"type":"Point"},"properties":{"title":"Odense","marker-color":"#5E74F3"}}
]</mapframe>[[Category:Infobox mapframe without OSM relation ID on Wikidata]]
London with Wikidata London (Q84) gives no coordinates and still works but shows a different type of map with a line around the city.:
I think the difference between Istanbul and London is caused by how mw:Help:Extension:Kartographer reads the "OpenStreetMap relation ID" linked at the Wikidata item for those cities (Istanbul and London). That's beyond my knowledge. Odense has no such Wikidata field so the coordinates are just pulled from the Wikidata item. That works. PrimeHunter (talk) 17:14, 14 December 2023 (UTC)
I can get the map by supplying the coordinates with |coord= or by supplying the wikidata id and setting |point=on or just adding |point=on. All show a point symbol for the coordinates. As a test, I gave it the wikidata ID for Cairo (Q85) it displays an OSM shape set by item OpenStreetMap relation ID (P402) and if I also add |point=on it displays both shape and point. It looks like there is an issue with OSM relation for Istanbul that makes the map fail, unless you tell it to also show a point. I think there has been an ongoing problem setting new OSM relations. — Jts1882 | talk17:26, 14 December 2023 (UTC)
We don't really know why this limitation is the way it is btw. It was setup when there was still a Maps team, but none of the current devs understand exactly how that index should be modified and why it is the way it is. If anyone wants to read through a lot of code and figure it out, please do :) —TheDJ (talk • contribs) 11:07, 15 December 2023 (UTC)
^The date given is the date used by Historic England as significant for the initial building or that of an important part in the structure's description.
^The "List Entry Number" is a unique number assigned to each listed building and scheduled monument by Historic England.
If I click on the Lat/Long coordinates, the GeoHack page gives me an OS ref of TQ2657774519, which matches the OS ref in the entry. If, however, I click on the OS ref in the entry, it gives me a GeoHack page with OS ref TQ2646474573 (51° 27′ 19.83″ N, 0° 10′ 46.59″ W). Looking at the template source, this seems to be a problem with #invoke:Ordnance Survey coordinates. What is this, and where should I report it?
Noticeboards, which are project-space pages. and article talk pages and user talk pages, have a button at the top right of each section marked "Subscribe". Clicking on this button puts the user on a list to receive emails notifying them of additions to the section, which is a watchlist-like feature. My problem has to do with scrolling these pages on a mobile device in desktop mode. It is too easy to hit the "subscribe" button by accident when only trying to scroll the page. There is a message saying that you have been subscribed, but then unsubscribing requires scrolling back to where one was scrolling to hit "Unsubscribe". Could a "Confirm" prompt be added to verify that the editor really wants to Subscribe, rather than taking an action that will clutter the inbox based on a single click? I am assuming that the developers didn't consider the use of desktop view on a mobile device, because the developers think that I should use mobile view when I am using a mobile device. However, many experienced editors are in agreement with me that desktop view is better than mobile view. (I haven't even tried using a mobile application, because I don't have a mobile application installed on my Android.) On a real computer, the screen is wider, and there is usually a real scroll bar, so that I don't have this problem with a real computer, but I would like to be able to scroll on an Android without accidentally signing up for inbox clutter that I have to turn off. Have I missed something, or is this a case where the developers have missed a misfeature because they have added a large misfeature of Mobile View? Does anyone have a different thought?
Robert McClenon (talk) 17:49, 14 December 2023 (UTC)
It gives a confirmation prompt but then subscribes no matter what you pick. There is probably an easy fix like something instead of e.preventDefault() but my JavaScript is very limited. PrimeHunter (talk) 21:52, 14 December 2023 (UTC)
I think the subscription feature uses a JavaScript event listener to process the click, so preventDefault() won't stop it from triggering. stopImmediatePropagation() is supposed to stop other listeners for the same element from being notified as well as stopping propagation to listeners for ancestor elements, but I don't think there's any way to guarantee that this listener would get triggered before the one for the subscription feature. isaacl (talk) 22:17, 14 December 2023 (UTC)
If it's worth the time and effort, could ask for a mw.hook().fire() to be added to the top of the subscribe code via Phabricator. Then you could hook into it with the mw.hook().add() API. –Novem Linguae (talk) 23:04, 14 December 2023 (UTC)
It can be done by removing the event listener (after first saving a reference to it). Then you would add back a new event listener which conditionally invokes the first one. – SD0001 (talk) 06:58, 15 December 2023 (UTC)
If I'm understanding the documentation for removeEventListener() correctly, this can only be done by knowing the event listener function and the value of the capture option used when registering it. Thus even if there is a named listener function handling the subscription, trying to refer to it would be highly coupled to the internal implementation. An officially supported hook as suggested by Novem Linguae would be a better approach. isaacl (talk) 18:10, 15 December 2023 (UTC)
Hooks in MediaWiki are asynchronous. They can only run an additional function in response to an event, not block another function which was supposed to run.
Removing listener can be done via jQuery: $('...').off('click'). But it appears there's no supported way to refer to an existing one – we would probably have to implement the subscribe logic on our own using the API. – SD0001 (talk) 13:12, 16 December 2023 (UTC)
Based on the documentation, the jQuery off() method will remove listeners added via the on() method; it's not clear to me that it would remove handlers added via the native JavaScript API. However as you noted, and much like removeEventListener(), it doesn't return a reference to any removed handlers.
Thanks for the info on hooks; I suspected this but wasn't sure after a quick reading of mw:Manual:Hooks. I still feel that an officially supported mechanism by the subscription feature is the best way to avoid coupling to the internal implementation. isaacl (talk) 17:13, 16 December 2023 (UTC)
Improvements to styling of Vector 2022
Before (?vectorzebradesign=0)After (?vectorzebradesign=1)
Are these screen shots intended to show the exact same page, in the same configuration, before and after the changes? If so, it looks like the column dedicated to content, the thing everyone comes to Wikipedia for, has gotten even narrower, when it was already much too narrow to begin with. I hope this is not the case.
(Edited to add this rant: I just confirmed this in another browser, logged out. At John Dalton, with the (fewer visible vertical elements) TOC showing, and the Tools sidebar displayed, the content column shrank from 670 pixels wide to 618 pixels wide, just 50% of my 1228-pixel-wide browser window. With custom CSS that makes Vector 2022 more usable, I am able to get the content column to 856 pixels wide, 70% of the available space. And this isn't even addressing the unused vertical space at the top of the window that means even more vertical scrolling for our readers. I've got that ancient phab link around here somewhere, maybe T325219.) – Jonesey95 (talk) 21:47, 11 December 2023 (UTC)
Same page on en.WP with Zebra and custom CSS
Welp, that was fun. With just a couple of edits to my common.css, I was able to get Zebra Vector 2022 (Vector 2023? Vector 2022 3.0? What are we calling this thing?) working pretty nicely, now with 897 pixels of content width (73% of horizontal space) and lots of visible text in the left and right sidebars. Other editors are welcome to copy some or all of my CSS file; caveat copier, though: I am not a CSS expert by a long shot, and my spacing preferences may leave too little white space for you. – Jonesey95 (talk) 02:59, 12 December 2023 (UTC)
Vielen Dank!I've been using an edited version of your great CSS tweaks for 5 months! However, I have edited them to be a bit more spacey and now there's a strange space to the right of the tools menu that only shows if there's rendered wikitext present. (e.g. doesn't appear in history or edit, appears in normal page view and edit after clicking preview). I can't find it in the DOM either. Any idea what may be causing it? Aaron Liu (talk) 03:18, 12 December 2023 (UTC)
As you saw, my CSS needed many adjustments when Zebra went into effect. I fiddled with it until it worked for me. I copied your CSS to my common.css but was unable to reproduce the problem. A full-window screen shot might help. What page were you on? – Jonesey95 (talk) 05:03, 12 December 2023 (UTC)
I've figured it out. I needed grid-template-columns:minmax(0,1fr)min-content; in .mw-body . No idea why it doesn't happen for you. Also, your CSS contains a column-gap: 6px that you soon override with column-gap: 16px . Why is that so? Aaron Liu (talk) 19:51, 16 December 2023 (UTC)
Reduces the content width, and seems to add new whitespace on the top and left of the ToC which will push some of that out of some screens. CMD (talk) 03:28, 12 December 2023 (UTC)
Yes, reducing the content width was apparently a deliberate design choice: Due to the change above, the content width gets slightly narrower when menus are pinned in both columns. (To be clear: I'm just reporting, not supporting, someone else's words.) – Jonesey95 (talk) 05:22, 12 December 2023 (UTC)
For me the pinning only affects the left column, unpinning the right column (tools) does not seem to change content width. CMD (talk) 07:45, 12 December 2023 (UTC)
While I like most of these changes, I hate the new dropdown and the new "show"/"hide" button. They are inconsistent with other occurrences and the new dropdowns have way too much padding-top and bottom and way not enough separation. I could maybe get over the show and hide, but the new dropdowns just look alien to the rest of the wiki (search boxes, regular boxes...) for no good reason and just look worse. I also think the new width is too narrow. Aaron Liu (talk) 01:28, 12 December 2023 (UTC)
For me, all dropdowns appear to have a 16px whitespace block at the bottom which looks awful especially under the logout button. And my twinkle menu now has some box within a box situation going on which also looks bad. Although I do like the changes somewhat. Terasail[✉️]03:42, 12 December 2023 (UTC)
See these CSS changes. The ...::after one at the end gets rid of the mysterious blank box. If the Wikimedia developers someday decide to populate that container with something fun, I won't see it, but for now, the space will be gone. Who knows if it's a bug or deliberate; the developers love putting unnecessary and excessive whitespace everywhere in Vector 2022, so that extra space at the bottom of each menu may be a design choice consistent with the rest of this skin's default aesthetic. – Jonesey95 (talk) 05:18, 12 December 2023 (UTC)
The contents menu needs to be of the same width and alignment with other menus. Currently, it’s shorter and right-aligned because of the arrows. We already have padding on the left and right, why can’t the arrows use some of it? Aaron Liu (talk) 12:09, 12 December 2023 (UTC)
I just noticed this change. It's getting there. At least the "popunder" behavior of the Page menu when opened after the User menu is fixed. I would still like to see the V icon change to ^ when the menu is displayed (either that, or click-away implemented like classic Windows). Sorry, I forgot which phab this was tracked under. David Brooks (talk) 18:47, 16 December 2023 (UTC)
Another qualm I have with the new user dropdown is there is no separation between Log Out (dangerous!) and everything else now. Aaron Liu (talk) 19:06, 16 December 2023 (UTC)
Watchlisting my own userpage (and therefore my own user-talk) so all of the automatic archives would be watchlisted automatically (catching an annoying type of vandalism).
Temporarily watchlisting a template (and therefore its template-talk) as several editors propose and hack on sandboxes or other development work.
Like the previous, but for an article (and therefore its talk), as several editors propose and work on major overhauls.
Watchlisting a template that has lots of subtemplates, and therefore automatically watching the whole bowl spaghetti and talkpages.
In all these cases, I would not have to specify the exact individual subpages or keep updating as new ones are created. DMacks (talk) 10:11, 17 December 2023 (UTC)
Making a script to watchlist a page and all its subpages is fairly easy (e.g. all of these: Special:PrefixIndex/User_talk:DMacks/). However, wanting to watchlist currently non-existent pages is not, and would require software changes. If the former would be of use, there are a few places to request that someone do that (could maybe even become a Twinkle feature). — xaosfluxTalk11:52, 17 December 2023 (UTC)
Small capital Œ indistinguishable from œ
The symbol ⟨ɶ⟩, the open front rounded vowel, should display as a small capital Œ, but instead appears identical to ⟨œ⟩ on Apple systems; that symbol represents the open-mid front rounded vowel, which is distinct.
That must be a limitation in the font used on the Apple system. They look different in Safari on my iPhone. We just send the character code. The rest is up to the user's software. It doesn't seem important so I advice against trying to send an image or force a specific font which might work better on some systems. PrimeHunter (talk) 22:04, 14 December 2023 (UTC)
My previous posts were about the mobile site on an iPhone. After seeing Nardog's link to Help:IPA#Rendering issues I tried the desktop site on iPhone and it shows the reported problem where "The symbol ⟨ɶ⟩" looks like the first image above. PrimeHunter (talk) 00:41, 15 December 2023 (UTC)
However, I didn't get Andika working on another device macOS Sonoma with Safari, even though I have it locally installed on this device. –LaundryPizza03 (dc̄) 23:19, 17 December 2023 (UTC)
Revert to Arial font
Hello, is there any way to revert to Arial font on talk and Wikipedia namespace pages? I have been doing this for years but now Wikipedia is forcing the Georgia font on talk pages. I think it's too big and I prefer Arial. Thanks, Nearly but not perfect (talk) 18:06, 15 December 2023 (UTC)
Other than headings, the Vector (legacy) skin doesn't set a specific font family except for a generic sans-serif. The font that actually gets displayed is controlled by your browser settings. So, second question: what is your browser? --Redrose64 🌹 (talk) 23:28, 17 December 2023 (UTC)
Template spacing help
This should be a pretty easy fix, I just don't know how to do this myself 😅 There have been some pretty significant formatting changes to Template:C21 year in topic recently per discussion, but now the title seems a little but smushed to the rest of the sidebar. Does anyone know how to space it out just a little bit so it looks better? Thanks! Johnson52423:09, 17 December 2023 (UTC)
@Izno: So... are you saying this is the proper layout as it is now? That's OK if it is, I just want to make sure I understand you correctly. Johnson52423:46, 17 December 2023 (UTC)
@Izno: Thank you for pointing that error out! It is fixed now, but do you have any ideas on how to fix my original spacing issue? Cheers! Johnson52423:54, 17 December 2023 (UTC)
Ah. Different issue entirely. What do you mean by smushed? That you think there should be more top and bottom spacing for the actual sidebar title? Izno (talk) 23:57, 17 December 2023 (UTC)
Works for me@Magnolia677: are you only seeing this at a certain resolution / screen size? Can you tell which pixel size you are looking at for possible thumbnail regeneration? — xaosfluxTalk12:48, 17 December 2023 (UTC)
@Xaosflux: I'm on a desktop computer, and the screen size and resolution do not remove the lines. However, I only see the lines when the image is loaded into a biographical infobox (I've tried "infobox person", "infobox boxer", and "infobox astronaut"). When I load the image into "infobox settlement", the lines go away. Thank you. Magnolia677 (talk) 14:48, 17 December 2023 (UTC)
@Xaosflux: The top image has three black horizontal lines (one cuts trough the bottom of his nose). The bottom image has no lines. There must be a bug in the infobox. Thanks. Magnolia677 (talk) 16:41, 17 December 2023 (UTC)
@Magnolia677: Do the lines extend to the border of the infobox? If not then it's more likely to be an issue with how your browser renders the MediaWiki rescaling at a certain size. {{Infobox settlement}} has default 250px. {{Infobox person}} depends on "Thumbnail size" at Special:Preferences#mw-prefsection-rendering. As a further complication, MediaWiki usually makes html with srcset with two sizes so your browser can choose depending on circumstances. What is your "Thumbnail size" setting? Do you see the lines in any of these pure jpg files: Original, 375px, 300px, 250px. They all work for me in Firefox. What is your browser? PrimeHunter (talk) 17:33, 17 December 2023 (UTC)
I have reproduced the problem. The default "Thumbnail size" at Special:Preferences#mw-prefsection-rendering is 220px. The image works at 220px but for users with the default 220px, MediaWiki makes a srcset which also has a 1.5x version at 330px. This version can be displayed by a browser in some circumstances and it has the described horizontal lines. It appears to be a problem in MediaWikis rescaling software. PrimeHunter (talk) 18:10, 17 December 2023 (UTC)
The current 2023 version by Xaosflux doesn't have the 330px issue when it's rescaled. You could restore or reupload the 2016 version and display it at 330px to see whether MediaWiki still has the problem. I have an offline copy of the 330px version with three lines. PrimeHunter (talk) 04:26, 18 December 2023 (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.
Tech News
The next issue of Tech News will be sent out on 8 January 2024 because of the holidays.
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 19 December. It will be on non-Wikipedia wikis and some Wikipedias from 20 December. It will be on all wikis from 21 December (calendar). There is no new MediaWiki version next week. [46][47]
Starting December 18, it won't be possible to activate Structured Discussions on a user's own talk page using the Beta feature. The Beta feature option remains available for users who want to deactivate Structured Discussions. This is part of Structured Discussions' deprecation work. [48]
There will be full support for redirects in the Module namespace. The "Move Page" feature will leave an appropriate redirect behind, and such redirects will be appropriately recognized by the software (e.g. hidden from Special:UnconnectedPages). There will also be support for manual redirects. [49]
Future changes
The MediaWiki JavaScript documentation is moving to a new format. During the move, you can read the old docs using version 1.41. Feedback about the new site is welcome on the project talk page.
The Wishathon is a new initiative that encourages collaboration across the Wikimedia community to develop solutions for wishes collected through the Community Wishlist Survey. The first community Wishathon will take place from 15–17 March. If you are interested in a project proposal as a user, developer, designer, or product lead, you can register for the event and read more.
Hello all. I noticed on Template:Xmas4, it has the 4 tildes for the person posting it but on the sample it only has 3 on display and when you do post it on a talkpage, it only shows the tildes. Is someone able to have a look at this and fix it so it does display the poster's name please? The C of E God Save the King! (talk)22:04, 18 December 2023 (UTC)
Template:Xmas4 has a box saying: "This template should always be substituted (i.e., use {{subst:Xmas4}})." That applies to most templates for user talk messages. Then it works. It deliberately uses 3 tildes in the middle of the text to show the username without a time stamp. PrimeHunter (talk) 23:15, 18 December 2023 (UTC)
And it's impossible to leave a signature or username for a template which is not substituted. At best it could detect it's not substituted and omit both tildes and signature but I don't think it should do that. PrimeHunter (talk) 09:48, 19 December 2023 (UTC)
On mobile view of the linked page, by default, the heading for Welcome to Wikipedia’s curated content is collapsed by default and I would like to set it so that it isn’t able to be collapsed. I am hoping to get some suggestions on how I would be able to do that. Thank you. Interstellarity (talk) 16:14, 18 December 2023 (UTC)
Don't use a level 2 heading if you have no others seems like a reasonable response. You cannot work around the collapsing behavior another way. Move the kead text to be an actual lead, remove the current level 2, and promote the remaining to level 2 headings. Izno (talk) 18:49, 19 December 2023 (UTC)
You can't see on mobile if the article is GA/FA, and there's no documentation explaining why
So obviously on desktop, you can see at the top-right corner if the article is a GA, or FA, or similar. Take England (GA), Canada (FA) and List of islands of Scotland (featured list). But when you switch to mobile view, you see none of that, so basically you have no idea if a long article is one we'd vouch for or is a bloated nightmare until you read well into the article, which most folks don't do anyway. (Donald Trump isn't a GA or FA as of writing, but it is 17.5K words long).
This template does not display in the mobile view of Wikipedia; it is desktop only. Read the documentation for an explanation.
So the problem is the explanation... isn't there in the doc. Looking to Page status indicators (MediaWiki) gives me nothing. So maybe there is a reason top icons won't display but I will never know why. And it's about a lot of elements.
Also, I would like to inquire about an alternative way to show an icon. A bit like a hybrid approach btw French Wikipedia and here. So if Wikipedia detects you're on desktop, display icon. If on mobile, display text at the top of page, a bit like here (You are reading a featured article rated in 2021). You can calculate the year based on the diff date. I hope this is possible to implement, at least for the more important ones like degree of protection and article status. Szmenderowiecki (talk) 14:02, 19 December 2023 (UTC)
importpywikibotfrompywikibotimportpagegeneratorssite=pywikibot.Site('en','wikipedia')query=r"""insource:/\| quote = You may also wish to consider using a Wizard to help you create articles\. See the \[\[Wikipedia\:Article wizard ?2\.0\|Article Wizard\]\]\.'''''/"""gen=pagegenerators.SearchPageGenerator(query,site=site)print('Querying...')print(len(list(gen)))
Throws
ERROR: Traceback (most recent call last):
File "/srv/paws/lib/python3.10/site-packages/pywikibot/data/api/_requests.py", line 682, in _http_request
response = http.request(self.site, uri=uri,
File "/srv/paws/lib/python3.10/site-packages/pywikibot/comms/http.py", line 283, in request
r = fetch(baseuri, headers=headers, **kwargs)
File "/srv/paws/lib/python3.10/site-packages/pywikibot/comms/http.py", line 457, in fetch
callback(response)
File "/srv/paws/lib/python3.10/site-packages/pywikibot/comms/http.py", line 333, in error_handling_callback
raise ServerError(response)
pywikibot.exceptions.ServerError: HTTPSConnectionPool(host='en.wikipedia.org', port=443): Read timed out. (read timeout=45)
Received wisdom from Help:Searching/Regex is that insource regex searches aren't indexed, while insource non-regex searches are. Either search for just the text with something like insource:"You may also wish to consider using a Wizard to help you create articles", or if it's critical that you also search for the punctuation, search for both. —Cryptic20:17, 18 December 2023 (UTC)
I don't know the language above, but do you need to escape the trailing apostrophe markup? Qwerfjkl, I'll be happy to generate a manual list of pages for you if that would help. – Jonesey95 (talk) 15:54, 19 December 2023 (UTC)
Jonesey95, I can generate a manual list. The problem is that the search query only returns 5000 pages at a time, so I'd have to rerun the thing a bunch of times. I'm fairly sure apostrophes don't need to be escaped, they are not special characters. — Qwerfjkltalk17:13, 19 December 2023 (UTC)
Got it. As a workaround, here are 5,000 results and 5,000 different results (with and without a space; URLs are not cooperating, but copy/paste should work): https://en.wikipedia.org/w/index.php?title=Special:Search&limit=5000&offset=0&ns0=1&ns1=1&ns2=1&ns3=1&ns4=1&ns5=1&ns6=1&ns7=1&ns8=1&ns9=1&ns10=1&ns11=1&ns12=1&ns13=1&ns14=1&ns15=1&ns100=1&ns101=1&ns118=1&ns119=1&ns710=1&ns711=1&ns828=1&ns829=1&ns2300=1&ns2301=1&ns2302=1&ns2303=1&search=insource%3A%2F\|+quote++%3D+You+may+also+wish+to+consider+using+a+Wizard+to+help+you+create+articles\.+See+the+\[\[Wikipedia\%3AArticle+wizard2\.0\|Article+Wizard\]\]\.%27%27%27%27%27%2F https://en.wikipedia.org/w/index.php?title=Special:Search&limit=9999&offset=0&ns0=1&ns1=1&ns2=1&ns3=1&ns4=1&ns5=1&ns6=1&ns7=1&ns8=1&ns9=1&ns10=1&ns11=1&ns12=1&ns13=1&ns14=1&ns15=1&ns100=1&ns101=1&ns118=1&ns119=1&ns710=1&ns711=1&ns828=1&ns829=1&ns2300=1&ns2301=1&ns2302=1&ns2303=1&search=insource%3A%2F\|+quote++%3D+You+may+also+wish+to+consider+using+a+Wizard+to+help+you+create+articles\.+See+the+\[\[Wikipedia\%3AArticle+wizard+2\.0\|Article+Wizard\]\]\.%27%27%27%27%27%2F Or if you want a list of 20,000 pages, try my sandbox. – Jonesey95 (talk) 04:29, 20 December 2023 (UTC)
Looking at #Increasing the post-expand include size just above led to me checking the NewPP report in the HTML source of a couple of pages. The NewPP output no longer displays limits. As an example, consider the source of Golden ratio which currently includes "Post‐expand include size: 370602". On 22 July 2023 a VPT report showed that page with "Post‐expand include size: 359791/2097152 bytes". Was that change mentioned somewhere? I found it useful to both see the limit and to show it to others. Johnuniq (talk) 08:02, 20 December 2023 (UTC)
The source does include that lower down ("postexpandincludesize":{"value":370602,"limit":2097152}); as far as putting that (back) in the newpp report - you're going to have to ask over at phab, use the project type "mediawiki-parser" on your ticket. — xaosfluxTalk11:06, 20 December 2023 (UTC)
Interesting, seems like a bug to me, so I filed T353793. In the meantime, you can still see the limits in the report shown below the wikitext editor when previewing, e.g. [50] (at the bottom under "Parser profiling data"). Matma Rextalk11:22, 20 December 2023 (UTC)
Category:Pages where post-expand include size is exceeded now has over 1,700 pages (many in mainspace) that are exceeding the post-expand include size. As our templates and modules get more complex, with many templates/modules calling other ones to perform various functions, this is causing various pages to hit this limit (expecially ones that call the same template over and over again, like {{lang}} oder {{sfn}} or other templates often used numerous times in the same page.
It seems to be me that it would be beneficial to bump this up a bit. It's long been set at 2 MiB (2,097,152 bytes), and an incremental increase to 2.5 MiB would solve the bulk of these cases. For my part, I'm going to have to undo a usability feature I added to a template recently because the extra template calls in it pushed at least one article over this limit.
There's a Phabricator ticket, T189108, open since 2018 asking for exactly that increase, but no action's been taken on it. I'm not sure what it takes to get it to happen. — SMcCandlish☏¢ 😼 04:54, 20 December 2023 (UTC)
You would have to prove that a modern machine with the same GB amounts of RAM and similar CPU to an five year old machine can have an bigger post-expand include size, whithout affecting cpu runtime or RAM usage. The Wikisources that have complained about this, moved their transcluded pages to the target page, solving the issue. For reference, wikisource transcludes a bunch of A4 pages from the "Page" namespace to mainspace. Snævar (talk) 08:46, 20 December 2023 (UTC)
I just tested one of the "offending" pages in question, and not only did it not load slower on my ancient Mac from 2010 (13 years old now!) than on my super-mega Windows edgelord gaming system that I built a year and a half ago with the fastest AM4 CPU then on the market and more RAM than god, it actually loaded faster, in all of view, edit, and preview, probably because I don't have something like 120 browser tabs in 3 browsers open on the Mac. The point being, it wasn't terribly slow, even on old-ass hardware. — SMcCandlish☏¢ 😼 11:14, 20 December 2023 (UTC)
@Snævar The Post-Expand Include Size primarily impacts server time, not client time. Almost all of the pages that exceed the limit do so because they contain nested templates, and those templates end up being double-counted every time there is a nesting (this includes templates that call Lua modules). For multi-level nesting the problem just compounds. If you subst all the templates on the page you can easily reduce the PEIS down well below the limit, and the client would get the same rendered HTML regardless of how many nested templates are present. In that case, a larger PEIS just means that the server may take longer to process the page before serving it if it cannot used the cached version. --Ahecht (TALK PAGE) 22:29, 20 December 2023 (UTC)
They do occasionally change these limits. They recently increased the maximum transclusion depth from 40 to 100, which I assume means that is rarely an issue now. — Jts1882 | talk10:27, 20 December 2023 (UTC)
Yes - the Wiktionary projects also recently had their Lua memory limit raised to 100MB, since Wiktionary projects tend to be very reliant on Lua. While template simplification is a good thing, it does have its limits. Theknightwho (talk) 21:53, 20 December 2023 (UTC)
These are hard decisions. Mostly because there are no takesies-backsies (if it turns out after 5 months that this was a very bad idea, it is hard to go back on it). There is also a pretty high level of 'one more lane bro' (Add a lane and drivers will fill it and you'll be stuck in traffic once more and ask for yet another lane) to something like this. I personally feel that higher page sizes are not a good idea, as they result in multi second downloads for most people which is just a bad experience for a website. The Post expand include size helps combat that. However it might be worth evaluating if the complexity increase is outpacing the article size increase and if it is worth raising that. Example if the PEIS is 1.2x the actual page size, it might effectively still produce reasonable downloads and reasonable complexity increase. But i'm not sure how we would measure that right now. —TheDJ (talk • contribs) 10:56, 20 December 2023 (UTC)
Well, it's only a tiny fraction of our pages; I don't think anyone's head will explode if they go to some of our longest lists or most complex table-based articles and have to wait a few extra seconds while all the rest of the site is quite snappy for them. I'm actually directly editing and previewing and re-editing some of these huge pages, and it's just not that big a deal. As for how to measure this stuff, I'm out of my depth on that. I'm a lot of kinds of nerd, but not an MW installation management nerd. — SMcCandlish☏¢ 😼 11:14, 20 December 2023 (UTC)
If you block a thread for 30+ seconds, that is quite a bit. Doesn't matter how many pages, it's how much cpu time it is relative to the other traffic. A page that takes 30x times as much time as a 'normal page' effectively takes the place of 30 page views/saves. That is a heavy ratio. And the effects of this are masked by all kinds of levels of caching, but those too have their limits. —TheDJ (talk • contribs) 11:18, 20 December 2023 (UTC)
This comes up frequently at DYK. See for example, WT:DYK#Oddity in the December 18 noms. It is perhaps the major reason we periodically switch to running two hook sets per day; once Template talk:Did you know/Approved exceeds the PEIS, we need to get it back down to below the limit to unbreak the page. It's really unfortunate that a major user-facing policy decision (i.e. how long a hook stays on the main page) is being driven by a technology limit like this. RoySmith(talk)15:29, 20 December 2023 (UTC)
While true, my browser has issues just rendering that page (and that's on an M1 mac on a superfast connection. That exemplifies that that page is just too big for comfort and the process should be redesigned. —TheDJ (talk • contribs) 17:38, 20 December 2023 (UTC)
Most pages that exceed the limit do so because of the double, triple, etc. counting of content served via nested templates, not because the actual rendered HTML is particularly large. --Ahecht (TALK PAGE) 22:34, 20 December 2023 (UTC)
Quickly testing these pages, most have issue with loading correctly. As most of Wikipedia's readers are already on mobile, and that's only going to increase, the solution is to fix the issues with these pages not making the problem worse. -- LCU ActivelyDisinterested«@» °∆t° 20:50, 20 December 2023 (UTC)
Re "what it takes to get it to happen" (skip the first paragraph if you like) : en.Wiktionary has this issue too, and also had a similar issue with pages exceeding the Lua memory limit; we filed a Phabricator request in 2017 about the latter, which got closed as wontfix because "you'll just want more later", and for 6 years after that, everyone reporting that Wiktionary was permanently broken got (as Tim Starling put it) rage-closed by folks on Phabricator for reporting duplicates of the old wontfix issue. Wiktionary accepted that some pages were permanently broken and provided people tips on how to "edit section" and then "preview section" so as to see whatever section they were interested in. And then, by pure chance, Tim Starling saw a screenshot of a broken page on Mastodon, saw the info the error message linked to, and the Phabricator ticket, and realized it was a simple fix and that it made more sense to periodically increase the memory limit rather than require Wiktionary to reconfigure our entire site, and so within two weeks he'd fixed the issue. His advice was: on Phabricator, tag all relevant software components, and tag all teams that might be responsible for maintenance, "since teams typically have triage meetings", but also look at the actual git logs, see who maintains the relevant code, and directly notify "developers who you think could do the work". He did also say "If you have a task which you think is low effort and high reward, you can pitch it to me directly." And in the same way that improved Lua garbage collection helped Lua-memory-limit issues independently of increasing the limit, it seems like there are some ?inefficiencies? in how the software calculates PEIS which could be addressed here, which might help prevent pages from exceeding the PEIS limit even if that limit isn't increased, e.g. apparently the size of some things is counted multiple times under certain conditions; this was reported but closed as wontfix years ago, but maybe if we figure out who is best qualified to give it a second look... -sche (talk) 21:20, 20 December 2023 (UTC)
Speaking as somebody who has a lot of experience doing both commercial and open source software development, it shouldn't be the responsibility of the person filing such a report to figure out which dev team to route it to. Sure, it's nice if you can accurately tag it, but fundamentally a bug report should be "This is what's supposed to happen, this is what actually happened, and this is how you can reproduce the problem". RoySmith(talk)21:28, 20 December 2023 (UTC)
@RoySmith It shouldn't be (and I completely agree with you), but I assure you that @-sche is simply trying to help you avoid the absolute nightmare we at Wiktionary had with the Phabricator. Theknightwho (talk) 21:56, 20 December 2023 (UTC)
I've checked a few articles, and the good news was that the only problems I could see were bottom-of-the-article navboxes that nobody ever uses anyway and that should probably be mostly removed (although it sucks to write articles so they conform to our software limit instead of making the software work for our articles). Are there any mainspace articles with more serious problems? (I am aware that the PEIS causes issues in project spaces like FAC and DYK, but would like to focus on mainspace). —Kusma (talk) 22:12, 20 December 2023 (UTC)
As an example, check out the pages User:Ahecht/sandbox/navboxes templates and User:Ahecht/sandbox/navboxes modules. The size of the HTML they produce is virtually identical, but the first has a post-expand include size of 24,806 while the second is 10,453. If you go all the way to User:Ahecht/sandbox/navboxes expanded, the size of the HTML produced is virtually identical but the post-expand include size is zero. More practically, I was able to remove the error on the first content page on the list, 2018 in spaceflight, just by changing some template calls to module calls in the footer. I further reduced it by an additional 500kb by switching {{flagicon}} for {{#invoke:flagg|main|cxxlo}}, again without changing the HTML output at all. --Ahecht (TALK PAGE) 00:20, 21 December 2023 (UTC)
There are several cases of just severe over-coding, invoking module after nested module to output a single line of text that could just be text in some cases. — xaosfluxTalk00:39, 21 December 2023 (UTC)
This reminds me of the games people used to play, trying to write a 1-line C++ program which produced the longest error message when compiled. This typically involved template abuse[51]. RoySmith(talk)00:45, 21 December 2023 (UTC)
Searching for no-linking text only
What's the best way to search for a string, but exclude links to it?
In Cirrus Search, it's almost "Foo Bar" insource:/[^\[]Foo Bar[^\]]/. That won't find Foo Bar at the start or end of the article: Cirrus Search lacks ^, $ and lookaheads which we'd need for that. (Without insource:, it would add spurious results.) If you are using some other type of search, such as one with PCRE, please let us know and we can produce a regexp which actually works. Certes (talk) 16:12, 21 December 2023 (UTC)
Error on ptwiki timeline function on user interactivity
Hi. There is a problem with the timeline function on the user interactivity tool in the Portuguese Wikipedia, as you can see here. It is working fine on the English Wikipeda. The error message says the person to contact is Σ and I did, but then saw that the user has been inactive for 18 months to the day. Thank you and I bid you a restful, peaceful and happy time over the end-of-year lull. Rui ''Gabriel'' Correia (talk) 10:57, 22 December 2023 (UTC)
Error
Buenas. Resulta que llevo 2 días sin editar. Al momento que le doy guardar en una edición me sale el mensaje:
Error
Our servers are currently under maintenance or experiencing a technical problem. Please try again in a few minutes.
See the error message at the bottom of this page for more information.
Soy de la Wikipedia en español y esta es mi cuenta.Soy de Colombia. Quisiera saber a qué se debe este problema y cómo buscar una solución. Gracias. Elías (talk) 04:43, 22 December 2023 (UTC)
Thanks for the ping. I reverted the change as this sounds worse than the issue @Trappist the monk referenced.
I likely won't look at this again, but from what I can see, when a link is clicked the class on the element changes from "mw-enhanced-rc mw-changeslist-line mw-changeslist-edit mw-changeslist-line-watched" to "mw-enhanced-rc mw-changeslist-line mw-changeslist-edit"
e.g. mw-changeslist-line-watched is dropped so the override background-image rule in the gadget stops working and shows the default background-image.
I am not sure how to resolve this as I am not too familiar with your setup @Novem Linguae but if you could comment on the ticket with details about what gadgets and scripts you have installed maybe we can get to the bottom of this. Jon (WMF) (talk) 21:29, 22 December 2023 (UTC)
I found the original, but other web sites have amp in other places in the URL. An automated way to find and flag these would be useful. 3MRB1 (talk) 03:51, 23 December 2023 (UTC)
I'd like to change the monospace font used in the source editor etc, but today I'm having difficulty reading what little information I can find on it thanks to my focus acting up. What class would I have to refer to in my CSS to change it? (And if anyone has any particularly good reference material for messing with Wikipedia CSS, please let me know!) Suntooooth, it/he (talk/contribs) 00:43, 24 December 2023 (UTC)
Wikipedia just uses your browser's default monospace font. So if you don't mind changing it on other websites too, you can change it in browser settings. In Chrome go to chrome://settings/fonts, in Firefox search for "Fonts" and click the "Advanced..." button the wub"?!"00:57, 24 December 2023 (UTC)
Neither Vector 22 or Vector 10 are working for me.
2022 in my normal size screen hides the top half of the tool bar and instead of preview, etc, over an inch of gray screen. Victor legacy overlays the left hand side with the Wikipedia logo, etc over the edit field. Doug Wellertalk15:16, 22 December 2023 (UTC)
Oddly it works perfectly on Safari and Chrome on my iPad. Ctrl+F5 gave me the editing tools on Windows 11 but left that large gray field at the bottom. That's on Chrome, Firefox works fine. Doug Wellertalk18:52, 22 December 2023 (UTC)
@PrimeHunter Logged out and in safe mode I don't get the gray space at the bottom so I can preview, see changes, save. Logged in I have the gray space at the bottom. Different editing tools each time, neither the same as on my iPad. Off to bed now, thanks for your help and patience. Doug Wellertalk20:56, 22 December 2023 (UTC)
Everything is now working with Victor 2022 on iOS, Firefox, Opera, Vivaldi and Edge, only broken on Chrome
Which puzzles me. It's fortunate as I can't blank any page using Chrome as I can't save anything because of the gray space was the publishing tools should be. Doug Wellertalk08:44, 23 December 2023 (UTC)
Feel free to post a screenshot. Because the problem goes away in safe mode, it means it's one of your gadgets or user scripts. Try blanking your User:Doug Weller/vector-2022.js as mentioned above (and any other common.js/global.js/[insert-skin-name-here].js/common.css/global.css/[insert-skin-name-here].css files) and see if the problem goes away. If the problem goes away, it's a user script. If it doesn't, it's a gadget. –Novem Linguae (talk) 09:03, 23 December 2023 (UTC)
Thanks. Forgot aobut that. It's useful but my main concern right now is that it's only broken in Chrome, I don't want to break it in other browsers. Tonight I'm going to look at all my scripts and get rid of any I can do without. Doug Wellertalk15:03, 23 December 2023 (UTC)
Victor 2020 works (and 2010 no longer works) but I lose the editing tools I want and I still am being told as I'm not an Admin I can't upload a screenshot.
Again, it's only Chrome where I don't see the editing tools I have on other platforms. I'll try to figure out their names. Sorry to be such a pain. Doug Wellertalk19:47, 23 December 2023 (UTC)
Odd that you can't upload a screenshot. Both enwiki and commons allow non-admins to upload screenshots. Next time it happens, can you paste the exact error message? Word for word? That way I can search the code for the error message and see where it's coming from. –Novem Linguae (talk) 19:52, 23 December 2023 (UTC)
@Novem Linguae Thanks but it's not code, just Something went wrongThis page is currently protected, and can be edited only by administrators.
Some templates and site interface pages are permanently protected due to visibility. Occasionally, media files and gallery pages are temporarily protected because of editing disputes.
The reason for protection can be found in the protection log.
That sounds like the "interface administrators only" message. Are you trying to upload a .js or .css file by accident, instead of a screenshot which is normally something like .jpg or .png? What's the name of the file you are trying to upload? –Novem Linguae (talk) 20:20, 23 December 2023 (UTC)
Your toolbar is made by wikEd at Special:Preferences#mw-prefsection-gadgets. Does the problem go away if you disable that on a pencil icon at the top right of edit pages? Or disable wikEd in preferences? It will remove the toolbar. I know you like it but it's just for testing. The wikEd toolbar should have an additional group of icons to the right similar to File:WikEd screenshot.png but your screenshot is cut off. Do you have that icon group at the right when wikEd is enabled? PrimeHunter (talk) 23:18, 23 December 2023 (UTC)
@PrimeHunter Yes, I've got the pencil icon. When I click on it the gray area disappears and I have the publishing tools as well as the cite dropdown. But lose the strikethrough, etc toolbar. And somehow I've lost the search toolbar which is a bit of a shame. However, using the pencil icon, I can switch between adding citations easily and being able to preview and save, etc. But no search box. It must be a Chrome issue because on every other browser I've tried I have it all, including the search box. So I guess I'll give up on Chrome for editing right now. Doug Wellertalk08:00, 24 December 2023 (UTC)
@Novem Linguae Thanks. Looking back at earlier problems, I think you are right. Really appreciate your help. Life is difficult enough for me at the moment because of my health to have this as well,. being able to edit Wikipedia helps keep me going. Doug Wellertalk12:41, 24 December 2023 (UTC)
On petscan, the category names (without "Category:") go into the "Categories" box in the "Categories" tab, all the namespace boxes in the "Page properties" tab should be checked (the default is just mainspace, but both those categories are mostly or entirely Talk: - all the pages in the intersection are, anyway), then click the "Do it!" button. You end up with this, which agrees with the search result above. —Cryptic18:16, 23 December 2023 (UTC)
.mw-tag-markers{display:none;} in your CSS should do it. (I haven't looked very hard for side effects. You don't mind them going away on diffs and so on, too?) —Cryptic21:54, 23 December 2023 (UTC)
Hi! Thank you so much, this seems to have worked perfectly. Does Custom CSS apply for everyone who looks at my contributions, or only me? This also seems to only work on the "My Contributions" page, not the "View History" page of a specific article. Pauliesnug (talk) 07:47, 24 December 2023 (UTC)
Just you. It is technically possible to hide the 2017 editor tag only from your edits on page history, but it's so complicated it might slow down the time it takes to render the page. If you don't care if it's hidden for other users as well, you might as well hide the tag altogether (by removing :is(...) ). Nardog (talk) 08:21, 24 December 2023 (UTC)
Hi. What preference did I bump to change the edit window to a divided pane? (edit 50% on left, the article as-is 50% on right) I would like to restore the edit window to full width. I checked your archive and hope this isn't a FAQ. Thank you. -SusanLesch (talk) 17:36, 24 December 2023 (UTC)
I just noticed that HTML entities seem to be getting parsed more agressively than they used to. For example:
"& n d a s h ;" (with spaced removed) should render as "–" and still does. Live example: "–"
"& a m p ; n d a s h ;" (with spaces removed) used to render as "& n d a s h ;" (with spaces removed) but is now rendering as "–". Live example: "–"
Was this change intentional? It seems like it will result in incorrect rendering in articles that discuss HTML entities, and in some old talk page discussions. -- Beland (talk) 23:46, 21 December 2023 (UTC)
And you can't prevent the parsing by using a numeric entity for either the ampersand or the trailing semicolon (wrapping with <syntaxhighlight lang="wikitext" inline="1">...</syntaxhighlight> sort of works):
That's useful information. It looks as if the unescaping (i.e. turning & into &, etc.) is happening exactly twice, rather than once (as it should) or arbitrarily often (as it would if done in a "while something to do" loop). Presumably two bits of code disagree on whether a string passed between them has already been unescaped or still needs to be changed. Certes (talk) 01:00, 22 December 2023 (UTC)
This issue appears a couple of days ago, the images in article are slow to load or not load at all. I thought it would resolve by itself, but it doesn't appear to be happening. I don't know if it has something to do with article size since it mostly affects large articles, although I also see the same thing in Wikimedia Commons. (A Commons issue?) I see it using multiple computers and different browsers, but mobile appears to work fine. Hzh (talk) 19:06, 24 December 2023 (UTC)
I've seen comments on Reddit that suggests that this is a UK specific issue. It's been bothering me for the last few days also. I don't really know where to complain about this. Hemiauchenia (talk) 23:10, 24 December 2023 (UTC)
Thanks, but I wouldn't know how to write a proper technical report. For clarity specifically upload.wikimedia.org domains (and thus images in articles) appear to be affected. Hemiauchenia (talk) 23:53, 24 December 2023 (UTC)
Sup, mid and subscript in one template?
I want to output something like 13 5 9, but I struggled to find one single template that suits this senario. Is this really possible? Or I just have to do a bit of workaround with {{su}} and {{sup}} as I did? Unnamelessness (talk) 12:31, 24 December 2023 (UTC)
For some reason, this image (which is hosted locally) doesn't render and the image page itself says "File not found: /v1/AUTH_mw/wikipedia-en-local-public.27/2/27/Ignatyevo.jpg". Clicking on the other revisions generate a server error message saying "Error: 404, Not Found at Mon, 18 Dec 2023 17:25:30 GMT". Note that this was discussed on Discord. QueenofHearts ❤️ (she/they 🎄 🏳️⚧️) 17:26, 18 December 2023 (UTC)
As far as I can tell, that is a totally different image, which, by pure coincidence, was posted to Commons with an incredibly similar name. QueenofHearts ❤️ (she/they 🎄 🏳️⚧️) 07:37, 20 December 2023 (UTC)
Post an bug to phabricator, in a worst case scenario, they might need to pull the image from backups. SRE-swift-storage are responsible for the thumbnail, not sure about where base images are stored. Snævar (talk) 08:54, 20 December 2023 (UTC)
Confirmed. Original is gone and likely has been since at least 2019. Wikimedia didn't have backups of media files until 2021, so there is no recovery possible and it is probably best to delete the page. Based on the google results it is clear that it was an extract of the Nasa worldwind set (satellite photo) of the airport. Likely by the Landsat or MODIS earth observing satellites. —TheDJ (talk • contribs) 14:40, 20 December 2023 (UTC)
This probably won't work, since they haven't edited since 2017, but @Timvasquez: as the original uploader, do you happen to have this file somewhere or know where to find it? Because it looks like the file is irrecoverable and the only solution is to overwrite it. QueenofHearts ❤️ (she/they 🎄 🏳️⚧️) 20:08, 20 December 2023 (UTC)
Thank you very much! However, I now have the issue of I'm not able to export it to Commons, because one of the revisions is missing a file. Is there any way to resolve this? QueenofHearts ❤️ (she/they 🎄 🏳️⚧️) 22:08, 20 December 2023 (UTC)
Ouch, I didn't notice that, thanks. In fact I failed to notice "A user with 11 edits" on the contribs. What I looked at when I only saw 6 edits was user rights which doesn't say confirmed or auto confirmed. Perhaps that's something to do with the age (created June 2006)? Johnuniq (talk) 08:41, 26 December 2023 (UTC)
I thought I remembered seeing "autoconfirmed" or "confirmed" under the Special:ListUsers link for "user rights" on someone's contribs page. However I can't find an example showing that now. Johnuniq (talk) 09:04, 26 December 2023 (UTC)
An implicit member is someone who became autoconfirmed the normal way, i.e. by the combination of account age and total edits (as noted above, deleted edits are counted toward that total). Somebody who isn't an implicit member will have been confirmed by admin action. --Redrose64 🦌 (talk) 10:26, 26 December 2023 (UTC)
What can cause templatestyles to not affect the page?
Hi, I've used the {{sronly}} template for a few things before and have noticed that sometimes the template styles (which in this case hide the visual text) will never make it to the webpage leaving the text visible. I had thought after a discussion on the talk page that maybe it was just linking, but I've run across other situations where the CSS doesn't work unless it's defined inline:
There is at least one bug with using templatestyles tags inside a link, see phab:T200704, with one or two different symptoms, most of which come down to "doesn't work where it's supposed to". Izno (talk) 00:13, 27 December 2023 (UTC)
The inevitable pile of redlinked maintenance categories that I can't resolve on my own...
Category:Television by decade by country has been moved to Category:Television by decade and country, but as so often happens has been left non-empty by failure to update a template-transcluded category in the implicated template — the problem is that I can't even find the category in the template in order to fix it manually: it has to be coming from {{Decade in television by country category}}, because that's literally the only thing on any of the pages in the redlink at all, but that category declaration is nowhere to be found in either the template or its documentation, which means it's being smuggled in from another multilevel recursive transclusion. Can somebody locate and fix this?
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) 13:34, 29 December 2023 (UTC)
Rater options changed?
Today I created a redirect which I intended to rate under WikiProject Scuba, but there was no option for redirects. This is a new development as I have done this often, and as it is a redirect with possibiities, there is good reason to link it to the project. Does anyone know why this happened? Is it a bug or is someone messing around with project settings? I have previously noticed that some projects can no longer be selected on Rater, but as I am not a member, have assumed it is because they are inactive and someone has gained broad consensus to remove them from the list. This is not the case with WikiProject underwater diving alias scuba Cheers, · · · Peter Southwood(talk): 08:07, 24 December 2023 (UTC)
Thanks, It appears that Rater is automatically rating redirects as redirects, which is good, but giving the user only non-redirect classes as options, which is weird and wrong. Cheers, · · · Peter Southwood(talk): 05:26, 25 December 2023 (UTC)
@Pbsouthwood: This is because the banners autodetect redirects, there is no need to set |class=redir explicitly - just leave it blank, or omit the param entirely. This is true for all classes other than Stub, Start, C, B, GA, A, FA, List, FL and FM - each of these needs to be set explicitly, either on the individual WikiProject banners or on the enclosing banner shell.
Hi Redrose64, Automatic detection is a great feature, but Rater should not give the other (non)options and exclude the relevant (non)option. I have used Rater hundreds, possibly thousands of times for redirects, and it used to give all the options allowed by each project, now it gives all the non-options except the non-option that one would expect. That is just confusing.
I am well aware that {{R with possibilities}} goes on the redirect page, I mentioned it basically because a redirect with possibilities could have an importance to a specific project other than N/A, and when it looks like you can't select redirect as a class because it is not on the list but everything else is, it gives the impression that there is no correct way to allocate it as a redirect with real importance for the project, even though apparently it is possible but you have to bypass Rater. The information provided by Rater is conflicting, therefore confusing. Some (most?) editors may just give up using it.
A more intuitively understandable response would be for Rater to explicitly state/display that the class "Redirect" has been automatically detected, and not show any other class option in the left dropdown, as they would not be valid, while still showing all the theoretically possible importance options in the right dropdown. Cheers, · · · Peter Southwood(talk): 06:39, 26 December 2023 (UTC)
I think the adjustments being made at Module:WikiProject banner have generally been impacting rater adversely. Which I think says more about rater and its level of maintenance than about the changes being made. Izno (talk) 19:15, 24 December 2023 (UTC)
@Evad37 updated the script yesterday, and it now has much better support for the new unified class ratings approach. I'm happy about this change. In terms of redirects, it sounds like the right thing to do is to add a WikiProject and just leave the class blank/unrated, and it will magically do the right thing. That's a change in behavior, but IMO not worse (maybe even better, as it's less work, after I learn the new system). WhatamIdoing (talk) 18:32, 29 December 2023 (UTC)
Internet Archive library and Wayback Machine blocked by ISPs (Parental filters)
I frequently access the Internet Archive's library or Wayback Machine while editing at Wikipedia, but since swapping over to Three broadband it looks like these sites are blocked by parental filters [Update: in the UK]. At the moment, I still have BT broadband and that works just fine.
In Chrome, attempted access results in this response:
"This site can't provide a secure connection. archive.org sent an invalid response ..."
This can be solved by going to the broadband user account and entering credit card details for a card check to verify age, but it won't accept debit cards. Alternatively, you can go into one of the relevant phone stores with a card and photo ID, assuming there is a phone store within reach.
However, I (for one, now a pensioner) can do neither, and I will note that already when I signed-up to Three they carried out a credit check and have my age.
Another way to obtain access (it works for me on Three) is to open an Opera or Avast secure browser (and probably others) and switch to a virtual private network (VPN).
Perhaps you know of other ways around such issues, and about the technicalities?
Thanks: on my Windows PC I'm already using Google DNS. It's okay, I would only use VPN in another browser to access the I.A., and would edit here at Wikipedia as normal. Esowteric + Talk + Breadcrumbs11:08, 30 December 2023 (UTC)
I'm using Vodaphone broadband in the UK and Internet Archive is working fine for me at the moment, which suggests it isn't to do with the online safety act. Although the 2019 article that the OP linked to doesn't work.Nigel Ish (talk) 14:33, 30 December 2023 (UTC)
@Esowteric and these "Parental filters" are not something you can enable or disable as the administrator? Is your ISP forcing these on you with no way to opt out? — xaosfluxTalk16:48, 30 December 2023 (UTC)
@Xaosflux My Three broadband comes with (something like) parental filters on by default, and you have to log into your account on their web site to disable this, providing credit card (not debit card) details, or go to a Three shop with a card and photo ID. There's no facility to disable this from the local admin web-interface / control panel. Esowteric + Talk + Breadcrumbs17:07, 30 December 2023 (UTC)
OK, so it is a service you are getting from your ISP. This isn't something the English Wikipedia technical team can help you with. — xaosfluxTalk17:10, 30 December 2023 (UTC)
Note that I'm assuming that my not being able to access the domain via Three is due to parental filters, and that the domain is filtered due to it wittingly and unwittingly hosting adult material with no age verification. Esowteric + Talk + Breadcrumbs17:11, 30 December 2023 (UTC)
Hi. In this Picture of the Day discussion, there is consensus that a video file of Steamboat Willie can be promoted to POTD on January 1, 2024. This would work in the following way: the POTD would appear at midnight UTC on January 1 with a placeholder image. At midnight EST (5:00 UTC), the film enters the public domain in the US, so the placeholder gets immediately substituted with the video file as soon as Commons admins undelete c:File:Steamboat Willie (1928) by Walt Disney.webm through {{#ifexist:}}. The draft for the POTD blurb is here. Someone in the discussion mentioned about implementing a JavaScript countdown in the POTD blurb, so I created this script (largely derived from this one) that substitutes every <div style="SWcountdown" data-event="" data-end="[date]"></div> with an inline live countdown to that date followed by "until Steamboat Willie enters the public domain" and such. The countdown disappears when the video is uploaded. I've tested it on my sandbox and it seems to work well when the script is installed. The question is, how can I request that the script is displayed for every user temporarily on the Main Page, POTD template and/or other relevant pages? I imagine that an user experienced with JS is required to inspect the script before that happens, right? ObserveOwl (chit-chat • my doings) 19:59, 26 December 2023 (UTC)
Alright then. Are you suggesting an RfC to see if the film could be that day's POTD, or about the countdown? By the way, if it ever happens, it should probably be closed at least 48 hours before January 1st per guideline. ObserveOwl (chit-chat • my doings) 21:24, 26 December 2023 (UTC)
It is too late for an RFC. RFCs normally need to be kept open for a week (because some people only edit on the weekends or vice versa). It's already December 27th. January 1st's POTD content is always expected settled by the end of December 29th (about 48 hours in advance). That means we have just today, tomorrow, and the following day (the 29th) to sort it out. We will have to make do with "just" normal discussions on the highest-traffic village pump. WhatamIdoing (talk) 00:09, 27 December 2023 (UTC)
This is not an appropriate use of JavaScript on Wikipedia; there's nothing particularly extraordinary about new imagery entering the public domain. Izno (talk) 23:41, 26 December 2023 (UTC)
RfCs are not needed here, as there's already consensus in that discussion I linked. The JavaScript stuff is just a bonus, not obligatory. Although I think it is a nice touch, the blurb could be without a countdown. ObserveOwl (chit-chat • my doings) 07:34, 27 December 2023 (UTC)
My initial reaction as also to wait until Jan 2; but if people want to run it on the first, has the idea of switching the picture at midnight eastern time instead of midnight UTC been considered? Is that reasonable/feasible? LittlePuppers (talk) 05:13, 28 December 2023 (UTC)
Well, we waited a century -- during which time copyright law was turned into a demented hell on earth, partially for the sake of keeping this specific mouse imprisoned -- so I say we shouldn't wait a second longer than necessary to bust him out of his cell. jp×g🗯️11:39, 28 December 2023 (UTC)
This seems pretty complicated for a one-off exceptionally-fun but ultimately niche case. Let the admins/templateadmins fight to first-post it manually. Give a 10% raise in admin salary to whoever's edit gets it done without jumping the gun. DMacks (talk) 20:05, 30 December 2023 (UTC)
One concern is that main page content should be properly reviewed by a local administrator to ensure the correct change is being made here on the English Wikipedia, it should not be dependent on the actions of contributors on another project. — xaosfluxTalk22:12, 28 December 2023 (UTC)
Meh we rely on Commons for all the mainpage media, and there's plenty of consensus from local admins in the linked discussion. Agree that a javascript countdown timer is not a necessary addition to the larger proposal. And if it happens that no Commons admin happens to be active to undelete the media at UTC−5 00:00:00 and we're stuck with a placeholder for a few hours, that doesn't feel like a particularly serious consequence. Folly Mox (talk) 19:41, 30 December 2023 (UTC)
You are invited to join the discussion at WP:ERRORS § Today's POTD. There seems to be an issue with being able to play the video currently on the Main Page on a mobile device. Technical knowledge would be greatly appreciated, as what seems to be causing this is a line in the Main Page’s TemplateStyles. Best, —a smart kitten[meow]11:24, 1 January 2024 (UTC)
Spurious edit conflicts
I have recently noticed a recurring issue where I am sometimes taken to the edit conflict page instead of read mode when saving edits. It seems that these edits are being submitted twice, as they do pass through. –LaundryPizza03 (dc̄) 11:50, 1 January 2024 (UTC)
It could still be a bad mouse or device reporting double clicks sometimes, at least if all the cases were with the same device. Double-clicking is the usual suspect for this. PrimeHunter (talk) 22:56, 1 January 2024 (UTC)
Hi, in Vector 2022, if we want to select title of an article and we accidentally extend the selected area to select Table of Content button with the icon , then the text copied in clipboard would wrongly be (for example for the article "Wikipedia"):
Toggle the table of contents
Wikipedia
This is not the intended clipboard and the first line is not required, i.e., only "Wikipedia" as the title of the article is enough and the text "Toggle the table of contents" makes the clipboard data wrong.
The solution to this problem could be applying this style:
.vector-page-titlebar-toc{-webkit-user-select:none;/* Safari */-ms-user-select:none;/* IE 10 and IE 11 */user-select:none;/* Standard syntax */}
Finally, I should note that this unexpected behavior in clipboard functionality is frequently occurred for me both in English Wikipedia and in Persian Wikipedia. Thanks, Hooman Mallahzadeh (talk) 04:30, 2 January 2024 (UTC)
My userscript doesn't work (API calls issue?)
I've been trying to make this script in response to a userscript request with my scarce JS knowledge. However, the API call has some strange behavior and pipe remains undefined when I'm inside the scope but suddenly becomes defined outside the scope. Is this some weird thing with promises? If so, how do I wait for it to finish? Aaron Liu (talk) 04:45, 1 January 2024 (UTC)
These are jQuery promises, not ES promises. If I await it, the editor gives me an error "Missing ";" before statement" Aaron Liu (talk) 15:20, 1 January 2024 (UTC)
Await should work with jQuery promises but don't think it's supported for scripts here. I think your issue is that for a promise, all code that is meant to run using the data gotten from the promise needs to be inside the .done/.then. Galobtter (talk) 16:08, 1 January 2024 (UTC)
Ugh, that's stupid. Now the problem is that the requests are done inside a loop, and I have to return the string outside the loop. However as this is inside the .done, the str gets prematurely returned. Any thoughts? Aaron Liu (talk) 16:27, 1 January 2024 (UTC)
If you're talking about the string I return, that variable is declared outside the loop. If you're talking about the API data I get, that was originally declared outside the loop and I got undefineds as the promises didn't finish yet. Re: .then instead of .done: Doesn't .then basically just make the callbacks chainable? Anyways I've done that and it seems to have had no effect. Aaron Liu (talk) 16:42, 1 January 2024 (UTC)
Async–await is supported on all compatible browsers. They're not supported only for gadgets which undergo validation and minification. If the editor complains about this, you can silence it by putting /* jshint esversion: 8 */ at the top. – SD0001 (talk) 16:42, 1 January 2024 (UTC)
Ah, that makes sense. However, I'm looking to make this an AutoEd module... So there's no other way to do this, I guess? I'll make it standalone then. To SD0001: that comment disables all hints with a "Incompatible values for the 'esversion' and 'esnext' linting options." warning next to the comment Aaron Liu (talk) 16:51, 1 January 2024 (UTC)
Erm, this isn't related to Wikitext. AutoEd loads modules inside a function it calls without an await. If one wanted to load the module into AutoEd, they'd have to use a hacked version of it that uses async for every function, which isn't very portable. Aaron Liu (talk) 17:10, 1 January 2024 (UTC)
If AutoEd is assuming the function to be synchronous, you cannot make any API calls from it at all. There's no way to "synchronously" get the value from a promise. – SD0001 (talk) 18:05, 1 January 2024 (UTC)
AutoEd runs functions one at a time to make substitutions on the text. This model isn't well-suited to run asynchronously, because the output from one set of substitutions can affect the next set. isaacl (talk) 18:26, 1 January 2024 (UTC)
re the comment: hmm weird, looks like MediaWiki's version of Ace also applies the deprecated esnext: true flag, with which esversion isn't compatible. So you'd need to do /* jshint esversion: 8, esnext: false */ - SD0001 (talk) 18:07, 1 January 2024 (UTC)
For some reason, bot edits are now showing up on my watchlist, even though I have them excluded in my preferences, (and have for years). Anyone know why this might be? And how to fix this? Thanks - wolf03:15, 2 January 2024 (UTC)
Do you have any pinned item in "Saved filters"? If so delete it or remove it as default. Otherwise, are you accessing your watchlist through a bookmark? If so the bookmarked URL may contain the bots flag. Nardog (talk) 09:08, 2 January 2024 (UTC)
(edit conflict) It appears fixed now. I had fsr checked some filters, but then they wouldn't stay unchecked after. I had almost 200 bot edits showing (of 1000 edits/30 days). Only after closing all browsers w/ wp open, then deleting cache & browsing data would the filters finally stay reset. Now only showing <40 bot edits, so hoping to call this resolved. Thanks for the responses folks, it is appreciated. Happy New Years & Cheers - wolf10:05, 2 January 2024 (UTC)
I want to pay with it in make some graphs and summary statistics of things that aren't covered by the filters, e.g. time of day content is just likely to get added or removed, or total added / removed by each editor.
I could try to copy paste it off the screen, but that's really fiddley, exporting as csv or tab delimited would be ideal.
Any special consideration for undoing an archive change?
An editor archived all of the topics on Talk:Neutron. This was entirely good faith, I assume a simple misunderstanding. I've set the page up for automatic archiving, and I would like to revert that one archive step. Is that practical? I will need to copy the topics back and delete the archive page manually?
WhatLinksHere only find pages which currently link to a transclude the template, also if it's via a redirect. If it has code to transclude the template depending on parameters but doesn't activate the code on the current rendering of the template page itself then it's not found. insource:"TennisMatch3" finds pages with "TennisMatch3" in the source but doesn't find redirects to {{TennisMatch3}} and doesn't say whether they use the template or contain the string for another reason. For example, {{ATP Cup box}} shows an example call with {{TennisMatch3}} in parameters but it doesn't use the template itself. PrimeHunter (talk) 14:58, 3 January 2024 (UTC)
Dumb author name scraping
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? BD2412T14:45, 1 January 2024 (UTC)
It happened in [54]. 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)
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. -- GreenC04:20, 2 January 2024 (UTC)
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)
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. -- GreenC18:02, 3 January 2024 (UTC)
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)
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)
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)
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)
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)
Vector legacy (2010) drawing infoboxes very poorly
(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)
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.
@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)
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)
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. Rummskartoffel11:12, 6 January 2024 (UTC)
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}} oder {{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)
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)
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.
Vector (2022) still has limited width with the skin setting turned off
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)
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)
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)
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)
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:
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)
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:
@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)
Tech News: 2024-02
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
mediawiki2latex is a tool that converts wiki content into the formats of LaTeX, PDF, ODT, and EPUB. The code now runs many times faster due to recent improvements. There is also an optional Docker container you can install on your local machine.
The way that Random pages are selected has been updated. This will slowly reduce the problem of some pages having a lower chance of appearing. [55]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 9 January. It will be on non-Wikipedia wikis and some Wikipedias from 10 January. It will be on all wikis from 11 January (calendar). [56][57]
DOI links give generic citations in VE citation tool
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?
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 | talk14:25, 7 January 2024 (UTC)
Does anyone know how to change the RefToolbar so that it is compatible with sfn?
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)
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)
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)
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,
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)
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.
Trey from the WMF Search Patform Team here. I only have a few things to add:
Certes is right that a custom program on a dump is the most straightforward way to go.
Almost any regex search with a wildcard will time out on enwiki because there's just too much data to grep through.
Behind the scenes we use an index of wikitext letter trigrams to speed up regex searches, so insource:/\{\{cn\}\}/doesn't time out because it's searching for articles containing {{c, {cn, cn}, and n}} before grepping, which really cuts down on the amount of text to grep because those are not common trigrams.
Any non-regex terms you can add to a regex search also helps kill the wiki a little less! Since the expanded text of templates is indexed, any expanded {{cn}} template will also have citation indexed, so a query like citation insource:/(\{\{[Cc]itation needed.*){100}/ above is more efficient than without citation, because it limits the grepping to 1.3M enwiki articles instead of 6.7M articles. "citation needed" with quotes is better yet, since it limits the grepping to 0.5M articles. (e.g., "citation needed" insource:/(\{\{[Cc]itation needed.*){100}/)
Including "citation needed" in your query also excludes results where the template is only used in comments, since those are not expanded. A query like insource:/\{\{cn\}\}/ -citation gives pages that only have {{cn}} in comments.