Wikipedia:Village pump (technical)
Page for discussing Wikipedia technical issues
From Wikipedia, the free encyclopedia
| Policy | Technical | Proposals | Idea lab | WMF | Miscellaneous |
If you want to report a JavaScript error, please follow this guideline. Questions about MediaWiki in general should be posted at the MediaWiki support desk. Discussions are automatically archived after remaining inactive for 5 days.
Frequently asked questions (see also: Wikipedia:FAQ/Technical) Click "[show]" next to each point to see more details.
If something looks wrong, purge the server's cache, then bypass your browser's cache. This tends to solve most issues, including improper display of images, user-preferences not loading, and old versions of pages being shown. No, we will not use JavaScript to set focus on the search box. This would interfere with usability, accessibility, keyboard navigation and standard forms. See task 3864. There is an No, we will not add a spell-checker, or spell-checking bot. You can use a web browser such as Firefox, which has a spell checker. An offline spellcheck of all articles is run by Wikipedia:Typo Team/moss; human volunteers are needed to resolve potential typos. If you have problems making your fancy signature work, check Help:How to fix your signature. If you changed to another skin and cannot change back, use this link. Alternatively, you can press Tab until the "Save" button is highlighted, and press Enter. Using Mozilla Firefox also seems to solve the problem. If an image thumbnail is not showing, try purging its image description page. If the image is from Wikimedia Commons, you might have to purge there too. If it doesn't work, try again before doing anything else. Some ad blockers, proxies, or firewalls block URLs containing /ad/ or ending in common executable suffixes. This can cause some images or articles to not appear. For server or network status, please see Wikimedia Status. If you cannot reach Wikipedia services, see Reporting a connectivity issue. |
Search function malfunction on mobile v3
Space bar resetting the search term was recently fixed at phab:T418172. But now there is a new problem. Now the search suggestions are always one step behind the typed term. For example when searching for Trapdoor. Typing T, nothing shows up. At Trap it shows Transnistria while omitting the p letter.
Confirmed affected sites: en.m.wikipedia.org, en.m.wiktionary.org
Phone model: OnePlus 3, operating system: Android 9 / OxygenOS 9.0.6, browser: Firefox 148.0.1
Aloysius Jr (talk) 23:23, 15 March 2026 (UTC)
- This is still happening. I think it's related to the previous fixes because when I write Trap and then press spacebar it acknowledges the letter p. Aloysius Jr (talk) 22:42, 19 March 2026 (UTC)
- I just tested this on Chrome, no problem there. Now I see why this post isn't getting any traction. Firefox mobile usage share is 0.5-2%! Who will save us? Aloysius Jr (talk) 12:29, 23 March 2026 (UTC)
- Aloysius Jr, you should file a bug report on phabricator. — Qwerfjkltalk 13:00, 23 March 2026 (UTC)
- You're right, thanks. Last time I did this Anne Tomasevich very kindly did the report for me. The Phabricator account requires an email account and since this account isn't connected to an email address I never made an account there. I'm not expecting people to do stuff for me of course it's just that I was excited ideologically when I noticed that this is one of the few places on the web where an account doesn't require an email address. Aloysius Jr (talk) 14:08, 23 March 2026 (UTC)
- Aloysius Jr, really? I was under the impression Phabricator uses unified login, so your Wikipedia account would work. — Qwerfjkltalk 17:45, 23 March 2026 (UTC)
- @Qwerfjkl: It does work, but if you add a comment you automatically get added as a subscriber, and by default will be sent emails for every subsequent post - or even amendment. You can amend your settings for this, at phab:settings/panel/emailpreferences/. It's just like with Wikipedia notifications. --Redrose64 🌹 (talk) 12:11, 24 March 2026 (UTC)
- Aloysius Jr, really? I was under the impression Phabricator uses unified login, so your Wikipedia account would work. — Qwerfjkltalk 17:45, 23 March 2026 (UTC)
- You're right, thanks. Last time I did this Anne Tomasevich very kindly did the report for me. The Phabricator account requires an email account and since this account isn't connected to an email address I never made an account there. I'm not expecting people to do stuff for me of course it's just that I was excited ideologically when I noticed that this is one of the few places on the web where an account doesn't require an email address. Aloysius Jr (talk) 14:08, 23 March 2026 (UTC)
- Aloysius Jr, you should file a bug report on phabricator. — Qwerfjkltalk 13:00, 23 March 2026 (UTC)
Webfonts don't load
Hey, I noticed for a while that some imported webfonts on my CSS page from db.onlinewebfonts.com can't load anymore as they used to. Is Wikimedia blocking it or does the website block Wikimedia now? --Esperfulmo (talk) 21:30, 18 March 2026 (UTC)
- We recently blocked loading of external resources from domains that aren't specifically whitelisted in response to a security incident. See phab:T419265. * Pppery * it has begun... 21:34, 18 March 2026 (UTC)
- https://www.base64encode.org/ + https://stackoverflow.com/questions/26867893/converting-and-rendering-web-fonts-to-base64-keep-original-look sapphaline (talk) 21:46, 18 March 2026 (UTC)
- Thanks both for taking the time and attention to answer me. How can we ask to allow the aforementioned website to be able to load some basic fonts for me and anyone? This website has too many essential fonts. Using my phone I can't see them and I can't install fonts on it. --Esperfulmo (talk) 01:40, 19 March 2026 (UTC)
- Converting to base64 isn't a solution for fonts which have too many characters and for dozens of fonts to load. It's impractical unless we want to just load a font which has a character or two. --Esperfulmo (talk) 01:43, 19 March 2026 (UTC)
- You may want to remove the
Content-Security-PolicyHTTP header/modify it so that it includes your domain, then. There are multiple browser extensions to do so, e.g. "simple-modify-headers" (Firefox, Chrome). sapphaline (talk) 08:45, 19 March 2026 (UTC)- Would heavily advise NOT doing that, unless you know what any of those words mean. —TheDJ (talk • contribs) 10:53, 19 March 2026 (UTC)
- Well, obviously this needs to be restricted to
en.wikipedia.org. sapphaline (talk) 11:21, 19 March 2026 (UTC)
- Well, obviously this needs to be restricted to
- I tried to use alternative extensions since those are for the desktop version and I need to load the fonts on my phone, not my desktop which has enough fonts and it's easy to download on. I was not very sure how to implement the suggested solution with alternative extensions, so I failed to load any of my imported fonts. I want to load them on arz.wikipedia.org where they are imported from my common CSS on en.wikipedia.org. I believe a simpler exception could be made for db.onlinewebfonts.com for the case of loading fonts, am I right? I read the phabricator, but could understand the relation to my issue. They were talking about java script. I am not trying to load java script. --Esperfulmo (talk) 13:53, 19 March 2026 (UTC)
- https://cdnjs.toolforge.org/ is a whitelisted domain, owned by WMF, that has some fonts. The "Fonts-linux" package in particular might help. If not, file a task at phabricator with all of the domains you are using and tag it with ContentSecurityPolicy. Snævar (talk) 17:04, 19 March 2026 (UTC)
- https://fontcdn.toolforge.org/ provides proxy access to the fonts available from the Google font server. Note having a proxy doesn't completely eliminate privacy concerns and thus I was told (but can't locate the discussion right now) that the toolforge font server's purpose isn't to be used on Wikipedia. However it is an available resource if you are willing in essence to download the fonts from the Google font server. isaacl (talk) 17:57, 19 March 2026 (UTC)
- Thank you all. I filed an issue. I hope it gets solved. --Esperfulmo (talk) 22:17, 19 March 2026 (UTC)
- Would heavily advise NOT doing that, unless you know what any of those words mean. —TheDJ (talk • contribs) 10:53, 19 March 2026 (UTC)
- You may want to remove the
- Update
- As of now, I see the fonts load. Thanks everybody! --Esperfulmo (talk) 21:34, 23 March 2026 (UTC)
Request for Comment: VisualEditor automatic reference names
Hi, I’m Johannes from Wikimedia Deutschland’s Technical Wishes team. We are considering to work on Community Wishlist/W17: Improve VE references' automatic names and reuse. This has been a long-term issue for wikitext editors (see e.g. WP:VisualEditor/Named references) which has been among the top-voted wishes in several Community Wishlist Surveys, e.g. 2017, 2019, 2022 or 2023.
We would like your input on the solutions proposed on our project page. We are considering several options, which can be combined if desired by the community.
- Changing the default pattern for automatically generated reference names (currently
":n", e.g.":0",":1"...) to use the reference type instead (e.g."book_reference-1"). - Providing a simple mechanism for communities to configure a different default name.
- Generating automatic reference names based on the domain name (if it’s a web citation).
- Generating automatic reference names based on template parameters (e.g. "title" or "last"+"first") – defined by the community.
Feedback
Visit our project page to read about our proposal in detail and share your thoughts on metawiki – or just comment below.
Please note: We will only implement a solution if there’s clear consensus among the global community. Our intention is not to build the perfect solution, but to find a simple and lean one that alleviates the pain caused by auto generated names. We are aware that some experienced VisualEditor users might prefer an option to manually change reference names in VisualEditor, but such a UX intervention is difficult to achieve across reference types and thus out of scope for our team, we can only improve the auto-naming mechanism. We are happy about suggestions for improving certain details of the proposed solutions. Any other feedback and alternative proposals are also welcome – even though it’s out of scope for us, it might still be relevant for future work on this topic.
Please support us interpreting consensus by clearly indicating your opinion (e.g. by using support/neutral/oppose templates). We are aware of WP:NOTVOTE, but given that we are facilitating this discussion with users from different wikis, potentially commenting in their native language, clearly indicating your position helps us avoid misunderstandings.
Thank you for participating! --Johannes Richter (WMDE) (talk) 11:46, 19 March 2026 (UTC)
- Did you mean to put an RFC template here on VPT? We do not typically hold RFCs on VPT, and the text block above contains a link to the RFC on Meta, where the proposed solutions are explained in more detail. Maybe let's not have two separate RFCs? – Jonesey95 (talk) 12:34, 19 March 2026 (UTC)
- I don't think this is meant to be be a RFC here, I'm not sure it 's meant to be a RFC in on meta either. I think the purpose is to get feedback on the proposal, which should probably be done on meta so as not to split the discussion. -- LCU ActivelyDisinterested «@» °∆t° 12:47, 19 March 2026 (UTC)
- The purpose is getting feedback on our proposed solutions and community consensus to move forward. It would be ideal for us if everyone commented on meta where we're gathering input from the global community, but we want to make it easy for users to participate – if people feel more comfortable discussing on VPT, they are welcome to leave a comment here. We will share key points from one discussion with the other. If you feel it's more useful to ask all users to reply on meta, that's fine with me as well. Johannes Richter (WMDE) (talk) 13:10, 19 March 2026 (UTC)
- I don't think this is meant to be be a RFC here, I'm not sure it 's meant to be a RFC in on meta either. I think the purpose is to get feedback on the proposal, which should probably be done on meta so as not to split the discussion. -- LCU ActivelyDisinterested «@» °∆t° 12:47, 19 March 2026 (UTC)
- @Johannes Richter (WMDE) I use this method (what the luddites would call a function) in User:Polygnotus/Scripts/DeduplicateReferences Polygnotus (talk) 12:38, 19 March 2026 (UTC)
- Thanks, your script is one of the tools we saw when investigating how communities currently deal with automatic reference names. Given that citation templates vary greatly across different wikis, we wouldn't be able to just implement such a mechanism, that's why we're proposing a solution where communities can define which template parameters VE should use to generate a reference name. Johannes Richter (WMDE) (talk) 13:33, 19 March 2026 (UTC)
- +1 for making the default configurable by interface admins. Each wiki's community should be the ones deciding the preferred naming conventions, not developers. And perhaps allow each user to override even that. Conventions for citations are so controversial already we have WP:CITEVAR; we don't need more drama (hence so many options in RefRenamer). Nardog (talk) 14:30, 19 March 2026 (UTC)
- But too much configurability is also not good so yes to sitewide settings but no individual preferences please. Polygnotus (talk) 14:32, 19 March 2026 (UTC)
- The individual way to do it is to manually change the reference name in text anyways. I don’t care what solution is used as long as it is reasonably unique site-wide, mitigating hard to to detect errors of incorrect citations being used. ~ 🦝 Shushugah (he/him • talk) 14:35, 19 March 2026 (UTC)
- "too much configurability is also not good" - that's... a strange thing to hear from someone who develops user scripts. sapphaline (talk) 15:54, 19 March 2026 (UTC)
- As for me, I'm still sad that MySkin was removed. sapphaline (talk) 15:58, 19 March 2026 (UTC)
- @Sapphaline Worse is better. Polygnotus (talk) 16:44, 19 March 2026 (UTC)
- But too much configurability is also not good so yes to sitewide settings but no individual preferences please. Polygnotus (talk) 14:32, 19 March 2026 (UTC)
- I'd prefer one of these
- Use
|id=if present - Use
|ref=if present - Use name provided by editor
- Use
- These are more likely to provide names meaningful to future editors. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 16:06, 19 March 2026 (UTC)
- I don't see the difference between our proposal on content-based reference names and your ideas 1 & 2? It would be up for the communities to define which template parameters should be used to automatically create a reference name.
- We are aware of the desire to allow VE users to manually change/create a reference name, but that's out of scope for our team, as it involves complex changes to the existing VE citation workflows. Johannes Richter (WMDE) (talk) 16:31, 19 March 2026 (UTC)
- (Summoned by bot) @Trappist the monk: This might be a stupid question, but could CS1 be adjusted to fill the
|ref=parameter with the reference type automatically if there was consensus to make such a change? mdm.bla 16:52, 19 March 2026 (UTC)- cs1|2 templates can't modify wikitext so whatever value is assigned to
|ref=(if it is even present) came from an editor. cs1|2 does create a<cite>tag attribute,id=when it has at minimum a contributor/author/editor. So from this:{{cite book |author=EB Green |title=Title}}
- cs1|2 creates this (look for the
<cite>tag):<templatestyles src="Module:Citation/CS1/styles.css"></templatestyles><cite id="CITEREFEB_Green" class="citation book cs1">EB Green. ''Title''.</cite><span title="ctx_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=book&rft.btitle=Title&rft.au=EB+Green&rfr_id=info%3Asid%2Fen.wikipedia.org%3AWikipedia%3AVillage+pump+%28technical%29" class="Z3988"></span>
- —Trappist the monk (talk) 17:07, 19 March 2026 (UTC)
- Forgot to mention that anything assigned to
|ref=exceptharvoverrides the anchor id created by cs1|2:{{cite book |author=EB Green |title=Title |ref=custom anchor id}}<templatestyles src="Module:Citation/CS1/styles.css"></templatestyles><cite id="custom_anchor_id" class="citation book cs1">EB Green. ''Title''.</cite><span title="ctx_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=book&rft.btitle=Title&rft.au=EB+Green&rfr_id=info%3Asid%2Fen.wikipedia.org%3AWikipedia%3AVillage+pump+%28technical%29" class="Z3988"></span>
- —Trappist the monk (talk) 17:19, 19 March 2026 (UTC)
- Forgot to mention that anything assigned to
- cs1|2 templates can't modify wikitext so whatever value is assigned to
- This needs to be closed in favour of the meta discussion, m:Talk:WMDE Technical Wishes/References/VisualEditor automatic reference names#Request for comment. Splitting the discussion seems like a bad idea. -- LCU ActivelyDisinterested «@» °∆t° 16:19, 19 March 2026 (UTC)
- I'll leave a comment on meta, but any of these would be an improvement to the current method which effectively hacks around the technical prohibition of using numbers as refnames. CMD (talk) 08:16, 20 March 2026 (UTC)
Returning HTML comments from Lua Module
I would like to include an HTML comment with debugging information in the string I return from a Lua module
e.g.
return "<!-- DEBUG comment --><br/>TEST<br/>"
However, it seems that the parser is stripping out HTML comments as I do not see the HTML comment in the page source but I do see TEST in the rendered page. Is there a way to prevent the HTML comments from being stripped out? RedWolf (talk) 19:10, 19 March 2026 (UTC)
- If you are expecting the debug comment to 'show' in an html rendering, it won't because by their very nature, html comments are not rendered visibly. I hacked a simple test to illustrate. If I write:
{{#invoke:Sandbox/trappist the monk/html comment|main}}
- that returns:
TEST
- if I write:
{{code|lang=html|{{#invoke:Sandbox/trappist the monk/html comment|main}}}}
- that returns:
<!-- DEBUG comment --><br/>TEST<br/>
- I can also write:
{{#invoke:string|find|{{#invoke:Sandbox/trappist the monk/html comment|main}}|DEBUG comment||true}}
- that returns: 6 – the 'D' in 'DEBUG' is the sixth character in
<!-- DEBUG comment --> - So the html comment is being returned; but you can't see it in an html rendering.
- Have you considered
mw.log()ormw.logObject()? - —Trappist the monk (talk) 21:41, 19 March 2026 (UTC)
- You can use Special:ExpandTemplates to see html comments. Bawolff (talk) 22:16, 19 March 2026 (UTC)
- If I run ExpandTemplates with {{#invoke:Sandbox/trappist the monk/html comment|main}}, why can't I see <!-- DEBUG comment --> under PREVIEW when I view the Page Source (not the rendered output)? It's not in the RAW HTML output either. I know the HTML comment won't be in the rendered output but why doesn't it show up when I use FF's "Page Source" menu item? RedWolf (talk) 22:59, 19 March 2026 (UTC)
- Html comments are removed before outputting the html. Special:expandtemplates has a checkbox to disable this but you cannot disable on a normal page. Bawolff (talk) 00:23, 20 March 2026 (UTC)
- If I uncheck "Remove Comments" on ExpandTemplates it still strips the HTML comments from Page Source. RedWolf (talk) 17:31, 20 March 2026 (UTC)
- In Special:ExpandTemplates with 'Remove comments' checked, if I write this:
{{#invoke:Sandbox/trappist the monk/html comment|main<!--p-->}}
- I get the test string 'TEST' with the html comment suppressed or invisible in the html rendering. When I uncheck 'Remove comments' I get this:
- Script error: The function "main<!--p-->" does not exist.
- That suggests that the 'Remove comments' checkbox applies to input only. See the (rather sparse) description at mw:Help:ExpandTemplates § Remove comments. I suspect that Editor Bawolff is correct that html comments are stripped before final page rendering (and others are added – 'NewPP limit report' and the like).
- —Trappist the monk (talk) 17:57, 20 March 2026 (UTC)
- If you want the html comment to appear in wikitext, you can subst your module. I did that for
{{cite book/French}}etc in Module:CS1 translator. You can see that with these search results. The code to create that is at lines 654–673. - —Trappist the monk (talk) 18:16, 20 March 2026 (UTC)
- In Special:ExpandTemplates with 'Remove comments' checked, if I write this:
- If I uncheck "Remove Comments" on ExpandTemplates it still strips the HTML comments from Page Source. RedWolf (talk) 17:31, 20 March 2026 (UTC)
- Html comments are removed before outputting the html. Special:expandtemplates has a checkbox to disable this but you cannot disable on a normal page. Bawolff (talk) 00:23, 20 March 2026 (UTC)
Accidental click on multi-notification's 'mark as read'

I clicked a little too fast and clicked on the blue mark as read button for notifications for some multi-notification item for Commons like the one on the right when on this Wikipedia.
I haven't noticed any loss of notifications so I don't know
- if the marking failed fully or
- if there is some safeguard to prevent multi-items being marked as read when on another project or when it's too many items or
- whether it did successfully mark some notifications as read that I didn't mean to mark as read.
Is there any way to check if I did lose some notifications and see which? I only know of one way:
- comparing the numbers of a screenshot of Special:Notifications to the current numbers. However, the last screenshot I made of the left panel of that page is too old.
For example, what would be great is a way of seeing 'read' notifications sorted by datetime of them being marked as read, not by the datetime the notification was created. Maybe there is a tool for that or it's possible in principle but one would need to code something for it.
If you think the problem of accidental clicking on mark as read should be solved, see this wish – one can vote on it:
W472: Confirmation dialog to prevent accidental clicks on Mark All As Read for notifications.
Help with retrieving the notifications if some did get marked as read would be very appreciated. Prototyperspective (talk) 23:41, 19 March 2026 (UTC)
- And in regards to the wish: whether or not there are confirmation dialogs could be made a preference; eg with these possible values
- Show confirmation dialog for all marking
- Just for multi items (say 20 notifications from one page)
- Just for mark all as read
- Never
- I've lost quite a few notification due to accidental clicks on the blue mark as read button and again if you know how to retrieve/see these, please let me know. Prototyperspective (talk) 17:37, 20 March 2026 (UTC)
- Hi. We can see all our local notifications (read and unread) on the local wiki they originated from. E.g. That example notification in your screenshot from Commons should be available to you at c:Special:Notifications. Also, coincidentally, next week's edition of Tech News has a related entry about a new notification-archive download tool. HTH. Quiddity (WMF) (talk) 19:08, 20 March 2026 (UTC)
- I accidentally marked them as read here on English Wikipedia. I have many notifications on that other wiki and even if I were the wade through hundreds of notifications to try to find the ones I accidentally marked as read there is no way to see which I marked as read recently to identify the ones I lost. If I were to wade through hundreds of read notifications on that wiki, I can't retrieve them even with that time effort because I can't remember which of all of these I have actually read and which not. Interesting news about the notification archive tool. Hopefully, it will at least allow for backups if nobody is interested in preventing large data losses by the volunteers. Prototyperspective (talk) 19:15, 20 March 2026 (UTC)
- A user on the wish page claimed clicking on the blue dot button on notifications like the one in the image (multi-item notifications from another project) would not actually mark these notifications as read (those 16 in the example) but instead just unroll them so they're not bundled anymore in some queue if I understood things correctly.
- Does somebody know more – is that the case?
- I think if there was no bug or safeguard when users mark lots of notifications as read from another project, then I think I must have lost those that I accidentally clicked and can't do anything about it assuming I don't just have very few notifications in which case I could just go through all read notifications and try to guess which of these I have not yet read (here one thing that would help would be some info on the 'read' items whether the item was clicked or just marked as read). Prototyperspective (talk) 11:12, 24 March 2026 (UTC)
- I accidentally marked them as read here on English Wikipedia. I have many notifications on that other wiki and even if I were the wade through hundreds of notifications to try to find the ones I accidentally marked as read there is no way to see which I marked as read recently to identify the ones I lost. If I were to wade through hundreds of read notifications on that wiki, I can't retrieve them even with that time effort because I can't remember which of all of these I have actually read and which not. Interesting news about the notification archive tool. Hopefully, it will at least allow for backups if nobody is interested in preventing large data losses by the volunteers. Prototyperspective (talk) 19:15, 20 March 2026 (UTC)
- Hi. We can see all our local notifications (read and unread) on the local wiki they originated from. E.g. That example notification in your screenshot from Commons should be available to you at c:Special:Notifications. Also, coincidentally, next week's edition of Tech News has a related entry about a new notification-archive download tool. HTH. Quiddity (WMF) (talk) 19:08, 20 March 2026 (UTC)
Firefox won't scroll horizontally on Template:Navbox timeline
If a usage of Template:Navbox timeline does not fit within the available area, Firefox will not allow users to scroll to see the remainder of the timeline, whereas Chrome will. There are various examples on the testcases page. I am not sure if the issue is specific to this template, or if multiple templates are affected. ~2026-17443-39 (talk) 01:11, 20 March 2026 (UTC)
- @~2026-17443-39: It works for me in Firefox 148.0.2 (64-bit) on Windows 11. The scroll bars are just more narrow and might be overlooked. Are you sure it doesn't work for you? Try hovering your mouse over the bottom of the box. Is there no scroll bar or doesn't it work to click it? Please name a specific example. PrimeHunter (talk) 02:19, 20 March 2026 (UTC)
- I can partially reproduce this, if I make my browser-window very narrow. E.g. phab:F73288607 screenshot showing 3 logged-out windows, one with the full contents of {{Boeing 7x7 timeline}} visible, one with a narrower window that is cutting off content but doesn't have scrollbars, and one that is 639px wide that does show scrollbars. I'm not sure if that's what the temp-account user is describing, nor how to resolve it. HTH. Quiddity (talk) 20:02, 20 March 2026 (UTC)
- No scrollbars are evident in Firefox 148.0.2 under Windows 10 (normal zoom, 1280px wide), even when hovering near the bottom. In Safari on an iPad almost all of the testcases are sideways scrollable, although the scrollbar only appears once I touch and slide sideways; it vanishes a couple of seconds after I lift my finger. The ones without sideways scrolling are the first pair (Basic) and the tenth pair (
|border=child). --Redrose64 🌹 (talk) 22:27, 20 March 2026 (UTC) - This is being caused by the two properties
table-layout: fixed+width: 100%. Fixed is very particular about how things should be displayed; removing either of those properties at least makes it so you can display the whole space (with their own perhaps-obvious detrimental effects). These are added in {{navbox timeline}}. The reason it works at lower ranges is because the table is assigneddisplay: blockdue to various responsive treatments upstream which causes the fixed layout to be removed. - Perhaps the best fix is to not put timelines in navboxes. :eyeroll: Izno (talk) 05:40, 21 March 2026 (UTC)
- I don't think either
table-layout: fixedorwidth: 100%directly has anything to do with it, other than making it so the table itself doesn't expand outside of its container due to the large content. It seems that Firefox just doesn't currently supportoverflow: autoon table cells (unless you enable the preferencelayout.tables.scrollable-cells); apparently they tried to fix it, but then had to put it behind a preference because other stuff broke, and it seems they need to rewrite some table code to make it work. A workaround may be to put theoverflow: autoon a<div>inside the table cell instead of putting it on the cell directly. Anomie⚔ 13:52, 21 March 2026 (UTC)
- I don't think either
- No scrollbars are evident in Firefox 148.0.2 under Windows 10 (normal zoom, 1280px wide), even when hovering near the bottom. In Safari on an iPad almost all of the testcases are sideways scrollable, although the scrollbar only appears once I touch and slide sideways; it vanishes a couple of seconds after I lift my finger. The ones without sideways scrolling are the first pair (Basic) and the tenth pair (
- I can partially reproduce this, if I make my browser-window very narrow. E.g. phab:F73288607 screenshot showing 3 logged-out windows, one with the full contents of {{Boeing 7x7 timeline}} visible, one with a narrower window that is cutting off content but doesn't have scrollbars, and one that is 639px wide that does show scrollbars. I'm not sure if that's what the temp-account user is describing, nor how to resolve it. HTH. Quiddity (talk) 20:02, 20 March 2026 (UTC)
Automatic spelling detection for Template:convert?
Template:Convert has the parameter |sp=us for when en-US is wanted instead of en (generic). Would it be possible to have the template detect Template:Use American English and automatically apply such changes, similar to how the CS templates automatically adapt to Template:Use mdy dates? CMD (talk) 09:57, 20 March 2026 (UTC)
- This has been suggested a couple of times. My most recent response is at Template talk:Convert#Spelling (permalink). In brief, my view is that there should be a central method for parsing a page once rather than having multiple modules use their own ideas of how that should be done. Johnuniq (talk) 03:24, 21 March 2026 (UTC)
- It would need to be aware of quotes and other uses where the 'other' spelling is the correct one, even if the page uses a different spelling. -- LCU ActivelyDisinterested «@» °∆t° 17:06, 23 March 2026 (UTC)
- A {{Not a typo}} template may be utilized to prevent automatic fixing words intentionally written 'wrong', but that would require manual verifying and template-annotating each article prior to automatic processing.
--CiaPan (talk) 18:44, 23 March 2026 (UTC)
- A {{Not a typo}} template may be utilized to prevent automatic fixing words intentionally written 'wrong', but that would require manual verifying and template-annotating each article prior to automatic processing.
- It would need to be aware of quotes and other uses where the 'other' spelling is the correct one, even if the page uses a different spelling. -- LCU ActivelyDisinterested «@» °∆t° 17:06, 23 March 2026 (UTC)
File upload not working
Is the file upload feature broken at the moment? I have tried to upload a file to this Wikipedia project (not Commons), and I get this error each time: Upload failed: Could not create directory "mwstore://local-multiwrite/local-public/1/1b". Simeon (talk) 19:45, 20 March 2026 (UTC)
- It worked now: File:Dochters film poster.jpg. Simeon (talk) 20:02, 20 March 2026 (UTC)
Discrepancies in GrowthExperiments log
Please could somebody familiar with how GrowthExperiments take a look at Wikipedia talk:Mentorship#Who selects people to mentor?, where a number of discrepancies in the GrowthExperiments log are coming to light. --Redrose64 🌹 (talk) 12:56, 21 March 2026 (UTC)
Image/WebP file not showing metadata
File:MacBook Neo (256GB Citrus; 2026).webp is one of a series of images I uploaded (see gallery above), and while the other images in the series show metadata correctly, this image does not. I re-downloaded the uploaded file to confirm the presence of the metadata, and it's there according to File Properties in Windows Explorer (Windows 11). Any ideas as to what about this specific image is causing a problem? —Locke Cole • t • c • b 16:03, 21 March 2026 (UTC)
- Comparing output of exiftool for File:MacBook Neo (256GB Citrus; 2026).webp and File:MacBook Neo (512GB Blush; 2026).webp, the only two things that jump out are: 1) the "bad" one has "Auto Tone Digest" and many "Paint Correction ..." entries, while the "good" one does not; and 2) the "good" one has GPS coordinates, but the "bad" one does not.
- I have zero knowledge of how it works, the following is just my educated guess at what could have happened. You can try uploading the photo one more time – it is possible that the metadata extraction during the uploading process got confused/crashed/interrupted/whatever and it simply never finished. I don't remember if MediaWiki allows uploading the exact same file on the exact same filename, so you might need to edit it a bit. If uploading it once more works, you could re-upload the original over the second version. —andrybak (talk) 21:56, 21 March 2026 (UTC)
Push notifications for page changes?
I have a set of about 50 pages (all the currently active WP:FAC nominations) that I want to monitor with a bot. Is there some way to get a push notification when any of them have changes? I could put them on my bot's watchlist, but then I'd have to synchronously poll mw:API:Watchlist every so often which would either be too slow (if I poll infrequently) or too load-intensive (if I poll like a crazed spider monkey). I expect most pages to have an average of a couple of edits per day, so an aggregate of about one notification every 10 minutes. RoySmith (talk) 17:20, 21 March 2026 (UTC)
- I'm not sure about notifications, but you could enable "Email me when a page or a file on my watchlist is changed" in the user preferences of the bot, and make sure that email is received by the bot. I don't know if there's any delay between the edit and the email though. Also, instead of the API, it's also possible to get an RSS feed of your watchlist. --rchard2scout (talk) 01:09, 22 March 2026 (UTC)
- No idea about push notifications but polling Special:RecentChangesLinked could be tried as an alternative to polling the watchlist. You would need an enwiki page with links to the 50 current FAC articles. On that page, click "Related changes" under the tools sidebar to see the URL. The Atom feed might be useful. On the other hand, I just tried mw:Special:MyLanguage/API:Watchlist feed. Clicking API watchlist last hour gives a very fast result. Johnuniq (talk) 02:02, 22 March 2026 (UTC)
- @Samwilson, are there any plans to add a parameter label=<label id> to Action API action=query/list=watchlist, so you can fetch only the watchlist changes for the labeled lists? IKhitron (talk) 10:27, 22 March 2026 (UTC)
- @IKhitron: It's not possible yet, but I've created phab:T420839 to track it. The current priorities are to add the watchstar popup and VE functionality, but after those are done I think sorting out the missing API functions will be next up. Sam Wilson 11:14, 22 March 2026 (UTC)
- OK, I think I can get this down to two API calls: one to grab WP:FAC, and a second mw:API:Info call to query for the
lastrevidof each of the individual pages and keep a small cache. The only remaining question is that API:Info says "Maximum number of values is 50 (500 for clients that are allowed higher limits)". How do I know if my bot has the higher limit? RoySmith (talk) 13:58, 22 March 2026 (UTC)
- OK, I think I can get this down to two API calls: one to grab WP:FAC, and a second mw:API:Info call to query for the
- @IKhitron: It's not possible yet, but I've created phab:T420839 to track it. The current priorities are to add the watchstar popup and VE functionality, but after those are done I think sorting out the missing API functions will be next up. Sam Wilson 11:14, 22 March 2026 (UTC)
Why does CharInsert no longer appear for me?
The other day, I noticed that CharInsert has stopped appearing as part of the source editing interface, and it's still absent as of today, so if I want to use an arrow in my edit summary, I now have to open up the special characters menu, look for it, then cut and paste it from the wikitext. In case you want to know, I'm using Firefox and the legacy Vector skin. – MrPersonHumanGuy (talk) 18:24, 21 March 2026 (UTC)
Works for me in Firefox under Windows also Safari on an iPad. --Redrose64 🌹 (talk) 19:03, 21 March 2026 (UTC)
- I use Windows 10 on a desktop PC. – MrPersonHumanGuy (talk) 00:02, 22 March 2026 (UTC)
- Is "CharInsert" enabled at Special:Preferences#mw-prefsection-gadgets? Does it work if you log out? PrimeHunter (talk) 19:08, 21 March 2026 (UTC)
- CharInsert is enabled by default, and I've never disabled it when I went into Preferences. – MrPersonHumanGuy (talk) 00:04, 22 March 2026 (UTC)
- Did you check it still has a checkmark? Try to disable it, save preferences, enable it again and save. Reload an edit page with Ctrl+F5, not just F5 or a reload button. Does CharInsert work if you are logged out? PrimeHunter (talk) 01:04, 22 March 2026 (UTC)
- As a matter of fact, pressing Ctrl+F5 at an edit page appears to have resolved the issue. Once I tried it here (which I just did a moment ago), CharInsert came back, and so far, I haven't had to repeat the trick (or try logging out) when going to edit other pages. Hopefully CharInsert will stay for good. – MrPersonHumanGuy (talk) 10:47, 22 March 2026 (UTC)
- Did you check it still has a checkmark? Try to disable it, save preferences, enable it again and save. Reload an edit page with Ctrl+F5, not just F5 or a reload button. Does CharInsert work if you are logged out? PrimeHunter (talk) 01:04, 22 March 2026 (UTC)
- CharInsert is enabled by default, and I've never disabled it when I went into Preferences. – MrPersonHumanGuy (talk) 00:04, 22 March 2026 (UTC)
Have the notification rules changed?
This edit notified me, but judging by WP:MENTION it should not have done because Create a new comment; modifying an earlier comment does not work.
Have the rules changed? --Redrose64 🌹 (talk) 21:50, 21 March 2026 (UTC)
- I'd wager that it worked because the comment they were editing was unsigned, and they added a sig in the edit in which they added the ping. Schazjmd (talk) 21:59, 21 March 2026 (UTC)
Script warning
Hi all, I was editing Trucial States, and the following appeared in the Preview window: Script warning: TRUCIALSTATES ("Trucial States") is not a recognized country in ISO 3166-1 (Module:ISO 3166). It's now the UAE, (AE/ARE/784). I saw that the list of "Pages included in this preview" includes {{ISO 3166 code}}, so I imagine it's being invoked somewhere. Does it matter? Can it be fixed? MinorProphet (talk) 10:48, 22 March 2026 (UTC)
- In the infobox:
|iso3166code=omit - —Trappist the monk (talk) 11:15, 22 March 2026 (UTC)
- Great, that seems to have worked. Thanks very much as usual for your timely and helpful reply. MinorProphet (talk) 11:48, 22 March 2026 (UTC)
Question about a search box functionality while reading articles on Wikipedia
Is it intentional that when the empty search box is clicked or tapped three suggestions pop up? These three are usually the related articles that are located near the bottom of a webpage on a mobile device. This happens on desktop and mobile. Aloysius Jr (talk) 13:50, 22 March 2026 (UTC)
- Yes, it's a new feature which was added last year. See mw:Reading/Web/Content Discovery Experiments/Search Suggestions for more information. the wub "?!" 14:09, 22 March 2026 (UTC)
- Much appreciated. Aloysius Jr (talk) 11:44, 23 March 2026 (UTC)
XTools articleinfo sometimes omits page watchers
I'm not sure if it's a bug, or if there is some mysterious reason that XTools articleinfo tool sometimes omits the tally of page watchers of an article. Examples:
- Liberation of France – 56 watchers
- Mercury (planet) – 1,085 watchers
- Pornography Act (Austria) – (stat absent)
Is there a pattern? Mathglot (talk) 21:11, 22 March 2026 (UTC)
- Because the number of watchers is not generally reported when the number of watchers is less than 30. This is the same limitation as in info. Izno (talk) 21:16, 22 March 2026 (UTC)
- What Izno said. Compare these two API responses: Liberation of France and Pornography Act (Austria); notice how the latter is missing the
watchersprop entirely. XTools uses the same endpoint. — DVRTed (Talk) 21:36, 22 March 2026 (UTC) - mw:Manual:$wgUnwatchedPageThreshold is set to 30 for Wikimedia wikis in https://noc.wikimedia.org/conf/highlight.php?file=InitialiseSettings.php. PrimeHunter (talk) 22:26, 22 March 2026 (UTC)
- Thanks. Is this intended as some sort of privacy thing, i.e., that it is somehow harder to reverse engineer who is watching, perhaps to avoid pinning a target on users for vandals or something? It took me about two or three minutes to navigate from articleinfo for Pornography Act (Austria) to the authorship tool (21 unique) thence to contribs on the Talk page (no authorship tool on Talk pages) and pick up 4 more uniques for a total of 25. So, now I have 25 possible watchers, and I also have their usernames, which the tool doesn't give you. Sure, we can't be certain they are all watchers, or who the non-contrib watchers are, but it seems like only in that last group is one's privacy being protected. Or is there another reason for the 30 limit? Mathglot (talk) 01:59, 23 March 2026 (UTC)
- Just to avoid giving easy targets for vandals. Izno (talk) 03:21, 23 March 2026 (UTC)
- @Mathglot: See Special:PageInfo/Pornography Act (Austria)#mw-pageinfo-watchers. If you're an admin, you'll see a figure here; if not, you'll see "Fewer than 30 watchers". Compare Special:PageInfo/Mercury (planet)#mw-pageinfo-watchers which presently shows 1,036 for everybody. --Redrose64 🌹 (talk) 18:18, 23 March 2026 (UTC)
- Ooh, that's helpful! It would be great if Articleinfo tool could just emit the same string; that is much better than saying nothing, because you can tell that no bug or oversight is involved. (And would prevent a repeat of this VPT question.) Maybe I should file a Phab enhancement ticket to request it. Mathglot (talk) 18:37, 23 March 2026 (UTC)
- To pick up on your post of 01:59, 23 March 2026 (UTC): not all people who edit a page will watch it. There are gnomes who use AWB to zip through hundreds or thousands of pages making small adjustments, I don't imagine that they would necessarily want to add all these pages to their watchlists (I think that's the trap that Rich Farmbrough fell into, he ended up with a watchlist that was so long that the MediaWiki software couldn't handle it). Conversely, there are people who will watch pages that they've never edited, for various reasons (personally, I might watch a page if: it's a page that's being discussed on some board like VPT; it's one of a set of articles that I eventually wish to bring to some sort of consistency; it's a redirect with possibilities, and I have some sources, so I want to see if somebody expands it to an article; a LTA that I've blocked has been targetting that page; etc.). --Redrose64 🌹 (talk) 21:08, 23 March 2026 (UTC)
- Ooh, that's helpful! It would be great if Articleinfo tool could just emit the same string; that is much better than saying nothing, because you can tell that no bug or oversight is involved. (And would prevent a repeat of this VPT question.) Maybe I should file a Phab enhancement ticket to request it. Mathglot (talk) 18:37, 23 March 2026 (UTC)
- Thanks. Is this intended as some sort of privacy thing, i.e., that it is somehow harder to reverse engineer who is watching, perhaps to avoid pinning a target on users for vandals or something? It took me about two or three minutes to navigate from articleinfo for Pornography Act (Austria) to the authorship tool (21 unique) thence to contribs on the Talk page (no authorship tool on Talk pages) and pick up 4 more uniques for a total of 25. So, now I have 25 possible watchers, and I also have their usernames, which the tool doesn't give you. Sure, we can't be certain they are all watchers, or who the non-contrib watchers are, but it seems like only in that last group is one's privacy being protected. Or is there another reason for the 30 limit? Mathglot (talk) 01:59, 23 March 2026 (UTC)
Making this discussion easier to follow on a talk page
On this page on my mobile device, the discussions are very hard to follow while on other vital article pages, they have collapsible discussions. Any idea on how to make this mobile friendly? Interstellarity (talk) 21:23, 22 March 2026 (UTC)
- @Interstellarity: There are 1138 images. Sections are not collapsible in mobile if there are more than 1000 images. PrimeHunter (talk) 23:18, 22 March 2026 (UTC)
- @PrimeHunter I don’t see any images on the page. Can you point out to me what’s causing the problem? I might be able to do something about it, but I’ll discuss it before I do anything. Interstellarity (talk) 23:21, 22 March 2026 (UTC)
- File:Círculos Concéntricos.svg occurs 1084 times from {{VA link}}. PrimeHunter (talk) 23:22, 22 March 2026 (UTC)
- Is there a way to fix the template so that it doesn’t happen again or should I consider splitting the pages for this to work to fix the issue? I will get community consensus before splitting the page if that’s the route you think is best. Interstellarity (talk) 23:32, 22 March 2026 (UTC)
- Sort of following you here because I reverted one of your edits ...if you want to talk about it we can.... That said drop the file spam talk (why do you need all these little images?). Ideally, a page should have no more than 100 images (regardless of how small). See MediaWiki:Limit number of images in a page. Make your talk pages (or any page) as accessible as possible so you're not deterring readers/editors. Moxy🍁 23:45, 22 March 2026 (UTC)
- I’m all for making our talk pages as accessible as possible and that’s exactly what I’m trying to accomplish here. We might have different ideas on how to fix this, but I am 100% willing to talk about with you. This is why I suggested splitting the page: to make the talk pages more accessible without the information overload and figuring out where to scroll, which makes it less accessible. Interstellarity (talk) 23:54, 22 March 2026 (UTC)
- All you need to do - it would fix accessibility on every archive page and current page - is drop the icons at {{Vital article link}} for real characters like + or (GA). That said there still is Help:Template limits...but they're much higher. Images like File:Escape suspect.svg are meaningless anyways to those outside the project.Moxy🍁 00:09, 23 March 2026 (UTC)
- Another way would be using a circle emoji in template:vital article link. Probably a blue circle for vital and red for not vital article. But then again, that page is 430kb, which is on the bigger side. Snævar (talk) 07:43, 23 March 2026 (UTC)
- I have removed the image from the template sandbox. See {{Vital article link/testcases}} for how it looks. Also, you could probably get the page under the 1,000-image mark by eliminating the improper use of the template in headers, which causes a few different problems. – Jonesey95 (talk) 13:35, 23 March 2026 (UTC)
- Another way would be using a circle emoji in template:vital article link. Probably a blue circle for vital and red for not vital article. But then again, that page is 430kb, which is on the bigger side. Snævar (talk) 07:43, 23 March 2026 (UTC)
- All you need to do - it would fix accessibility on every archive page and current page - is drop the icons at {{Vital article link}} for real characters like + or (GA). That said there still is Help:Template limits...but they're much higher. Images like File:Escape suspect.svg are meaningless anyways to those outside the project.Moxy🍁 00:09, 23 March 2026 (UTC)
- I’m all for making our talk pages as accessible as possible and that’s exactly what I’m trying to accomplish here. We might have different ideas on how to fix this, but I am 100% willing to talk about with you. This is why I suggested splitting the page: to make the talk pages more accessible without the information overload and figuring out where to scroll, which makes it less accessible. Interstellarity (talk) 23:54, 22 March 2026 (UTC)
- Sort of following you here because I reverted one of your edits ...if you want to talk about it we can.... That said drop the file spam talk (why do you need all these little images?). Ideally, a page should have no more than 100 images (regardless of how small). See MediaWiki:Limit number of images in a page. Make your talk pages (or any page) as accessible as possible so you're not deterring readers/editors. Moxy🍁 23:45, 22 March 2026 (UTC)
- Is there a way to fix the template so that it doesn’t happen again or should I consider splitting the pages for this to work to fix the issue? I will get community consensus before splitting the page if that’s the route you think is best. Interstellarity (talk) 23:32, 22 March 2026 (UTC)
- File:Círculos Concéntricos.svg occurs 1084 times from {{VA link}}. PrimeHunter (talk) 23:22, 22 March 2026 (UTC)
- @PrimeHunter I don’t see any images on the page. Can you point out to me what’s causing the problem? I might be able to do something about it, but I’ll discuss it before I do anything. Interstellarity (talk) 23:21, 22 March 2026 (UTC)
These images could be moved into WP:TemplateStyles if there is a specific use needing it. Izno (talk) 18:18, 23 March 2026 (UTC)
- I thought emoji too, I found 🎯: Unicode Character 'DIRECT HIT' (U+1F3AF). However I can't find one without the dart, or that's monochrome. --Redrose64 🌹 (talk) 19:07, 23 March 2026 (UTC)
- If you want to put in that much effort into replacing this icon, you can draw vector graphics from Lua modules these days: mw:Extension:Scribunto/Lua reference manual#SVG library. Matma Rex talk 01:19, 24 March 2026 (UTC)
Duplicate reference included mysteriously in template expansion.
I'm stuck. Selenium has a ref with a cite template that I cannot track down. I know it is being introduced by the last clause of a {{MultiReplace}} in the helper template {{Element-symbol-to-oxidation-state-entry}}.
- Testing just the helper template does not show the extra citation, Template:Element-symbol-to-oxidation-state-entry/sandbox (just one Greenwood).
- The extra citation appears in {{Infobox selenium}} and {{Infobox element}}.
- If I use Special:ExpandTemplates with
{{Infobox element}}and context page "Selenium" and uncheck all of the boxes, the phantom ref appears. But if I check the Remove comments box, my ref goes away and a different one appears. - Selenium oxidation data should not trigger the last clause of the MultiReplace. Elements which do trigger the last clause, eg {{Infobox copper}}, do not show the extra Greenwood citation, just the one it should.
I've run out of things to try. Johnjbarton (talk) 03:37, 23 March 2026 (UTC)
- @Johnjbarton: All {{MultiReplace}} parameters in {{Element-symbol-to-oxidation-state-entry}} are evaluated whether they are used or not so the ref is always evaluated even if there is no asterisk to replace with the ref. If a ref has been parsed during page evaluation then it's added to the references list even if the code containing the ref is not part of the generated wikitext. That's one of several odd quirks with refs, maybe because they are part of an extension mw:Extension:Cite and its authors just did what seemed practical or easy when integrating refs into MediaWiki. If you want to avoid it then you can do a test to see whether there is an asterisk and only make the replacement then. PrimeHunter (talk) 17:14, 23 March 2026 (UTC)
- Thanks! Your analysis makes sense. But I've not found a way to solve this. I need to MultiReplace the string in both the * and not * cases. I tried
#tagbut as far as I can tell that fails. The order of evaluation is completely opaque :-( - Maybe a better fix is to emit
<ref name=GE2/>when the * is used and tell editors to add a list-defined ref at that name. The ref here is widely used among the Elements so this is not a terrible choice. Johnjbarton (talk) 22:59, 23 March 2026 (UTC)- @Johnjbarton: I meant to make a test before the replacement like . PrimeHunter (talk) 23:48, 23 March 2026 (UTC)
- Hmm. So I really don't understand the problem. I originally thought all of the templates would be expanded, then the lower level operations would work through the logic: modifying content optionally gives a final stream of content. If the input string does not contain
<sup>*</sup>then the ref is not in the final content. - But you said "the ref is always evaluated", which matched the failures I saw. How can the
ifexpralter the outcome? Both branches will be expanded ? so the ref will alway be expanded? If the problem is that both branches of the conditional are evaluated, how can another conditional help? Johnjbarton (talk) 00:12, 24 March 2026 (UTC)- @PrimeHunter OK Thank you! I put your version in and it works. I don't understand it but it works. Johnjbarton (talk) 00:25, 24 March 2026 (UTC)
- @Johnjbarton: if-expressions generally use lazy evaluation where only one of the then and else parts is evaluated. The template language also uses lazy evaluation for parameters so a parameter is only evaluated if the template asks for the value, but {{MultiReplace}} apparently does ask for the value of all the parameters even if some of them may not affect the output. PrimeHunter (talk) 01:29, 24 March 2026 (UTC)
- Hah, of course it does. Thanks, that makes a lot more sense. Johnjbarton (talk) 03:03, 24 March 2026 (UTC)
- @Johnjbarton: if-expressions generally use lazy evaluation where only one of the then and else parts is evaluated. The template language also uses lazy evaluation for parameters so a parameter is only evaluated if the template asks for the value, but {{MultiReplace}} apparently does ask for the value of all the parameters even if some of them may not affect the output. PrimeHunter (talk) 01:29, 24 March 2026 (UTC)
- @PrimeHunter OK Thank you! I put your version in and it works. I don't understand it but it works. Johnjbarton (talk) 00:25, 24 March 2026 (UTC)
- Hmm. So I really don't understand the problem. I originally thought all of the templates would be expanded, then the lower level operations would work through the logic: modifying content optionally gives a final stream of content. If the input string does not contain
- @Johnjbarton: I meant to make a test before the replacement like . PrimeHunter (talk) 23:48, 23 March 2026 (UTC)
- Thanks! Your analysis makes sense. But I've not found a way to solve this. I need to MultiReplace the string in both the * and not * cases. I tried
- You wrote "Template:Element-symbol-to-oxidation-state-entry/sandbox (just one Greenwood)", but there are two like in {{Infobox selenium}}, just in another order. It's a named ref and the first occurrence comes early so it's reference 1. The d at reference 1 is caused by Selenium but doesn't link anywhere since the reference is not actually used for Selenium. Only b from iron (Fe) is used and has a working link. PrimeHunter (talk) 17:49, 23 March 2026 (UTC)
- Yes, sorry, I should have said that only one comes from the "Se" entry. Johnjbarton (talk) 18:06, 23 March 2026 (UTC)
Announce: wiki mail verification tool
Headers and contents of emails sent using Special:EmailUser are signed with DKIM by the Wikimedia mail servers and can be verified provided that the email is available in raw format including all relevant headers. I developed a tool to do just that. It works completely in the browser using WebAssembly and JavaScript to do the verification in order to address any privacy concerns.
My main motivation for writing this tool was that we have cases in German language Wikipedia where users are harassed using emails sent this way and administrators have to verify that such emails forwarded to them are genuine. I assume there are similar cases in the English language Wikipedia. Count Count (talk) 06:31, 23 March 2026 (UTC)
Archive bot
Is there any bot that can be used to archive specific talk page discussions based on a filter rather than just date? Specifically, I'm hoping to automate archival of KiranBOT notifications regarding threads being archived. Teahouse suggested asking here. ChompyTheGogoat (talk) 08:41, 23 March 2026 (UTC)
- ChompyTheGogoat, {{User:ClueBot III/ArchiveThis}} with
|archivenow=should work. — Qwerfjkltalk 09:49, 23 March 2026 (UTC)
Tech News: 2026-13
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Weekly highlight
- Wikimedia site users can now log in without a password using passkeys. This is a secure method supported by fingerprint, facial recognition, or PIN. With this change, all users who opt for passwordless login will find it easier, faster, and more secure to log in to their accounts using any device. The new passkey login option currently appears as an autofill suggestion in the username field. An additional "Log in with passkey" button will soon be available for users who have already registered a passkey. This update will improve security and user experience. The screen recording demonstrates the passwordless login process step by step.
- All wikis will be read-only for a few minutes on Wednesday, 25 March 2026 at 15:00 UTC. This is for the datacenter server switchover backup tests, which happen twice a year. During the switchover, all Wikimedia website traffic is shifted from one primary data center to the backup data center to test availability and prevent service disruption even in emergencies.
Updates for editors
- Wikimedia site users can now export their notifications older than 5 years using a new Toolforge tool. This will ensure that users retain their important notifications and avoid them being lost based on the planned change to delete notifications older than 5 years, as previously announced.
- Wikipedia editors in Indonesian, Thai, Turkish, and Simple English now have access to Special:PersonalDashboard. This is an early version of an experience that introduces newer editors to patrolling workflows, making it easier for them to move from making edits to participating in more advanced moderation work on their project.
- The Special:Block now has two minor interface changes. Administrators can now easily perform indefinite blocks through a dedicated radio button in the expiry section. Also, choosing an indefinite expiry provides a different set of common reasons to select from, which can be changed at: MediaWiki:Ipbreason-indef-dropdown.
- Mobile editors at several wikis can now see an improved logged-out edit warning, thanks to the recent updates from the Growth team. These changes released last week are part of ongoing efforts and tests to enhance account creation experience on mobile and then increase participation.
View all 36 community-submitted tasks that were resolved last week. For example, the bug that prevented mobile web users from seeing the block information when affected by multiple blocks has been fixed. They can now see messages of all the blocks currently affecting them when they access Wikipedia.
Updates for technical contributors
- Images built using Toolforge will soon get the upgraded buildpacks version, bringing support for newer language versions and other upstream improvements and fixes. If you use Toolforge Build Service, review the recent cloud-announce email and update your build configuration as necessary to ensure your tools are compatible.
- The API Portal documentation wiki will shut down in June 2026. API keys created on the API Portal will continue to work normally. api.wikimedia.org endpoints will be deprecated gradually starting in July 2026. Documentation on the API Portal is moving to mediawiki.org. Learn more on the project page.
Detailed code updates later this week: MediaWiki
In depth
- WMDE Technical Wishes is considering improvements to automatically generated reference names in VisualEditor. Please check out the proposed solutions and participate in the request for comment.
Tech news prepared by Tech News writers and posted by bot • Contribute • Translate • Get help • Give feedback • Subscribe or unsubscribe.
MediaWiki message delivery 16:49, 23 March 2026 (UTC)
Database error
I tried to visit Special:Log/Materialscientist and ended up getting an error message:
To avoid creating high database load, this query was aborted because the duration exceeded the limit. If you are reading many items at once, try doing multiple smaller operations instead.
[1fbe34c9-0299-4efd-89e3-312f624fd6f9] 2026-03-23 20:11:50: Fatal exception of type "Wikimedia\Rdbms\DBQueryTimeoutError"
I understand what it means, and obviously it's a standard server message for "sorry, too much happening right now; I can't help you and everyone else". But (1) any idea what's going on right now? I've never gotten this kind of message when trying to view user logs before, and it's the default size (I assume 50 entries), not something massive like 5000 entries. (2) How could one view a smaller version of a user log page, without accessing it and then clicking the option for 20?
Nothing wrong, and nothing urgent at all; I've just never seen this before and I was curious. Nyttend (talk) 20:16, 23 March 2026 (UTC)
- (1) Probably the sheer number of log entries Materialscientist has performed is making the server unhappy. Anyway it worked for me. See also phab:T325062. (2) Manualy specify a limit param in the URL: https://en.wikipedia.org/wiki/Special:Log/Materialscientist?limit=20. * Pppery * it has begun... 21:36, 23 March 2026 (UTC)
Recently (past few weeks) a larger than normal percentage of 429 rejections from the WMF API:
HTTP Error: 429 - Your bot is making too many requests. Please reduce your request rate or contact bot-traffic@wikimedia.org
IABot
IABot seems to have had a backend malfunction that has caused it to be effectively down for several days. Is there anyone who can restart it? -- Beland (talk) 02:23, 24 March 2026 (UTC)
- I think User:GreenC understands how that works. Johnuniq (talk) 03:06, 24 March 2026 (UTC)
Another odd error
I have this error:
Save failed: (blocked): Your username or IP address has been automatically blocked by MediaWiki. The reason given is: You are currently unable to edit Wikipedia. You are still able to view pages, but you are not currently able to edit, move, or create them. Editing without an account from $1 is disabled as it is a private IP range. This is probably a result of a problem with your Internet connection. You may be able to edit if you log in or create an account.
- Start of block: 00:55, 24 March 2026
- Expiration of block: no expiry set
- Intended blockee: 172.16.5.0 Your current IP address is 172.16.5.0. Please include all above details in any queries you make.
This can't be right; 172.16.x.x is the private IP range used for internal local networks. (In this case, probably Kubernetes.) Anybody know what is going on? Hawkeye7 (discuss) 03:43, 24 March 2026 (UTC)
- This seems to be the error you get if you a script running on a Toolforge K8s node is trying to edit Wikipedia while logged out. Most likely your bot got logged out and needs to log in again. * Pppery * it has begun... 04:06, 24 March 2026 (UTC)
Line spacing with syntaxhighlight
I am seeing some excessive vertical spacing when using syntaxhighlight. For example:
{{ABC}}— Martin (MSGJ · talk) 10:40, 24 March 2026 (UTC)
- There is an invisible character, left-to-right mark, in your comment before
<syntaxhighlight></syntaxhighlight>. You might accidentally copy it from somewhere. Since it is placed after a line break, it creates a separate parahraph. It is visible as a bullet if you enable syntax highlighting in the standard editor. Jack who built the house (talk) 15:49, 24 March 2026 (UTC)