Jump to content

Commons:Village pump/Technical

Add topic
From Wikimedia Commons, the free media repository
Latest comment: 3 minutes ago by Geagea in topic File:PION Labs logo.svg

Shortcuts: COM:VP/T • COM:VPT

Welcome to the Village pump technical section
Technical discussion
Village pump/Technical
 Bug reports
 Code review
Tools
 Tools/Directory
 Idea Lab



This page is used for technical questions relating to the tools, gadgets, or other technical issues about Commons; it is distinguished from the main Village pump, which handles community-wide discussion of all kinds. The page may also be used to advertise significant discussions taking place elsewhere, such as on the talk page of a Commons policy. Recent sections with no replies for 30 days and sections tagged with {{Section resolved|1=--~~~~}} may be archived; for old discussions, see the archives; recent archives: /Archive/2025/12 /Archive/2026/01.

Please note
 
SpBot archives all sections tagged with {{Section resolved|1=~~~~}} after 1 day and sections whose most recent comment is older than 30 days.

Category missing in all of a cat's subcats

[edit]

I noticed at Category:Electricity production trends in the United Kingdom that the category is missing the category Category:Charts of the United Kingdom.

The Electricity production trends categories don't have a template so all their categories are set not via template but separately on all the categories.

Is there a way to add [[:Category:Charts of ''name of country'']] (where name of country is simply whatever comes after "Electricity production trends in" in the category title)) to all of these categories without having to go through all the categories and adding it manually?

If not, is there a tool to which this functionality could possibly be added? Prototyperspective (talk) 11:41, 17 November 2025 (UTC)Reply

Another example is that I'd like to add the navbar that's included in Category:Videos of short films from 2024 to all subcats of Category:Videos of short films by year. The only thing that varies there for every 10 cats is |decade=202 and it would already suffice if one could add this for 10 cats at a time from the parent category page with an interface like Cat-a-lot. Prototyperspective (talk) 21:58, 7 December 2025 (UTC)Reply

Image thumbnail problem

[edit]

Could take anyone a look at File:Vittore Carpaccio - Young Knight in a Landscape - Google Art ProjectFXD.jpg. For some reason the image doesn't show up in any page, where it's used. Armbrust (talk) 17:07, 30 November 2025 (UTC)Reply

All i see in the logs i have access to is a bunch of "first byte timed out" errors. Would probably need to file a bug on phabricator to get it more investigated. If the file is a progressive jpeg, try uploading as a baseline jpeg. Bawolff (talk) 20:03, 30 November 2025 (UTC)Reply
Fixed by Yann (thanks!) with the following comment/explanation: c:User:Rillke/bigChunkedUpload.js: baseline instead of progressive. How to see more files affected by this issue so they can be fixed as well or alternatively is there a code issue about making also such progressive files show up properly? Prototyperspective (talk) 11:28, 3 December 2025 (UTC)Reply
Essentially the issue is that progressive jpeg take much more memory to resize, so large files are much more likely to time out. Bawolff (talk) 16:31, 3 December 2025 (UTC)Reply

Here is a list of some files with rendering failures according to the logs (Note that this is only a sample based on files that someone tried to look at at a specific point in time, so consider this list to be rather incomplete. Its also possible other unrelated errors could make a file show up on the list. For many of them, the failures are only at very large sizes, e.g. an 8000px thumbnail):

List of some files with rendering failures

Bawolff (talk) 18:52, 3 December 2025 (UTC)Reply

Thanks! I put the list into a collapsed template (mainly for those users who scroll through this page) – hope that's okay (if not just remove it).
Is there an issue on phabricator about this already? Especially when considering how many files are affected. And do you know of a way to bulk correct the files or to change the thumbnail code to make also those display? Maybe rerunning the code to resize them (not sure if you mean resizing the thumbnail or sth else) if it timed-out earlier? Prototyperspective (talk) 22:20, 3 December 2025 (UTC)Reply
Its somewhat expected that edge cases happen with thumbnail rendering. Some of these cases are only for creating 16000px wide thumbnails, in which case the response on phabricator would probably be - don't do that. In other cases it seems like a server overload happened. e.g. for the NBC logo that has been deleted for years. Someone requested that file like 500 times in a 1 minute interval, which seemed to trigger some sort of short term overload which got it on this list, but for our purposes that is a false positive. Anyways I don't think you'll get much response from phabricator unless there is a specific pattern that has been identified. Bawolff (talk) 22:32, 3 December 2025 (UTC)Reply
Its somewhat expected that edge cases happen with thumbnail rendering That it's expected doesn't mean it can't or shouldn't be fixed. I'm not saying it should necessarily be prioritized by whoever does it (and maybe a volunteer fixes this). Some of these cases are only for creating 16000px wide thumbnails That's something I don't understand based on the info so far: aren't thumbnails sized down to a size of an average around 300px? e.g. for the NBC logo that has been deleted for years It doesn't matter for deleted files. Anyways I don't think you'll get much response from phabricator That doesn't mean bugs shouldn't be filed there...e.g. when it's not normal anymore for bugs there to stay unfixed for 1+ decade but more development going on in a more positive future.
.
I don't know how you created this list of broken files. The main question basically remains of how to proceed to get these files fixed. For example, if you can create a list like that, couldn't that be fed into a script that corrects these? unless there is a specific pattern that has been identified another approach would be to retry failed thumbnail renderings every once in a while and with larger timeouts. Prototyperspective (talk) 22:42, 3 December 2025 (UTC)Reply
The list is generated by looking at the "varnishslow" log (in logstash) for entries that have a 503 status code and are for a thumbnail from commons [logstash is not available to the public, you have to ask to get access]. This is basically a list of every web request that took so long that a timeout occured. There is a lot of things that can cause a failure here. It could be a transient issue, it could be a broken file, it could be that the file is simply too big and a timeout is reached, it could be some other sort of bug. I haven't particularly investigated individual cases, to do something about them would require determining the cause, and its likely different files have different causes. Re That's something I don't understand based on the info so far: aren't thumbnails sized down to a size of an average around 300px - users can request bigger thumbnail sizes. I think there is some max where we refuse to serve them. If a user asks for a huge thumbnail and it doesn't work, that is arguably a non-issue. Users who need a file that large should just download the whole thing. Re That doesn't mean bugs shouldn't be filed there...e.g. when it's not normal anymore for bugs there to stay unfixed for 1+ decade but more development going on in a more positive future - I don't think anyone really has the goal of 0 thumbnail rendering failures. Some of these may be legitimately broken files where failure is the correct thing to do. I'm not saying don't file bugs at all, but some narrowing things down is probably a good idea. If the bug report is simply: something sometimes breaks for unclear reasons, its not a useful bug report. A future dev could just pull the same list I did if they are interested, by which time the files in question on the list would probably be entirely different. Bawolff (talk) 23:12, 3 December 2025 (UTC)Reply
Thanks! This helps the eventual solution – e.g. if somebody later gets back to this problem and would like to find out how the failing files were identified. to do something about them would require determining the cause I think that applies to just one of the two approaches of what could be done here: the solving of the cause problems. But the other approach would be to just fix the individual files, e.g. by retrying them with a larger timeout or something and/or by running some script to regenerate the thumbnails. I think the first approach or also doing the first approach would generally be preferred but one could also just fix those files like the individual file that Armbrust asked about was fixed but at scale for many files at once. If a user asks for a huge thumbnail and it doesn't work, that is arguably a non-issue Yes but not that the thumbnail is then broken also for users who request a smaller thumbnail in the category or search results views as is the case with the files in the list. Some of these may be legitimately broken files I didn't name 0 files as a goal or a near-term goal. The thing is that the preview on the file page often loads just fine, just not the thumbnail in the category view. If the bug report is simply: something sometimes breaks for unclear reasons, its not a useful bug report. Not as useful as it should be ideally but again: one could do what has been done with the single specific one file File:Vittore Carpaccio - Young Knight in a Landscape - Google Art ProjectFXD.jpg but for all or many files in that list. For things that change many files, usually scripts are used so it's a technical task and it would solve the bug of thumbnails missing. Prototyperspective (talk) 23:42, 3 December 2025 (UTC)Reply
Just to clarify, thumbnails (unless they already exist) are generated live. So there is no script to run to regenerate them. They are generated anytime someone views the image thumb's url. By the same token, this is where 8000px thumbnails can come from, the user just has to modify the url to change the size. Some files that are near the edge could in theory be fixed if the limits were higher, but limits are also there for a reason, to make sure no individual file hogs all the resources and prevent other files from rendering. Bawolff (talk) 23:48, 3 December 2025 (UTC)Reply
Ahh that's the key part I didn't know, thanks. I thought they were already generated at the upload and then sit on the server.
Wouldn't it make sense and improve performance and reduce server-load if thumbnails weren't re-regenerated every time one opens a Commons page at least for one or a few standard sizes?
(Nevertheless, while it would be done differently and may be more difficult, one could maybe have a scrip edit the files and upload new versions in a batch to fix a fraction of the affected files in the list.) Prototyperspective (talk) 23:53, 3 December 2025 (UTC)Reply
I think a couple common sizes are auto-generated. And recently they've become bucketed (i.e. If you're off from the common size by a little bit, it returns the common size instead of a brand new file). To be clear, this is only the first time its viewed. Once a thumbnail of a specific size is generated, its saved for later. Bawolff (talk) 23:57, 3 December 2025 (UTC)Reply
Some of these things just don't belong here. We are a pretty generic file host. There is no way we can account for anything and everything in the world. It's fine that a museum wants to host a 33000 x 66000 pixel progressive png in a pipeline they specifically designed for that to work and to facilitate archival quality. But we haven't done that extra work, so it will fail. This is expected. There are limits to everything. Just because someone can build the burj khalifa, doesn't mean you can place it on a lot designed for townhouses. —TheDJ (talkcontribs) 18:37, 4 December 2025 (UTC)Reply
I think we're talking past each other a bit. I was saying one should probably just generate and store a thumbnail of max 300 px size and then serve that. If there's some issues within the file, then the file can be edited. For example, exactly like it was done for File:Vittore Carpaccio - Young Knight in a Landscape - Google Art ProjectFXD.jpg except by some script for all the problematic files. I doubt not filing any code issue and just letting this info sit here and collect dust in the archives is a good approach even when this is not an important problem. Prototyperspective (talk) 18:54, 4 December 2025 (UTC)Reply
Is this even possible? AFAIK the PNG limit is at 65535 pixels. TIF may server larger pixel dimensions --PantheraLeo1359531 😺 (talk) 18:12, 8 December 2025 (UTC)Reply
serve* --PantheraLeo1359531 😺 (talk) 18:13, 8 December 2025 (UTC)Reply
I think you are mixing up jpg and png. Max dimensions of a png are 4294967295 x 4294967295 afaik. 65535 is the jpg limit. ("progressive" usually refers to jpgs, although pngs have a similar feature called adam7. I dont know if adam7 affects memory usage patterns when scaling in pngs the way progressive does in jpg. I think it doesn't but i dont know). Regardless, i think TheDJ's point is that we are optimizing for the average case. There are always going to be cases outside of that. The exact number so to speak is besides the point. Bawolff (talk) 19:05, 8 December 2025 (UTC)Reply

Sorting files in a category by the number of categories the files have?

[edit]

Is this possible somehow?

It would also be useful for Wikipedia articles.

Here, in the case of Category:NIH BioArt, I'd like to either filter for files with only the (currently) 2-3 meta-categories set or sort the files by number of categories to add categories to those files that miss them.

Most of these files don't have categories set and currently one can't filter away the ones that already got some. Categorization request here. Prototyperspective (talk) 00:47, 3 December 2025 (UTC)Reply

Does https://quarry.wmcloud.org/query/99622 work? Bawolff (talk) 02:48, 3 December 2025 (UTC)Reply
Very cool. I imagine this could be useful in many cases. Maybe a script/gadget could be built using this that enables seeing the results (1) from the click of the button on the category page (e.g. under Tools "List files by number of cats") and (2) in wikitext so you can click the file links. Currently, a main problem here is that the file links are not clickable.
Moreover, it would be better if not needed if the files had a preview/thumbnail. Do you think it would be possible to make this a native sorting feature? Because with native MediaSearch sorting one could search e.g. use deepcategory:Category:NIH BioArt and then just sort by number-of-categories.
A hacky way would be, if I'm not mistaken, to export it as wikitable and then using VsCodium or a text-editor like it to search and replace prepend [[File: and append |thumb]] and then paste it manually into some page on Commons. Prototyperspective (talk) 21:20, 4 December 2025 (UTC)Reply
Some sort of gadget or toolforge tool would be possible (To be clear, I'm not necessarily volunteering to make one). Its unlikely this would be built into mediawiki, as this category sorting method becomes slow for very large categories, which is a no-no for anything built into mediawiki. Bawolff (talk) 21:53, 4 December 2025 (UTC)Reply
For your hacky way - we can do that with quarry. Try this link https://quarry.wmcloud.org/run/1046668/output/0/wikitable which outputs a table you can paste into commons. Bawolff (talk) 22:16, 4 December 2025 (UTC)Reply
Thanks! I think it his can be quite useful in many occasions for dynamically improving categorization of some cat by users (works best with flat cats and is especially useful for categories with new files such as after batch uploads). Maybe it could also be useful for some report, e.g. for files uploaded as part of challenges like the Commons:Wiki Science Competition 2025 where good files often don't have categories other than the hidden meta categories set.
A problem is that one needs to put this on a Commons page and click preview to be able to use the results (being able to click the file links and see thumbnails if that's enabled). Also, currently one needs to fork the query and run it on Quarry which is impractical and inaccessible to most users.
If nobody here is interested in creating a tool for this, I'll probably make a wish in the m:Community Wishlist. It could be a gadget that adds an entry to the Tools menu like "List files by number of cats".
Query for the output https://quarry.wmcloud.org/query/99690
@Don-vip: By the way, you may be interested in that query – the example category is for a batch import you recently implemented (thanks again). Prototyperspective (talk) 19:20, 10 December 2025 (UTC)Reply
Nice! vip (talk) 19:42, 10 December 2025 (UTC)Reply

Need SQL query for geolocated images before date X

[edit]

Hello! I would like to create a current chart showing the number of geolocated images up to date X. I tried this at https://quarry.wmcloud.org/query/99798, but "WHERE img_timestamp < 20101206235959" is not recognized. How can this be fixed? Thanks --PantheraLeo1359531 😺 (talk) 13:05, 7 December 2025 (UTC)Reply

https://quarry.wmcloud.org/query/99805# is the basic way of how to do that, but i think a smarter method is needed as that query is very slow. Bawolff (talk) 19:02, 7 December 2025 (UTC)Reply
p.s. If an image has multiple sets of coordinates, its possible it might get counted twice. I'm not sure if its even possible to have multiple coordinates or how common that is. Bawolff (talk) 03:39, 8 December 2025 (UTC)Reply
@Bawolff: an image may easily have both {{Location}} and {{Object location}} templates with different coordinates. MKFI (talk) 09:48, 8 December 2025 (UTC)Reply
So we need a parameter that counts a single file as 1, not more? --PantheraLeo1359531 😺 (talk) 17:45, 8 December 2025 (UTC)Reply
I changed the query to do that, but that might make it slower so it might timeout. It might all be a moot point if no files have multiple coordinates. Bawolff (talk) 18:55, 8 December 2025 (UTC)Reply
Looks like it worked and did change the answer by quite a bit. Bawolff (talk) 19:15, 8 December 2025 (UTC)Reply
Thank you a lot :) --PantheraLeo1359531 😺 (talk) 19:28, 9 December 2025 (UTC)Reply
@Bawolff Following queries are faster during execution :) --PantheraLeo1359531 😺 (talk) 16:47, 10 December 2025 (UTC)Reply

Tech News: 2025-50

[edit]

MediaWiki message delivery 17:41, 8 December 2025 (UTC)Reply

Any way to identify files with closed captions missing the ClosedCaptions template?

[edit]

There are many files with TimedText subtitles but missing the {{Closed captions}} template (example: Special:Diff/1128180107).

The template is useful because it allows people to easily see that and which subtitles are available on thje file page and because it sets categories like Category:Files with closed captioning in English which can e.g. be used for filtering, searching, and glamorous usage scans etc.

Is there a way to identify these files and if possible also to add that template to these automatically? Prototyperspective (talk) 11:47, 9 December 2025 (UTC)Reply

Hello @Prototyperspective, I made this SQL query to look for files with TimeText pages, but without the {{Closed captions}} template. The query shows there are 7824 files with this case. Thanks. Tvpuppy (talk) 14:24, 10 December 2025 (UTC)Reply
Amazing, thank you very much! Do you know of a way to add the template to these files automatically? I think VisualFileChange could be used for that but it only works on categories, searches, etc and not Quarry outputs afaik. If not, it may be best to make a new thread at Bot work request so that a bot adds it based on the query export. Prototyperspective (talk) 14:53, 10 December 2025 (UTC)Reply
No, I don't. I would like to know if there's a way to do it automatically as well. Tvpuppy (talk) 15:26, 10 December 2025 (UTC)Reply

Enhance NoFoP-category with year the item becomes PD

[edit]

There's a suggestion at Template talk:NoFoP-category#Year that the images become free suggesting that {{NoFoP-category}} include a field to store the year that the subject becomes Public Domain, after which the category can be removed or replaced, and any deleted items undeleted. Posting here for visibility, input is appreciated. -Consigned (talk) 10:22, 10 December 2025 (UTC)Reply

Abuse filter makes it impossible to start a DR for timed text

[edit]

Apparently an abuse filter makes it impossible to start a DR for timed text, because it won't allow anything but timed text on the page. I am trying to start a DR for TimedText:WIKITONGUES-_Aminah_Abba_speaking_Wolof.webm.en.srt; here is what would go on that page:
{{delete|reason={{U|Jumbo T}} says (at [https://commons.wikimedia.org/w/index.php?title=Commons:Help_desk&oldid=1128926632#Delete_incorrect_captions?] that the captions here are wrong and this should be deleted, but that for some reason they could not start the DR themself. I have no Wolof, so I have no independent opinion of whether that is correct. Jumbo T, would you please explain in more detail what is wrong here (entirely wrong? needs editing? etc.)|subpage=TimedText:WIKITONGUES- Aminah Abba speaking Wolof.webm.en.srt|year=2025|month=December|day=10}}
Jmabel ! talk 22:04, 10 December 2025 (UTC)Reply

I also have no Wolof, but I reckon you can agree it's clear that she's not telling people how to change the language on Google Sites (not least because of all the English words that you can hear her say that don't seem to appear in the captions). Jumbo T (talk) 00:17, 11 December 2025 (UTC)Reply
Given that and the fact that Hellomyfriends1222, who created this file, looks to have been at least a borderline vandal, I will simply delete the file, but the technical issue remains: it is currently impossible to start a DR for timed text. - Jmabel ! talk 01:16, 11 December 2025 (UTC)Reply
Thanks JMabel. BTW, the Russian language subs at TimedText:WIKITONGUES-_Aminah_Abba_speaking_Wolof.webm.ru.srt are just a translation of the incorrect English subs into Russian and should be deleted too. Jumbo T (talk) 16:29, 11 December 2025 (UTC)Reply
I believe the same is true of some other special types of pages like translations and Data: pages. It might be nice if scripts like QuickDelete could be modified to allow nominations without a {{Delete}} template on the page when it's technically impossible to insert. Omphalographer (talk) 20:06, 11 December 2025 (UTC)Reply

Where do I put the GraphViz code?

[edit]

For the images generated with (my own) GraphViz code, where do I put the code? Example of my ugly first try: File:Deregulated electrical market - single purchaser configuration.svg. Викидим (talk) 21:37, 12 December 2025 (UTC)Reply

Unless someone can give you a better answer within a day or two, just put it on the file page inside a <code> element. At least it will be safe until someone works out a better way to do this. - Jmabel ! talk 23:50, 12 December 2025 (UTC)Reply

Receiving reply notification when subsequent DRs are added to daily page

[edit]

In the past few days I've noticed that after submitting a deletion request and having it posted to the daily page like Commons:Deletion requests/2025/12/14, I'm notified when new DRs are added below.

For example today I posted a DR at 10:52, and then I received a notification Yann replied in "December 14" when Yann added a new DR below it at 10:54, then I was notified again with Belbury replied in "December 14" when the next DR was added at 11:15. I first noticed this on Dec 10 and it seems to happen both after using "Nominate for deletion" link in the sidebar and also COM:VFC.

I don't think I changed my preferences recently, has the auto-subscribe logic changed? -Consigned (talk) 11:28, 14 December 2025 (UTC)Reply

Investigating a bit further - looks like posting a DR is automatically subscribing me to the == December 14 == heading at the very top of the page, under which each DR is added as a level 3 heading... -Consigned (talk) 11:42, 14 December 2025 (UTC)Reply

>Moving this to mediawikiwiki:Talk:Notifications#Automatically subscribed after adding to daily Commons DR page as it seems to relate to the Notifications tool rather than Commons functionality. Please respond there if you have any insight or if it also affects you. -Consigned (talk) 12:26, 14 December 2025 (UTC)Reply

 Comment, there's a related discussion at Commons:Village pump#Notifications. Thanks. Tvpuppy (talk) 17:05, 14 December 2025 (UTC)Reply
Is there a phabricator issue about this? Prototyperspective (talk) 16:21, 15 December 2025 (UTC)Reply

A solution has been implemented in phab:T412462. However, maintainers of AjaxQuickDelete will still need to update the gadget to use that solution to actually fix the immediate issue here. – Ammarpad (talk) 18:20, 16 December 2025 (UTC)Reply

I've added some notes, and a hesitant ping, at MediaWiki talk:Gadget-AjaxQuickDelete.js#Requested change to add discussiontoolsautosubscribe parameter, which will hopefully help resolve this. Quiddity (WMF) (talk) 21:08, 16 December 2025 (UTC)Reply
I encounter the issue even when I manually add a DR to the daily page like Commons:Deletion_requests/2025/12/17 (e.g. diff), following the steps at Commons:Deletion requests/Listing a request manually, not using VFC or "Nominate for deletion" button. I don't think that updating AjaxQuickDelete will solve the entire problem. -Consigned (talk) 09:45, 17 December 2025 (UTC)Reply

Love search result

[edit]
This is what pops up when you look up "love"

Why do images of Diddy appear when I look up "love" in Commons search??? Is this some kind of prank? – Howardcorn33 (💬) 13:15, 15 December 2025 (UTC)Reply

One of his stage names was "Love". Sjoerd de Bruin (talk) 13:28, 15 December 2025 (UTC)Reply
One of the reasons for why W471: When searching Commons, if there is a category with same or very similar title, show a hint/link (voting on it is open) would be useful (here to find relevant files especially when the search engine shows lots of unrelated files).
To answer your question: it seems to be because MediaSearch uses Wikidata a lot and specifically also the aliases defined there and one alias defined for Sean Combs (Q216936) is "Love". MediaSearch needs lots of improvements such as considering categorization much more than SD's Wikidata aliases if the latter is considered at all. Prototyperspective (talk) 15:30, 15 December 2025 (UTC)Reply
Ah, I see now. I hope the search function improves in this regard. – Howardcorn33 (💬) 15:38, 15 December 2025 (UTC)Reply

Tech News: 2025-51

[edit]

MediaWiki message delivery 18:59, 15 December 2025 (UTC)Reply

[edit]

Does this search query show all the categories with the legacy way to connect pages interwiki? I kind of doubt just this few categories have the dysfunctional legacy interwikilinks (they need to be moved to Wikidata). Prior discussion here by User:Immanuelle. Prototyperspective (talk) 16:38, 21 December 2025 (UTC)Reply

I did some testing just now with https://regex101.com/. You might want to replace [a-z][a-z] with another [a-z]{2,3} (not sure if that ever arises, but harmless), but at least according to that tester it should work pretty much as expected. Here was my test case, which covers a lot of ground:


[[en:whatever]]
[[de-au:Whatever it may be]]
[[de-au:Grün]]
[[ru-sum:что бы ни]]
[[en-us:Color]] [[en-gb:Colour]]

At least according to the tester, it matches all six as expected (including diacritic in Grün, non-Latin characters in Russian, and having the last two links on a single line).
@Prototyperspective: Offhand I'd have to guess something is wonky in using regexes in mediasearch; the same search string produces a LOT more results with Special:Search. - Jmabel ! talk 20:12, 21 December 2025 (UTC)Reply
Thanks. Even after making that change and using Special:Search instead (like so), it still only shows 27 results. Prototyperspective (talk) 21:04, 21 December 2025 (UTC)Reply
@Prototyperspective: Strange. If you eventually find things it missed, please let me know & I'll see what I didn't think of in designing the regex. - Jmabel ! talk 02:05, 22 December 2025 (UTC)Reply

Interface message for switching search type

[edit]

When you perform a search, the gray box at the top says Looking for Special:Search? (MediaWiki:Mediasearch-user-notice-title) Clicking on the link in that just opens a blank Special:Search page, though, removing your query. The "Switch to Special:Search" link in the upper right (MediaWiki:Mediasearch-switch-special-seach [sic]) works much better, preserving your query. Could we get the gray box link to do the same?

Cheers, Sdkbtalk 03:41, 22 December 2025 (UTC)Reply

For those looking for this (as I was), you can only see this message once, if you dismiss it, it's gone forever. —TheDJ (talkcontribs) 10:32, 22 December 2025 (UTC)Reply
The message is rendered without having access to the parameters. So fixing this requires filing a ticket in Phabricator. —TheDJ (talkcontribs) 10:32, 22 December 2025 (UTC)Reply
Done; see phab:T413392. Cheers, Sdkbtalk 19:30, 22 December 2025 (UTC)Reply

Sorting files in a category according to the first letter of a word: bot or RegEx

[edit]

The files are audio pronunciation files in Estonian. All the file names follow the pattern Et-<term>.ogg. They need to be categorized as [[Category:Estonian pronunciation|<first letter of the term>]] or [[Category:Estonian pronunciation|<term>]] (which I think is exactly the same). Since there are 6000 files, I would rather not click through 6000 links to do this manually. Some people have said that I should use RegEx, but I'm not very skilled in coding and I couldn't find anybody to do it for me either. But I was thinking that maybe this could be done by a bot, which would certainly make things easier. I haven't uploaded the files yet, I just want to know if that's possible and is anybody ready to do it. Joonas07 (talk) 17:39, 22 December 2025 (UTC)Reply

Yes I can do it easy in VisualFileChange  REAL 💬   17:59, 22 December 2025 (UTC)Reply
Does it require coding skills or no? In the sense that could I do it myself or better leave it to you? Joonas07 (talk) 20:57, 22 December 2025 (UTC)Reply
Not hard here is 1 way to do it you can see I did it here and subsequent edits

1st search

/\[\[[cC](?:ategory|at):[Ee]stonian[_ ]pronunciation\]\]/g

Replace

[[Category:Estonian pronunciation|%PAGENAME%]]

Next search

/\[\[Category:Estonian pronunciation\|Et\-/g

Final part not essential

/(\[\[Category:Estonian pronunciation.*)\.(?:og[ga]|mp3|wav|mid|midi|flac)/g
$1

 REAL 💬   00:00, 23 December 2025 (UTC)Reply

Thanks! Happy holidays! Joonas07 (talk) 10:02, 23 December 2025 (UTC)Reply

Tech News: 2025-52

[edit]

MediaWiki message delivery 21:42, 22 December 2025 (UTC)Reply

Scanned images with a front and back

[edit]

I’m currently uploading scans of old Carte de Visites. They have content on both sides. For each one, I am inclined to upload a side by side, and also each side individually - 3 files total - since there are different use cases. Is that correct? ~2025-42162-94 (talk) 08:28, 23 December 2025 (UTC)Reply

I'd upload the front and back as separate files, and use the other_versions field in the file description template to tie the two together, personally. The various projects have templates that can display two images side by side without the needf to have them as one file. The Squirrel Conspiracy (talk) 08:36, 23 December 2025 (UTC)Reply
That makes a lot of sense, thank you! I’ll try what you suggested. The Edinburgh Early Photography Archive (talk) 08:48, 23 December 2025 (UTC)Reply
@~2025-42162-94 and The Edinburgh Early Photography Archive: see also File:Bagley Hall, University of Washington, ca. 1910 - DPLA - 924e8711bb9227e3e04d29b7f683f919 (page 1).jpg for another good way of doing this. - Jmabel ! talk 19:30, 23 December 2025 (UTC)Reply
[edit]

When I add a Creator to Author, for example Creator:Samuel Alexander Walker, there's no automatic link anywhere in Creator to other works by him on Commons. I realise I can link manually to "Category:Samuel Alexander Walker". But I don't think the Category page auto-populates with works tagged with the Creator tag, and it can also contain lots of things that aren't by the Creator (eg portraits of the Creator), so it's not useful for the purpose of finding all (and only) works with a certain Creator tag.

Shouldn't there be some automatic page that displays all (and only) works tagged with a Creator tag, for example "Works:Samuel Alexander Walker"? Am I missing something? The Edinburgh Early Photography Archive (talk) 06:58, 24 December 2025 (UTC)Reply

@The Edinburgh Early Photography Archive: I don't think that would be at all easy to do automatically, especially because for a given author there may or may not be a subcategory along the lines of Category:Works by Samuel Alexander Walker, and if that category does exist it may or may not have subcategories of its own. (See Category:Works by Pablo Picasso for a fairly good example of how complicated that can get.)
So, in short: yes, things like that are good categorization; no, it is not readily automated. - Jmabel ! talk 20:01, 24 December 2025 (UTC)Reply

Public domain template

[edit]

Hi. I made a new public domain template specifically for Saint Helena, Ascension and Tristan da Cunha: {{PD-SHN}}. Please comment on it to see if I got the template code format correctly, thanks. – Howardcorn33 (💬) 22:53, 24 December 2025 (UTC)Reply

PD-Art vs PD-Scan tag for old photos

[edit]

For old photos in the public domain, where the digitiser asserts copyright (for example the National Portrait Gallery), and it is not clear if the digitisation of the old photo was done as a photograph or as a scan, is it more appropriate to use the PD-Art tag or the PD-Scan tag? The Edinburgh Early Photography Archive (talk) 23:45, 26 December 2025 (UTC)Reply

{{PD-scan}} is just a redirect to {{PD-art}}. - Jmabel ! talk 00:23, 27 December 2025 (UTC)Reply
I'm so sorry that I keep asking what are probably stupid questions! I thought PD-scan says "a mere mechanical scan or photocopy" whereas PD-art says "a faithful photographic reproduction". I was just wondering which one is more appropriate when it's not clear whether the digitisation was done via photo or scan. The Edinburgh Early Photography Archive (talk) 00:40, 27 December 2025 (UTC)Reply
@The Edinburgh Early Photography Archive: oops, my mistake: it's en-wiki that combined the two, Commons keeps them separate. But the former speaks volumes about how unimportant the distinction is. - Jmabel ! talk 17:46, 27 December 2025 (UTC)Reply

Discord Embeds Broken

[edit]

Most images sent to Discord from Wikimedia Commons do not load their preview images in the embed, ever since 2 or 3 days ago. Some do, my testing shows a trend that images featured on the current main page for Wikipedia tend to load, (sending myself https://commons.wikimedia.org/wiki/File:Neptune_Voyager2_color_calibrated.png on discord loads the preview image in the embed, for example) yet most others do not (sending myself the non-main page image https://commons.wikimedia.org/wiki/File:Neptune_-_August_19_1989_(39736246694).jpg refuses to load the preview image in the embed, as a counter-example), though it's not entirely consistent. Old Wikimedia Commons links sent in Discord have also had their previously loadable preview embeds fail to load upon cache refresh. Other websites like YouTube, (Fx)Twitter and Internet Archive still load embeds fine. I love sending images that I find in Wikimedia Commons to my friends a lot, so finding help would be important, though since my issue is technically based on another website / app, I'm not entirely sure if posting here is appropriate. Jd vance don't like me (talk) 02:58, 28 December 2025 (UTC)Reply

Probably relates to the recent limiting of what image sizes can be generated, same thing that messed up the TinEye and Google Lens tools. - Jmabel ! talk 07:39, 28 December 2025 (UTC)Reply
Do note that Discord has some custom mechanics for Wikipedia previews, not something we can easily fix from our side. Sjoerd de Bruin (talk) 09:46, 28 December 2025 (UTC)Reply
@Jmabel: Just for my reference, any links/announcements regarding image size generation limits? Haven't heard about that. ~Kevin Payravi (talk) 01:08, 5 January 2026 (UTC)Reply
Nothing that I'm aware of, but I gather this briefly broke the links to TinEye and Google Lens, because they were using a specific hard-coded size that is no longer supported. Can't recall where I heard that, sorry. - Jmabel ! talk 01:29, 5 January 2026 (UTC)Reply

How to specify a SVG file's default language?

[edit]

At File:Wissenschaftssprachen NatWi.svg how does one specify that default language there is German? It would be best if in the dropdown just had German and English and not (default language) and English. Prototyperspective (talk) 23:14, 3 January 2026 (UTC)Reply

It's not possible, without modifying the svg. A workaround here is to explicitly add German translations, the same way you added the English ones. (It may seems duplicative, since the default text is already German, but it would work.) If you do that, you'd get 3 options 'English - German - default' in the dropdown. – Ammarpad (talk) 18:22, 4 January 2026 (UTC)Reply
The semantics of SVG's systemLanguage attribute imply the default clause does not have a language. Consequently, a multilingual SVG file should have an explicit systemLanguage for each language and a default clause to cover all other languages. Otherwise, strange behavior happens. For example, if my browser prefers German over English, I expect to see German when displaying File:Wissenschaftssprachen NatWi.svg in my browser, but my browser will display English text. The browser does not know the default clause is German, so it displays the less-preferred (but known) English.
It would be nice if SVG Translate learned the default language and copied the default clause to an explicit language clause.
The simple fix is to use SVG Translate to translate the default language to German.
Glrx (talk) 19:42, 4 January 2026 (UTC)Reply

Chunked upload error

[edit]

Trying to use chunked upload to overwrite File:Port of Seattle map, 1918.jpg with an even higher-resolution file, I got an error that MwJSBot is not defined. File I was uploading shows in MS Windows "properties" as 100MB, but is apparently a tad too large for a non-chunked upload. - Jmabel ! talk 23:46, 4 January 2026 (UTC)Reply

Change in wiki commons rate-limiting? Getting frequent 429 - errors on wiki commons image requests

[edit]

I have an app (ios / android) that is displaying images from wikimedia commons (Hotlinking) that has been out for over a year. About a week ago I started seeing a massive increase of 429 - Too many Requests responses. Nothing on our side has changed, the requests come from user devices (and IPs) and the 429 responses come after about 10-20 requests (request volume hasn't changed either). This seems to be a problem for other projects as well as evidenced by this reddit thread and this bug report. I am sending a user agent with requests as recommended here. Bbbub (talk) 11:14, 5 January 2026 (UTC)Reply

File:PION Labs logo.svg

[edit]

There is some issue with File:PION Labs logo.svg. Don't know how to fix it. -- Geagea (talk) 11:25, 5 January 2026 (UTC)Reply