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/2024/12/Archive/2025/01.
Thank you! This solves it. I think there were also some templates that display image despite that the file is an audio but I can't find them anymore and think some of these change once structured data is added to the file which can take some time (but not all of them). So this may also be useful it various places. Prototyperspective (talk) 15:04, 2 December 2024 (UTC)[reply]
You can get the mime type via Lua reference manual - File metadata. But it doesn't seem that MediaWiki's internal concept of mediatype (image/video/application etc, that the search etc depend for categorising these) is exposed through that endpoint. Then again, this is a very internal definition and concept. —TheDJ (talk • contribs) 14:31, 5 December 2024 (UTC)[reply]
scale(<sx> [<sy>]), which specifies a scale operation by sx and sy. If <sy> is not provided, it is assumed to be equal to <sx>
Given that, the images are valid. Anyone has an authoritative SVG spec saying that different scaling factors for X and Y are invalid? Taylor 49 (talk) 23:31, 3 December 2024 (UTC)[reply]
The files are legitimate, so they should not need to be fixed. Fixing them is also problematic because the support for font-stretch="condensed" (the practical effect of the nonuniform scaling) is poor. Also, Commons does not support textLength.
There was an involved discussion about this problem a long time ago, but I could not find it. I also did not find something on Phabricator, so a Phabricator issue should be started. Someone can check whether this bug is already fixed in later versions of librsvg.
aufBuchen banner with the forest behind somewhat visible
When I upload and download a JPG file with the XMP.Attribution Name set to First Last <first.last@provider.tld>, Wikimedia Commons, for some reason, omits the space char between Last and < in the "When re-using this work, please credit" field in the Metadata section. This is the only name/email address combination that renders incorrectly in this section. So a bit odd really.
The given thumbnail provides an example. A small but not insignificant bug. I also searched the Village Pump archives but could not find any reference to this particular problem. Best, RobbieIanMorrison (talk) 21:22, 4 December 2024 (UTC)[reply]
I can confirm that this very minor bug exists. In the file in the field "author" there is ordinary space $20 followed by ordinary less-than-sign $3C. However in the field "please credit" there is ordinary space $20 folowed by a HTML-encoded less-than-sign. Taylor 49 (talk) 21:58, 4 December 2024 (UTC)[reply]
it's probably because it can be html, and it does some sort of normalisation on it. These parts of the MediaWiki code tend to be rather obscure. —TheDJ (talk • contribs) 14:13, 5 December 2024 (UTC)[reply]
Speaking as a programmer, it is usually worth fixing bugs. An artifact that is brushed aside as trivial may well indicate a more significant problem. And I would also argue that image metadata should be faithfully replicated in all cases. RobbieIanMorrison (talk) 07:33, 7 December 2024 (UTC)[reply]
I'm trying to help UN FAO upload some maps to Commons but I'm getting an error I've never seen before, the error happens when I try to upload the SVG files, does anyone know what the issue is and can someone suggest steps to fix them? I'm not very technical so a vague answer isn't really going to help a lot :)
The error message is maybe 20 pages long so I'll just copy the start of it in below.
A data URL allows arbitrary content. MediaWiki wants to limit that content, so it only accepts a limited number of data URL mimetypes. The mimetype is not specified in the data URL, so MediaWiki will reject the upload.
The element is probably trying to include a bitmap image. Inserting the appropriate mimetype between the colon and semicolon should fix the problem. Acceptable mimetypes are image/jpeg or /image/png.
Hi Glrx thanks very much for the technical explanation, unfortunately I'm far too much of a muggle to work out how to use it. Do you have a suggestion of which program to use and what buttons to press to resolve this? Thanks again, John Cummings (talk) 08:34, 6 December 2024 (UTC)[reply]
Use a text editor. Search for href="@data:. then add the appropriate mimetype after the colon and before the semicolon. Save the file. Then open the file in a browser to see that it displays correctly (Windows: locate file in the File explorer; then right click and open with Chrome/Edge/Firefox/whatever). If the file is good, then upload the file to Commons.
I'd like to sort audio files not by recency or relevance but by duration (maybe filesize would also work as a proxy for that). Is that possible somehow? I'd like to use it in a category of spoken Wikipedia audios for example to be able to choose between a very long or a shorter audio. It would probably be useful also for many other kinds of applications. Prototyperspective (talk) 16:39, 5 December 2024 (UTC)[reply]
Image duration is not an indexed field in the database I believe. It's only in the db's metadata structure of a file's record, which makes it difficult to query over a collection of files. —TheDJ (talk • contribs) 10:42, 6 December 2024 (UTC)[reply]
Thanks, that's helpful. However, that's just the filesize, not the duration and I'm looking for something that can be used in the UI (integrated into MediaSearch or SpecialSearch). Strange if it's not in the database since on the file page there is a field "Dimensions" which shows both the file-size and the duration (except for files with this bug). Prototyperspective (talk) 11:11, 6 December 2024 (UTC)[reply]
The file page is just one page, it can descend into the metadata relatively cheaply. Sorting is a 140+ million row operation and for that we need to pre-extract the information and put it into a separate column of the database and keep an index. We do this for width, height and file size, but not for duration. —TheDJ (talk • contribs) 15:54, 10 December 2024 (UTC)[reply]
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
Technical documentation contributors can find updated resources, and new ways to connect with each other and the Wikimedia Technical Documentation Team, at the Documentation hub on MediaWiki.org. This page links to: resources for writing and improving documentation, a new #wikimedia-techdocs IRC channel on libera.chat, a listing of past and upcoming documentation events, and ways to request a documentation consultation or review. If you have any feedback or ideas for improvements to the documentation ecosystem, please contact the Technical Documentation Team.
Updates for editors
Layout change for the Edit Check feature
Later this week, Edit Check will be relocated to a sidebar on desktop. Edit check is the feature for new editors to help them follow policies and guidelines. This layout change creates space to present people with new Checks that appear while they are typing. The initial results show newcomers encountering Edit Check are 2.2 times more likely to publish a new content edit that includes a reference and is not reverted.
The Chart extension, which enables editors to create data visualizations, was successfully made available on MediaWiki.org and three pilot wikis (Italian, Swedish, and Hebrew Wikipedias). You can see a working examples on Testwiki and read the November project update for more details.
Translators in wikis where the mobile experience of Content Translation is available, can now discover articles in Wikiproject campaigns of their interest from the "All collection" category in the articles suggestion feature. Wikiproject Campaign organizers can use this feature, to help translators to discover articles of interest, by adding the <page-collection> </page-collection> tag to their campaign article list page on Meta-wiki. This will make those articles discoverable in the Content Translation tool. For more detailed information on how to use the tool and tag, please refer to the step-by-step guide. [1]
The Nuke feature, which enables administrators to mass delete pages, now has a multiselect filter for namespace selection. This enables users to select multiple specific namespaces, instead of only one or all, when fetching pages for deletion.
The Nuke feature also now provides links to the userpage of the user whose pages were deleted, and to the pages which were not selected for deletion, after page deletions are queued. This enables easier follow-up admin-actions. Thanks to Chlod and the Moderator Tools team for both of these improvements. [2]
The Editing Team is working on making it easier to populate citations from archive.org using the Citoid tool, the auto-filled citation generator. They are asking communities to add two parameters preemptively, archiveUrl and archiveDate, within the TemplateData for each citation template using Citoid. You can see an example of a change in a template, and a list of all relevant templates. [3]
Last week, all wikis had problems serving pages to logged-in users and some logged-out users for 30–45 minutes. This was caused by a database problem, and investigation is ongoing. [5]
View all 19 community-submitted tasks that were resolved last week. For example, a bug in the Add Link feature has been fixed. Previously, the list of sections which are excluded from Add Link was partially ignored in certain cases. [6][7]
Updates for technical contributors
Codex, the design system for Wikimedia, now has an early-stage implementation in PHP. It is available for general use in MediaWiki extensions and Toolforge apps through Composer, with use in MediaWiki core coming soon. More information is available in the documentation. Thanks to Doğu for the inspiration and many contributions to the library. [8]
Wikimedia REST API users, such as bot operators and tool maintainers, may be affected by ongoing upgrades. On December 4, the MediaWiki Interfaces team began rerouting page/revision metadata and rendered HTML content endpoints on testwiki from RESTbase to comparable MediaWiki REST API endpoints. The team encourages active users of these endpoints to verify their tool's behavior on testwiki and raise any concerns on the related Phabricator ticket before the end of the year, as they intend to roll out the same change across all Wikimedia projects in early January. These changes are part of the work to replace the outdated RESTBase system.
The 2024 Developer Satisfaction Survey is seeking the opinions of the Wikimedia developer community. Please take the survey if you have any role in developing software for the Wikimedia ecosystem. The survey is open until 3 January 2025, and has an associated privacy statement.
I've slowly been requesting overwriting of various files from the IWM. They've been so gracious in correcting exposure on some of those (as I write this, I begin to wonder whether the modifications are enough to warrant copyright protection...), but there being multiple versions of a file seems to confuse Dezoomify, who only pulls the second version with lighting issues. With the help of ChatGPT, I could extract the exposure-corrected one, albeit at a reduced-resolution (1000x1300, compared to the 2400x3000 unedited) version of the edited image. Is there something I'm doing wrong in terms of extracting the edited image in all of its glorious pixels?
Hi. If Dezoomify does not yield the desired result, then we have other methods ... First, look at the code for the page. On the page mentioned in IWM, press Shift + Ctrl + i (works in Firefox) to see the code window. In the code window, you should use the "Network" tab. Then click on the desired picture (the one to the left?). You will see three lines of code, one with the "info.json" file that is used by Dezoomify, and two with jpg files that result in thumbs of the picture. If you click on the bottom jpg codeline, you will see this code: https://media.iwm.org.uk/cantaloupe/iiif/2/171%7C%7C950%7C%7Csuper_000000.jpg/full/251,/0/default.jpg
Now substitute "251" with "2400" and Voila! You get a picture in 2400 x 3111 (at least that is how it worked for me). You might try with larger values than "2400", but the system at IWM would block that option. Hope this helps. Cheers Rsteen (talk) 03:27, 11 December 2024 (UTC)[reply]
Great, I can do that going forward and many thanks! Often images that have been retouched aren't accessible beyond the 2400px, or so I've unreliably figured out after further digging in the source code. JayCubby (talk) 03:53, 11 December 2024 (UTC)[reply]
Interesting. I use the Dezoomify extension on Chrome, and I only get 1536 x 1990 pixels. Please note that this may not be in the public domain. They say "Press agency", not a government picture. Yann (talk) 09:59, 11 December 2024 (UTC)[reply]
Upload shown as "completed" in OpenRefine, but no upload has been made to Wikimedia Commons
Problem: I have prepared and reconciled all relevant metadata, filenames and filepath in OpenRefine for a test batch of 5 images, including schema preparation with "add media" to Wikimedia Commons. When I did a first technical trial run of the pipeline a month ago, my images successfully uploaded to Wikimedia Commons even though the metadata was far from perfect at the time. Now that I have worked on the data transformation, following the Wikimedia best practices as closely as I could, I have the same thing happening to me as this user here: https://forum.openrefine.org/t/no-errors-reported-and-the-percentage-completed-went-through-but-no-upload-done/899. I get a "100% percent completed" notification from OpenRefine but no uploads have been made (I checked under my user profile as well as searched for the file names).
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
Interested in improving event management on your home wiki? The CampaignEvents extension offers organizers features like event registration management, event/wikiproject promotion, finding potential participants, and more - all directly on-wiki. If you are an organizer or think your community would benefit from this extension, start a discussion to enable it on your wiki today. To learn more about how to enable this extension on your wiki, visit the deployment status page.
Updates for editors
Users of the iOS Wikipedia App in Italy and Mexico on the Italian, Spanish, and English Wikipedias, can see a personalized Year in Review with insights based on their reading and editing history.
Users of the Android Wikipedia App in Sub-Saharan Africa and South Asia can see the new Rabbit Holes feature. This feature shows a suggested search term in the Search bar based on the current article being viewed, and a suggested reading list generated from the user’s last two visited articles.
The global reminder bot is now active and running on nearly 800 wikis. This service reminds most users holding temporary rights when they are about to expire, so that they can renew should they want to. See the technical details page for more information.
The next issue of Tech News will be sent out on 13 January 2025 because of the end of year holidays. Thank you to all of the translators, and people who submitted content or feedback, this year.
View all 27 community-submitted tasks that were resolved last week. For example, a bug was fixed in the Android Wikipedia App which had caused translatable SVG images to show the wrong language when they were tapped.
Updates for technical contributors
There is no new MediaWiki version next week. The next deployments will start on 14 January. [10]
Validate offline and re upload. Or if they are valid and can be identified easily (by prefix for instance) it might be worth it to run a server side script, but this requires a Phabricator ticket. The cause might be that there is no XML DOCtype declared for these files. —TheDJ (talk • contribs) 12:21, 17 December 2024 (UTC)[reply]
Being more concrete with two example files.
File:Dam.svg (27 March 2015) does not have an XML processing instruction.
Rebecca Moore (en:Rebecca Moore (scientist)) is a friend of mine. When I told her what her picture was used in Wikipedia (because she never looked at her own page - haha), she was appreciative of the guy who uploaded it, but wished he had uploaded one she felt better represented her and her work. (Yes, the Rachel Carson Award was a great honor, but she thinks the work in Brazil with helping Chief Almir and the Surui tribe save their rainforest was more meaningful.) She gave me the original photo that they used in Time Magazine article April 15, 2016,(https://time.com/4295351/rebecca-moore-google-earth-outreach/) to replace it. I cropped the original 1600 x 1062 to be square, but then realized I couldn't upload it. Can someone please do that for me if I get you the photo? And how would I do that? My real email is martha.legare@gmail.com. Thanks much for your help, Martha MarleeRobertson (talk) 05:58, 19 December 2024 (UTC)[reply]
@MarleeRobertson: Hi, We need a written permission from the photographer, unless he/she uploads it her/himself. And it would be much better to have the original image first, it can always be cropped later. Please see COM:VRT for the procedure. Yann (talk) 09:41, 19 December 2024 (UTC)[reply]
Thanks. To make sure I'm clear: we (and thank you!) can replace the current photo as long as we get the Brazilian photographer's permission, not the guy who posted the current photo? The photo I have was taken for Google - I'm 99% sure they paid her for it. (I think the photographer was from Brazil - will check with Rebecca.) I appreciate your help. MarleeRobertson (talk) 04:09, 20 December 2024 (UTC)[reply]
This section is resolved and can be archived. If you disagree, replace this template with your comment. --廣九直通車 (talk) 09:26, 2 January 2025 (UTC)
This image, File:Papua New Guinea relief map.svg, is failing to load globally, and I don't see any indication as to why. The image hasn't been modified in a few years, and the image loads from the direct URL ([11]) and from the link on the description page, but doesn't load when it's placed inline. Is this some kind of database error? Ivanvector (talk) 13:55, 19 December 2024 (UTC)[reply]
I imported {{Q+}} and Template:Q+/doc from this project into Wikispecies, but the latter is behaving very oddly, and the former not working as expected, and I cannot figure out why. Wikispecies is a small project with few technical editors; can someone assist there, please? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits16:07, 19 December 2024 (UTC)[reply]
Hi! It seems that there is a problem during File-uploading here on commons. The page was created for my picture, all descriptions and categories can be seen, but the picture itself is missing. Instead I get following message: "Die Datei „mwstore://local-multiwrite/local-public/1/15/Bargischow_Kirche_lub_2024-07-27_img01.jpg“ befindet sich, innerhalb des internen Speicher-Backends, in einem inkonsistenten Zustand." Is this a known issue, is the issue being resolved, and is there a chance I can add my images to the page without having to go through all the work of describing and categorizing again? Thanks in advance Lukas Beck (talk) 20:03, 23 December 2024 (UTC)[reply]
Hi, I got an error while uploading a big video with Chunked Uploads, and many errors while deleting files. It seems there is some issue on the servers today. Yann (talk) 21:22, 23 December 2024 (UTC)[reply]
I looked at the cat again just now. The Alaska subcat was the only state subcat present, and that's been the case for several days now. For 2010 alone, there also exists 25 other such subcats (California, Colorado, Connecticut, Florida, Georgia, Illinois, Indiana, Iowa, Louisiana, Maryland, Massachusetts, Michigan, Minnesota, Mississippi, Nevada, New Jersey, New York, Ohio, Oregon, Pennsylvania, Rhode Island, Tennessee, Texas, Virginia, Wisconsin). The majority of those subcats had the template added to it within days after it was created in 2021. I don't know if they've been broken for 3+ years or if this is a recent thing. What I do know: I've been clearing my caches as I go along, which hasn't changed anything. I waited several days before originally reporting this in case it was an issue with a database catching up. I tried the null edit and it didn't appear to do anything. The examples I offered don't exist in a vacuum and shouldn't be treated as such. It's clear something is wrong and it appears rooted in the template. There's scores more states, scores more years, and other templates primarily designed to populate cats which aren't doing so. Another example is {{Earthquakes in Alaska}}. I made a simple edit to adjust its position on the tree, but those cats aren't being populated as such. In the future, if the templates' creators and maintainers aren't willing to address the problem, I'll simply comment them out and return to populating the cats by hand until something does change. RadioKAOS / Talk to me, Billy / Transmissions 21:22, 27 December 2024 (UTC)[reply]
I tested just now doing a null edit (an edit without changing anything) on Category:2009 in California by city and it got added to Category:2009 in the United States by city by state. The link gets placed by the California category, not the United States category. I think you may have done a null edit on the United states category, which yes, does nothing. Null edit the subcategory, not the parent. Also there are two caches here, one on your side, and one on the server side. If you do not see California in the United States Category, then clear your client cache (it might be ctrl+F5, maybe something else, it depends on your browser and operating system).
I then procceded with the other subcats you mentioned, the ones that exist, and now the United States category is at 24 entries. I did not find 4 entries in your list, they showed up as redlinks. If you have more of those, ask a bot operator to help you.
Waiting for this is going to take ages. I tested this a short while ago, so I know for sure. What happens is that every 30 days - when the server side cache expires, 30 items get added to the category. Also, the link on the subcategory gets added first and then it gets added to the parent a month later. So, if i did not null edit Category:2009 in California by city it would have gotten a link to the Category:2009 in the United States by city by state as the template was added, but California would not be in the United states category until a month later.
There is absolutely nothing the template can do here, this is a problem with the underlying software, MediaWiki, not the template. File a report on phab:, blaming the template is not going to change anything. I do not care if you will start adding the wikicode yourself without the template. Snævar (talk) 04:55, 28 December 2024 (UTC)[reply]
When you look at Category:Amastra, you'll find categories of which the name has changed by a vandal, e.g.
Ancient Amastra should be Amastra antiqua
Amastra Abbey should be Amastra badia
Amastra Baldwiniana should be Amastra baldwiniana
Amastra horny should be Amastra cornea
Hulled Amastra should be Amastra decorticata
Amastra fossil should be Amastra fossilis
Fraternal Amastra should be Amastra fraterna
Amastra Grayana should be Amastra grayana
Amastra Johnsoni should be Amastra johnsoni
Amastra Lahaina should be Amastra lahainana
and many more. When I press F5 on my keyboard, the proper name appears for a second or two and then changes again into the vandalised names. I have tried to find a solution for this, but I'm at a loss. Can someone put this right ? JoJan (talk) 10:16, 24 December 2024 (UTC)[reply]
I show you a partial clipboard of what I see on my computer (the arrows are mine): Partial clipboard of the Category:Amastra. Perhaps this will help solving this problem. JoJan (talk) 14:15, 25 December 2024 (UTC)[reply]
@JoJan It looks like you have Google Translate attempting to translate these. Try clicking the icon (next to the bookmark star in the address bar), then select Italian (or whatever it detects the source language as). —Mdaniels5757 (talk • contribs) 16:46, 25 December 2024 (UTC)[reply]
Hi!
It is common to smooth edges in raster images to make them look more aesthetic to the eye. But sometimes, it is desired not to apply this anti aliasing (for example for small 8-bit graphics). Otherwise it looks blurred. Is there a way to change the display style?
A pixelated arrow looks blurred, which is undesired in this case.
Yeah, I thought about it, too, but I guess there are several (unexperienced) users, who may not take the SVG conversion into account (and batch converting may be complicated, too)... --PantheraLeo1359531 😺 (talk) 14:51, 26 December 2024 (UTC)[reply]
EXIF data rendering omits space chars in some contexts
A recent example of faulty metadata rendering, click thru to view and search on "LNG boom"
I raised a similar issue previously and was told the code involved is intricate and quite possibly quirky and that the problem I described is relatively minor — in short, creating a fix is not especially likely. That earlier example involved ␣< sequences in email addresses.
For context, I specify my JPG metadata using YAML — a language with its own idiosyncrasies — and write that information to file using Python and ExifTool.
My latest upload (see thumbnail) used single quotes preceded (or followed) by spaces in the EXIF::Image Description metadata field, namely ␣'. This sequence was rendered faithfully to my terminal by Python and reported correctly on upload and download by both ExifTool and the gThumb image viewer. So I assume that the metadata is stored correctly in the JPG file.
But Wikimedia omits the leading (or trailing) space char in that sequence in the Description field and in the Metadata table. I duly edited the Description field to correct the corrupted string. But the faulty Metadata table rendering cannot be so easily fixed by users.
This presumed bug might not seem important but I can imagine use‑cases where confusion could result. In any case, I am raising the matter in the hope that a fix might be given some increased priority. Best, RobbieIanMorrison (talk) 22:46, 25 December 2024 (UTC)[reply]
Yes, it is another of those posts about GFDL :-) (the other were on main VP)
But the new is that
There should be no more disclaimers added to any new uploads and all template with disclaimers are now named so it is clear they have disclaimers. See w:en:Wikipedia:GFDL standardization.
Per December 27, 2024 there are estimated to be 153,000 files liecensed GFDL outside Commons. Please note that some license templates categorize files in more than one category so some files are counted more than once (mostly screenshots).
Below are categories with more than 1,000 files licensed GFDL as an example.
I could use some help to find the number of files that can't be moved to Commons because the files are not dual-licensed and they are uploaded after 15 October 2018. (They should go in the column "Date issue").
If anyone can think of a good way to automatize the count it would be great. Otherwise I can only think of checking the files manually (if there are only a few files) or with for example https://petscan.wmcloud.org/.
How to open x recent files uploaded by a given user in new tabs at once? It takes very long to click all of them. Alternatively, a way to replace a category based on the file title would be great as well but it probably doesn't work as well. Here's what I intend to do:
going through the recent uploads of Florin Talasman to replace category "Agricultural products" with whatever product the map is about which varies per file (e.g. kiwis).
For example, maybe it's possible to output files uploaded by a user with Wikimedia Commons Query Service in a column with the URLs, then select all those URLs (maybe copy pasting it somewhere) and drag them into the Firefox tabs to open them all in new tabs. That didn't work because I could only query for files created by the user but not files uploaded by that user (see also Make it possible to search by page author /contributor/ uploader).
Thank you! That's very helpful and I guess solves this issue. It should probably be in some relevant Help page so people looking for this functionality can find it. I think one can't filter files to only copy files that are videos of a specified resolution that way (which would be a workaround to solve phab:T377606) so a way to search files uploaded by a given user would still be useful. Prototyperspective (talk) 22:47, 31 December 2024 (UTC)[reply]
I tried write one but Special:ListFiles does not show the resolution directly, this is very slow because it has to load video metadata to fetch the width and height
(function(){functiongetVideoDimensions(videoUrl){returnnewPromise((resolve,reject)=>{constvideoElement=document.createElement('video');videoElement.src=videoUrl;videoElement.addEventListener('loadedmetadata',()=>{constwidth=videoElement.videoWidth;constheight=videoElement.videoHeight;resolve({width,height});});videoElement.addEventListener('error',(e)=>{reject(`Error loading video: ${e.message}`);});videoElement.load();});}constuploads=document.querySelectorAll('.TablePager_col_img_name a:nth-child(2)');consttoCopy=[];constpromises=[];for(constdofuploads){consturl=d.href;if(url.endsWith('.mpg')||url.endsWith('.mpeg')||url.endsWith('.ogv')||url.endsWith('.webm')){constpromise=getVideoDimensions(url).then(dimensions=>{constminimumWidth=1000;constminimumHeight=720;if((dimensions.width<minimumWidth)||(dimensions.height<minimumHeight)){toCopy.push(url);}}).catch(error=>{console.error(error);});promises.push(promise);}}Promise.all(promises).then(()=>{if(toCopy.length===0){alert('Nothing to copy');}else{navigator.clipboard.writeText(toCopy.join('\n')).then(()=>{alert('Copied '+toCopy.length+' items');}).catch(err=>{console.error('Failed to copy: ',err);});}});})();
Then I found VisualFileChange interface actually shows video dimensions, this is much faster
(function(){consttoCopy=[];consttitles=document.getElementsByClassName('jFileTitle');constdimensions=document.getElementsByClassName('jFileSize');constminimumWidth=1000;constminimumHeight=1000;for(leti=0;i<titles.length;i++){consturl=titles[i].href;if(url.endsWith('.mpg')||url.endsWith('.mpeg')||url.endsWith('.ogv')||url.endsWith('.webm')){constds=dimensions[i].textContent.replace(/^.*KiB/,'').replace('px','');constwidth=ds.replace(/^.* x /,'');constheight=ds.replace(/^ x .*/,'');if((minimumHeight>parseInt(height.replaceAll(' ','')))||(minimumWidth>parseInt(width.replaceAll(' ',''))))toCopy.push(url);}}if(toCopy.length===0){alert('No videos with width below '+minimumWidth+' or height below '+minimumHeight+' found on this page');}else{navigator.clipboard.writeText(toCopy.join('\n')).then(()=>{alert('Copied '+toCopy.length+' items');}).catch(err=>{console.error('Failed to copy: ',err);});}})();
@Prototyperspective for this specific case, since what you want can be extracted from the title, you could use com:vfc to batch replace "Category:Agricultural products" with "Category:{{subst:#invoke:String|sub|s={{subst:PAGENAME}}|j=-22}}s", and then fix any possible errors. RoyZuo (talk) 08:30, 1 January 2025 (UTC)[reply]
Requesting features on colour blindness acessibility templates
Hello, I have recently been working on adapting maps and other files to be readable by colour blind users. I'd like to add some (in my opinion, non-controversial) features to a few templates used in this area, but my wikitext comprehension isn't good enough for this. Any help would be greatly appreciated.
There are many image sets that are completely reliant on colour information for highlights, for example in math-related categories (see Category:Planar circle packings; red and blue for an image set without issues). Considering the nature of image sets in Commons, it would be better and easier to maintain if we could add the Colour blind accessible template to image set categories instead of each member individually.
I have managed to add such a parameter to the template above, adding anything to |set= adjusts the text on the notice box. However I haven't learned conventions for templates localised through autotranslate, and I suspect my implementation breaks good practice by mixing up parameters on the main template with parameters on the subpage ({{Colour blind accessible/en}}).
I'd like to request a once-over on the template from someone more experienced than me, as well as a switch that changes the category inserted by that template from Category:Images accessible to colour blind people to Category:Image sets accessible to colour blind people. Similarly, I'd like to request a feature for using the Colour blind template in categories for image sets, instead of just file pages. I'd be really thankful if anyone could help with this!
P.S. that the category tree for the cats used by these two templates is a bit messy right now because lots of files on it had the category added manually instead of through the template, I'm working on fixing them. Croomfolk (talk) 23:56, 28 December 2024 (UTC)[reply]
Umlauts should be sorted to display at the expected place
No. Category sorting in mediawiki is made with collations. Collations exist for individual languages, but generally not a group of them like latin languages. German Wikipedia can have category sorting designed for German, but there is not a solution for a multilingual wiki like Commons, Wikidata and Meta. You would also get into issues like how to categorize french specific letters next to german ones, in all of the over 100 languages wikimedia commons services with their images. While having a english focused category sorting collation seems very english specific, it really is not, because english has a lot of the basic letters that a lot of latin languages use. The main defect of using english sorting is that it is bad for asian languages. Snævar (talk) 06:56, 31 December 2024 (UTC)[reply]
@Prototyperspective: In addition to what Snævar said, it's apparently close to impossible to make a global change to Commons' category sorting order at all. We asked for a different change to category sorting a few months ago and the WMF sysadmins haven't been able to work out a way to do it without a lot of disruption: phab:T362494. --bjh21 (talk) 09:55, 31 December 2024 (UTC)[reply]
No. As I understand it, the problem isn't really the number of categories, it's the number of category links. That is, the number of connections between categories and members. So an empty category with one parent costs about the same as a file that's in one category. Templates that put millions of files into categories have a much more significant effect. --bjh21 (talk) 11:55, 31 December 2024 (UTC)[reply]
In that case a script that changes the sorting of cats starting with ö to o may be useful. how to categorize french specific letters next to german ones right next to them like the umlauts. Prototyperspective (talk) 10:59, 31 December 2024 (UTC)[reply]
I think the internationally/globally reasonable and expectable sorting would be sorting ö after o and è, é, ê, ë after e. That may or may not be the sorting commonly used in English but it doesn't matter much. It doesn't seem like an important issue however. Prototyperspective (talk) 12:38, 31 December 2024 (UTC)[reply]
I wonder if we can borrow some code from English Wiktionary. Is this something you want to replicate? English Wiktionary is multilingual in the sense that they have thousands of German page titles, Spanish page titles, etc. I believe they have a (semi-)automated system based on Lua modules that doesn't need DEFAULTSORT manually written on every page. You do need to specify the language per page through a template, though. whym (talk) 12:54, 31 December 2024 (UTC)[reply]
Interesting, there it's sorted well. Yes, replicating that would be great...above I suggested sorting umlauts separately after the letters but sorting them to be within them may be better (so ö is sorted as if it was o and é as if it was an e). I checked some of the pages and they didn't specify any sorting so it must be implemented differently if I'm not mistaken and I guess one could ask somewhere on wiktionary how it's been done. Prototyperspective (talk) 13:16, 31 December 2024 (UTC)[reply]
About half an hour ago while attempting to upload files, I started getting the "Wikimedia Error" page partway through, with "Request from [my IP] via cp1110 cp1110, Varnish XID 963178924
Error: 503, Backend fetch failed at [date and time]" at the bottom. This happens sometimes, usually due to my connection being weird, so I refreshed and the upload went through. But it's been getting steadily worse and now I basically can't upload at all, while the rest of my connection seems just fine. Something messed up on the backend maybe? The Bushranger (talk) 00:08, 1 January 2025 (UTC)[reply]
This section is resolved and can be archived. If you disagree, replace this template with your comment. Prototyperspective (talk) 12:59, 2 January 2025 (UTC)
View all 19 community-submitted tasks that were resolved last week. For example, a bug in the Add Link feature has been fixed. Previously, the list of sections which are excluded from Add Link was partially ignored in certain cases. [6][7]
This section is resolved and can be archived. If you disagree, replace this template with your comment. --廣九直通車 (talk) 09:26, 2 January 2025 (UTC) 
A pixelated arrow looks blurred, which is undesired in this case.
_Pudong_Yanggao_W0K-114_%E6%B2%AAA31192D_581_2024-12-08.jpg){kind=link}
![[11]](https://upload.wikimedia.org/wikipedia/commons/d/de/Papua_New_Guinea_relief_map.svg){kind=link}
_(Feb._2020).svg){kind=link}
