I left it open because it was the primary link in the Tech News announcement last week. Now that the week has passed, and no comments were left here, it can be resolved indeed. Thanks.

1. Summary of audit

@matmarex @Zabe As consultants/reviewers of the original implementation at T151682, do you have any insights to inform which choice we should make or otherwise have a preference for one approach vs another?

Priorities for me this week:

• T434469: Clientside SVG rendering not used when previewing during an edit, T419356: MathJax rendering missing after saving edit with VisualEditor, T422077: MathML rendering from DiscussionTools reply preview: These are likely all the same issue, just need to broaden MathJax init to use mw.hook('wikipage.content') instead of DOM-ready.
• T433928: mathmode= is bypassed by cache
• T433891: Improve Math preference labels
• T434686: Fix forcemathmode attribute

Re-tagging per mw:Maintainers. I originally filed it under Core, because we ran into these obstacles a few times when working on the CentralAuth extension, but it affects other teams' extensions and components like JobQueue as well. I was hoping to get to it as a side-project, but I haven't gotten around to that. Suggesting this as essential work for Ariel as a way to get familiar with the Config interface for extensions (and solve a big old DC hurdle). I can provide code review.

OK. I don't know what you're referring to in that case. Feel free to describe it in more detail in a new task.

@Ainali Can you share the following:

• a screenshot of the File description page on your wiki (before opening the Stockphoto gadget)
• the version of MediaWiki as displayed on Special:Version
• the version of the Stockphoto gadget as displayed in the line like "Revision: 2026-05-16" atop you copy of https://commons.wikimedia.org/wiki/MediaWiki:Gadget-Stockphoto.js
• right-click on the "Other resolution" links under the image preview and copy-paste the URLs here
• a screenshot of the Stockphoto gadget in a state where the problem is visible (ie showing a link to a non-standard size, which we know receives an error if you try to open or download it).

In T434357#12198955, @hashar wrote:

Thank you for having split the case of 3d2png to a standalone task (T434378)

The images now have python pointing to python3 which was worth adding. It resolves this task.

In T433891#12198333, @SalixAlba wrote:

Just one question, which one will be the default in the immediate future?

If we go with Blubber at T434378: CI broken for 3d2png.git, then this can be declined/reverted.

@DTotten-WMF wrote at Google-Wikimedia - Public meeting notes:

It seems like the new pages are getting indexed quite frequently I pulled a list of ~1400 URLs where the tag was recently removed and ran them through the indexing API last week. At the time there were only 10 that hadn't been indexed and 6 of those have been indexed since. but here are some examples of pages that are not.

https://en.wikipedia.org/wiki/220th_Brigade
https://en.wikipedia.org/wiki/Saline_District_Courthouse
https://en.wikipedia.org/wiki/2026%E2%80%9327_Dutch_Basketball_Cup
https://en.wikipedia.org/wiki/Tamara_Gabbe

@hashar wrote at change 1314972

There is some further gyp error:

Package x11 was not found in the pkg-config search path.
Perhaps you should add the directory containing `x11.pc'
to the PKG_CONFIG_PATH environment variable
Package 'x11', required by 'virtual:world', not found
Package 'xi', required by 'virtual:world', not found
Package 'xext', required by 'virtual:world', not found
gyp: Call to 'pkg-config --libs-only-L --libs-only-other x11 xi xext' returned exit status 1 while in angle/src/angle.gyp. while loading dependencies of binding.gyp while trying to load binding.gyp

Package x11 was not found in the pkg-config search path.
Perhaps you should add the directory containing `x11.pc'

That is provided by libx11-dev.

Then there is pkg-config --libs-only-L --libs-only-other x11 xi xext which would indicate libxi-dev and libxext-dev are also required.

Why is that repo requiring the X Window System client? Could it be a slop in the npm dependencies? We never had libx11-dev added in the CI image.

If we actually need those packages, I think we will want to craft a more specific image that ships the required libraries.

Between the WMF CI images for node18 and node20, we switched from Debian 11 Bullseye to Debian 12 Bookworm. This means the long-deprecated /usr/bin/python alias no longer exists, instead python2 or python3 must be used explicitly.

As mentioned in several places already, these proposals are inactionable. We haven't resorted to legacy UA-sniffing to determine payloads in ten years since T102318: Convert startup blacklist to feature test.

It seems we've a significant further increase to the memcached error as perceived by MediaWiki. Specifically "TIMEOUT" and "SERVER ERROR" more than trippled since last month. Happy to go to a differnet task, but since this one was active, I'm reporting it here in case it's already looked into.

@Jacobolus Have you noticed the target change to MathJax SVG, rather than only MathML? Ref T271001#12123535.

In T432727#12184474, @ihurbain wrote:

[…]

• the boldface greek is bold AND italic

In T432727#12191162, @ihurbain wrote:

As I understand it:

• in mathoid mode, we send the TeX directly to mathoid and let it do its thing (as I understand it: ship it to mathjax, which handles the full TeX->MathML->SVG conversion)
• in mathjax mode, we convert the TeX to MathML, and that's what we ship to mathjax-client-side.

[…] I'm also assuming that we might not want exactly the same conversion for "send MathML for the browser to be interpreted" and "send MathML for MathJax to interpret". […]

In T429706#12111832, @SD0001 wrote:

Note: Caniuse differs from MDN and incorrectly shows Chrome 85+. Confirmed using Browserstack that Chrome 80 supports this.

I added the following to https://meta.wikimedia.org/wiki/Tech/News/2026/33:

In T271001#12189893, @Jacobolus wrote:

[…] As one basic example, the "alignat" environment has completely wrong spacing. […]

In T271001#12188818, @STei-WMF wrote:

Hello @Krinkle, please see if the rewritten entry is accurate:

The Math extension supports the rendering of mathematical formulae. Soon formulae images will now be rendered in the browser instead of on the server. Wikibooks will see this change on 12 August, Wikisource on 19 August and Wikipedia from 20-27 August. MathML will continue to render server side and is supported in browsers without JavaScript. The SVG image is a progressive enhancement. This change is part of deprecating RESTBase and deprecating Mathoid. You can try this by selecting "Client side MathJax rendering" in your preferences.

I have questions:

1. MathML extension page redirects to Math. Does this mean they are the same extensions? If so, am I right to interpret this sentence "MathML will continue to render server side and is supported in browsers without JavaScript." to mean that till changes start on 12 August, Math (aka Math ML) continues to render server side?

Works in the UI like this. Nice!

Codesearch query: -microtime|microtime\([^)]*\) -

I suggest we verify (and, if reproducible, fix) this as part of work on the ActorStore at T428517: The user's row is being loaded three times on every page view.

In T380344#10340438, @Jdforrester-WMF wrote:

Historically, we've raised the browser support requirements based on a feature we want to use that we think is worth the trade-off. […]

In T431372#12092928, @Bugreporter wrote:

Because these endpoints see very little adoption outside of WMF owned features, we expect there to be a relatively short deprecation cycle;

Note they are used in default installation of Math in 1.42 and before (though all such versions are EOL), and later version if they enable the mathoid mode (it is disabled by default). See T373705 and https://www.mediawiki.org/wiki/Extension:Math#Math_output_modes

So before that we need to remove the Mathoid mode from Math.

Optimistic schedule:

I can't reproduce it. It uses MathJax for me in Firefox and Chrome:

@Jacobolus wrote:

I don't understand why this is targeting MathML rather than SVG rendering from MathJax. […] Or is this the SVG output?

Status update:

• Rollout phase 1: Enabled native MathML with MathJax as the default on group0 wikis (including test.wikipedia.org and mediawiki.org).
• Reduced MathJax client size by 20% from 1.59MB to 1.31MB. T414055#12175948
• Rollout phase 2: Enabled native MathML with MathJax as the default on wikis other than Wikipedia/Wikisource/Wikibooks. This includes Wikiversity articles, Commons file pages, and Wikidata talk pages.

Phase 2 rollout (wikis other than Wikipedia/Wikisource/Wikibooks, such as Wikiversity, Commons, and Wikidata).

If I understand correctly, what the bugs have in common is that they are using the 3-value and 4-value syntax, which is the "side-relative" or "edge offset" feature of background-position. If this is incorrect, let me know. If this is correct, but there are other bugs here as well, please create a separate task for those.

In T395347#12179109, @Krinkle wrote:

Doc updates:

• https://www.mediawiki.org/w/index.php?title=Compatibility&diff=8540290&oldid=8527134
• https://www.mediawiki.org/w/index.php?title=Template%3ACompatibility_browser&diff=8540292&oldid=8482458

Doc updates:

• https://www.mediawiki.org/w/index.php?title=Compatibility&diff=8540290&oldid=8527134
• https://www.mediawiki.org/w/index.php?title=Template%3ACompatibility_browser&diff=8540292&oldid=8482458