Thank you @Edtadros, great write up!

I reviewed the real user monitoring (RUM) we have in place for loadEventEnd and domComplete and didn't notice any obvious spikes:

Moved this to block to wait for deployment date. When ready, the patches associated with T331657 should be backported first, followed by the config change to enable page tools for anons

2023 update — this is still an issue. Here is the profile from my $100 Xiaomi Redmi 9A phone:

@Edtadros I will add QA steps, sorry for not including them earlier

In T331657#8690150, @Jdlrobson wrote:

I might be wrong but I'm not seeing the pinning functionality in the main menu, only in page tools. It's late in my day so I'll check again tomorrow. Maybe I've got my local setup in a muddle.

@Jdlrobson and I used a script to roughly measure the effect that adding languages to the mobile pages has on the html payload for the most popular pages in the month of February 2023.

@ovasileva @Jdlrobson I noticed this issue on beta and mediawiki today as well.

I am very interested in getting rid of the visual regression docker container and possibly switching to playwright. This issue has become worse because on arm64 machines we install a version of chromium that is unsupported by puppeteer (since they don't bundle a supported version themselves) and that has resulted in many of the output images looking totally different than the images generated on other machines

@Edtadros I will add QA steps soon

Sorry for the lack of response here! I've just submitted a PR to re-enable this . Thank you for fixing!

@Patafisik Thank you for the report. I tried replicating this on my Mac and Window (via Browserstack), but haven't been able to replicate this problem. Can you please check my recording at https://app.screencast.com/KfbS8Q0V1Uxay?conversation=2E9AB8aetLIJcobL533a0s, and let me know if I need to do something different to replicate it?

The srcset attribute was added on the mobile site as part of T326147 and was deployed everywhere on Feb 27. Therefore, I'm marking this as resolved

• Analyze FirstView android dashboard
• Analyze ParserCache dashboard
• Analyze NavigationTiming dashboard

CPU times, including HTML parse times, from the synthetic test results (without srcset, with srcset) on a Moto G (5) phone look pretty similar:

I'm OOO starting tomorrow Feb. 22 and will return Wednesday, March 1:

@Ladsgroup Regarding the possibility of parser cache issues between the mobile and desktop sites, @Edtadros found that the beta mobile site's infobox has a horizontal scrollbar on articles like "Dog":

FYI I analyzed the payload effects this change has on the top 100 most popular pages in January in T293303#8630702

In T293303#8337960, @SCherukuwada wrote:

@Krinkle Do you have a recommendation on how to measure the potential impact on experience/bandwidth for the affected users (regardless which option we chooose)?

This still looks like an issue:

In T326147#8593913, @DLynch wrote:

Re: test cases, DiscussionTools does its own transforms on talk pages on mobile web, so Editing has some interest in keeping an eye on changes that affect this.

Thanks for the explanation @Sgs! I see the patch to make it a soft dependency, so I'll probably wait until that's merged before re-enabling the extension

@Nux It has currently been deployed to all group 0 wikis (e.g. https://www.mediawiki.org/wiki/WYSIWYG_editor) and is scheduled to deploy to group 1 (e.g. cawiki) today and group 2 (e.g. enwiki) tomorrow

@Edtadros I think this is a pass based on Alex's comments on T319315:

//@ovasileva, this passes, based on the specs above. There is one peculiar behavior outside of the scope in the QA steps. When logged in the TOC when sticky sits a bit lower on the page. I'm not sure why this is or

@bwang The mediawiki-config change has been backported. Handing this back to you