Updated the doc @Oauth2.0!

@Krinkle, Although I initially thought adding LessMessage to the switch was the cause, but after separating the grouped options of dependencies, message, and LessMessage the bug seems to go away when $option = array_values( array_unique( (array)$option ) ); was disabled for dependencies.

Thanks for the follow-up! So I was able to see why we have the regression! Removing duplicates is causing the bug in VisualEditor.

Thank you!

Thanks for the heads up. I was able to reproduce on beta, but I couldn’t on my localhost. Is there a way to enable the "Editing Mode" option locally?

How I tested Foreign Wiki pages'. For subclasses with a foreign $this->getDB(), getting page info from Title will result to null as it looks at the local db.

It looks like I have to create the tool before setting up a repo for the project. I just applied for membership, and my request is still pending.

Okay! Thanks for the feedback @matmarex! I'll have those set up and let you know how it goes :)

@matmarex, please have a look and let me know what you think! I haven’t tested with a real client_id yet.
https://phabricator.wikimedia.org/F66343855

I think this is what fixed the bug for less.js https://github.com/less/less.js/pull/3291/files

To replicate the errors, I had to do $this->overrideConfigValue( CAMainConfigNames::CentralAuthEnableSul3, false ); in APiLoginTest.

@matmarex, were you able to reproduce the failed test locally?

So I tried doing this locally by adding $wgPHPSessionHandling = 'disable'; to my localsettings and running: ./vendor/bin/phpunit tests/phpunit/includes/session/.

Hello @Krinkle, this is what it looks like for titleinfo-user cache hit.

Is it safe to remove the test referencing the destructor function for the test files, or do we move that to the test to make the test pass? CC @Tgr @matmarex

Thanks @Tgr! I'll do that and see how it goes!

This is what I'm experiencing:

Hello @Tgr , I’ve been trying to come up with a scenario where session::save is called from destruct, but I haven't been able to.

Thanks to @Krinkle, we were able to increase our Grade A traffic percentage to 85%, which is a significant improvement. However, we still consider the result inconclusive due to missing grading data for iOS devices, which are a major part of our traffic.

Currently, our compatibility test successfully excludes browsers that do not support ES2017.

Here is the script I used to get the page views per month using all_sites_by_browser_family_and_major.tsv data.

Currently, async causes a bug when it appears as a literal in property expression;

Thank you!

In T389384#10653328, @jnuche wrote:

Error is gone from the logs, thank you for the fix @Hokwelum :)

I'm removing this as a train blocker

Summary of what was causing this bug:

1. Risky Patch! 🚂🔥

In T367441#10562250, @Krinkle wrote:

Re-opening for the ESLint TODO.

Yes, @mszabo! We'll be doing all that! That's the only way the fix would be in the next mw train!

Hello @Daimona! Sorry you had to go through all that. Hopefully, the fix should be included in the next MediaWiki deployment train, so core will be updated soon.

Sounds good!

In T385997#10549001, @Ladsgroup wrote:

I will drop the table and remove it from the catalog the week after to make sure all code has reached production.