Does monitoring perf have any negative perf impacts?

Way better, thanks. Can we rename the file to make it clear it's MariaDB config? How about .localmariadb.cnf?

Can you make sure CI passes before landing? It looks like you'll need to add some [ignore] lines to some .flowconfigs

Let's defer landing this for a while

Add comments

Remove brotli.js from keyserver in favor of built-in Node.js utility from zlib library

1. It looks like Brotli is included in Zlib in (at least) Node >= 18 LTS: https://nodejs.org/dist/latest-v18.x/docs/api/zlib.html#class-brotlioptions. Would it be possible to leverage this instead of brotli.js (which doesn't seem to have a "real" commit since ~2016)?

Remove brotli.js from keyserver in favor of built-in Node.js utility from zlib library

pendingToRealizedThreadIDsSelector is used for three things:

Allow SIDEBARs

This actually breaks threadInfoFromSourceMessageIDSelector... I'll need to make the change up the stack

Thanks, this looks super close!!

Note: that means that web clients will need to reload to run the cookie migration.

That one is trivial to fix... here you go: D9284

One quick note: my original idea in ENG-4981 was to "extend RobotextParams to include a sidebar parent's ThreadInfo". In this case you appear to have extended it to include all threads' parents' ThreadInfo, which I think is probably better. I thought about whether we could save some work by reducing scope to my original proposal, but probably not. But if you think of something let me know.

It's generally an expectation that if you put a diff back on a reviewer's queue, you have addressed that reviewer's comments. If you haven't, you should at least attempt to explain why you're putting the diff back in that reviewer's queue (maybe you have a question?). Otherwise you should hit the "Plan Changes" button to keep it away from reviewers' queues until you've addressed all of the comments.

Don't forget to add a comment to ENG-2199 before landing!