What is also bad is that the CC is still added in gerrit:
https://chromium-review.googlesource.com/c/v8/v8/+/1009904
https://screenshot.googleplex.com/4cmJ93pFTq6.png

I'd like to know where it's coming from. But apparently codesearch and gerrit know something that I don't.

Hmm, I have a hunch that the V8 auto-roller might be massaging those watchlist file. I'll double check that.

Indeed. We keep a separate watchlist file on the release branches that gets copied around.

@hablich: I think the bug is here https://codereview.chromium.org/1663003003/diff/20001/tools/release/create_release.py line 166. This uses "latest_version" for the WATCHLIST. But shouldn't it use master?

For version file and changelog it makes sense to make a diff from the last rolled version, since we manipulate it differently in each version. But I don't see how that makes sense for the watchlist file. That has regular updates on master, which we all miss.

Let me take a look and see what I can find ...

OK, yeah: you'll of course be more familiar with what your release scripts/process are doing.

What I can tell you is that for Code Search we pay attention to the version that is specified in the DEPS file.  And indeed the DEPS file shows the revision of a release tag, and the version of WATCHLISTS on that tag matches what appears in Code Search.

https://chromium.googlesource.com/chromium/src/+/917298c4c641a98e0722cc7e7d6cdcf5234211d7/DEPS#89

This is completely on us. We roll an old WATCHLIST into Chromium and Codesearch uses the v8 in src/v8 not master.

I'll fix that and get the watchlist updated.

I too have no clue why we are doing that.