We might be able to detect that the --strategy-option theirs merge has done something suspicious by trying to merge C^ into the final state (or vice versa). If it merges cleanly, I think we can be fairly confident that the changes are actually independent. However, I'm not sure offhand what we could conclude if it doesn't merge cleanly.

A particular example is:

In the general case, the problem is that some invocations of arc, including arc land --pick A, may select an arbitrary range of commits to land which have some ancestors that we do NOT want to land.

git diff ...

• Address code review comments

Couple minor inlines, but looks good to me. I added you to Blessed Committers so you should be able to land this yourself -- see the description of that project for guidance, or let me know if you run into issues.

Broadly, I believe these "certain changes" are approximately changes which include a merge commit.

• Text tidying.

This is actually very subtle.

In general, Phacility production hosts may interact with logs as several different users:

This ran into some filesystem permissions issues and needs followup.

When we are importing a commit, we often don't have any legitimate user to act as: the commit may come from an observed repository (so we never authenticated a pusher), and the "Author" and "Committer" strings are arbitrary and untrusted in Git and Mercurial.

All three of these are almost certainly diffusion.rawdiffquery, I think the first one just predated D21463 reaching production so it didn't show the method.

This deployed, and appears resolved.

I deployed this and nothing blew up, presuming this is resolved.

Presuming this is resolved until evidence to the contrary arises.

Calls to preg_* can still be swapped to phutil_preg_* to improve behavior, but this seems basically resolved.

Enable the SSH and HTTP application logs on the web, repo and admin tiers.