Jean-Paul Calderone

I think this was intentional. The Eliot action succeeds to reflect the fact that the test runner succeeded in "running" the test. The fact that the test itself may have...

These also come after a timeout of `magic_folder.test.cli.test_magic_folder.CreateMagicFolder.test_leave_no_folders_at_all`, not sure if that's related somehow.

Okay. Sure. Sounds like that would make a good, separate ticket too.

I also encountered this when trying to re-pack an Obelisk thunk to rev 41f97410cfa2e22a4ed9e9344abcd58bbe0f3287. ps tells me that nix-thunk has a child process `/nix/store/v41a4jzfgs36srbip4gjfh2fx10m2hmj-git-2.31.1/bin/git -C impl/ rev-list --left-right master...origin/master` and...

I also wouldn't have thought of Twisted first when thinking about where such functionality should live. If the intended user is pydoctor and you want to avoid the overhead associated...

> I'm a little confused about the status of this PR. Why'd the other one get closed? Note I didn't get any notification about this comment. I merely happened upon...

> > If we remove dependencies on `memory` and `foundation` , we need another source for constant-time comparisons and comparisons to an all-zero string. > > I don't know this...

An improvement here would be to separate what is logically the Functor part of `addCallback` (callbacks returning `T` - aka `fmap`) from what is logically the Monad part of `addCallback`...

The "sometimes" is for the same revision. Several times I've triggered a rebuild of a revision for which the build failed this way and it succeeds the next time.

I've seen it 3 or 4 times now so ... somewhat reproducible. But I haven't been able to trigger it on demand yet. I can try a bunch of builds...