I’ll take a look on Monday

Looks like my hack is not actually necessary, since bazel is already in nativeBuildInputs of the fixed output derivation and therefore any change to bazel will re-trigger the fixed output derivation build.

I seem to remember that I had some problems with fetchpatch not getting re-evaluated even though patchutils had changed, but apparently that is not the case here for some reason 🤔

Anyway, I'll remove the hack.

Can you put it in a sub-attributeset? Then one can do nix-build -A bazel.tests.downstreamPackages or something.

Done

@GrahamcOfBorg build bazel.tests.downstreamPackages

I just noticed that the downstream tests won't actually be run when running nix build -f. bazel.tests. Is that what you intended @Profpatsch?

I just noticed that the downstream tests won't actually be run when running nix build -f. bazel.tests. Is that what you intended @Profpatsch?

Oh, hm. Maybe we need to add recurseIntoAttrs.

@GrahamcOfBorg build bazel.tests

Adding recurseIntoAttrs indeed solves the issue.

@GrahamcOfBorg build bazel.tests

dm-sonnet is failing on linux.
bazel-watcher is failing on darwin (as noted in #65268).

@kalbasit can you take a look at bazel-watcher? @timokau you are maintaining dm-sonnet, can you look into it?

@Profpatsch bazel-watcher fix is in #65308.

you are maintaining dm-sonnet, can you look into it?

Not currently. If nobody else steps up, it'll have to wait for a week or two.

We should also add bazel-buildtools to the downstream tests. It doesn’t use bazel …

Not currently. If nobody else steps up, it'll have to wait for a week or two.

I’ll remove it from the dependency test for now and open a new issue.

Rebased on master & disabled dm-sonnet

@GrahamcOfBorg build bazel.tests

And we’re live.

Tests went through, autosquashing and merging.