Add dependencies to fix opam-dune-lint

[mbarbin]

So that it is easier to avoid duplicating entries, use a canonical order to sort the libraries.

The canonical order I used here was to put ocaml first, and then the rest of dependencies alphabetically.

Note that this PR is based on top of #49, so it is currently marked as draft.

• rebased on top of format dune-project using [dune format-dune-file] #49

[mbarbin]

I noticed the following item didn't pass in #48 in the list of checks:

❌ EXPERIMENTAL: (lint-opam) (failure: Lint failed for grpc-async.opam)

The motivation for this PR is to fix that check. Rebasing the PR across this change will also allows to verify that #48 doesn't introduce breaking changes for opam-dune-lint.

[mbarbin]

This satisfies opam-dune-lint when checked locally. Before merging we also need to verify the result of the check on the ci.

$ opam-dune-lint
grpc-async.opam: OK             
grpc-bench.opam: OK
grpc-eio.opam: OK
grpc-examples.opam: OK
grpc-lwt.opam: OK
grpc.opam: OK

[mbarbin]

• This PR fixes the opam-dune-lint (lint-opam) check in the ci

https://ocaml.ci.dev/github/dialohq/ocaml-grpc/commit/cbac7ecaebebb366c2825fb8ba69e2a6a42e6e9b/variant/%28lint-opam%29#L808-814

/src: (run (shell "opam exec -- opam-dune-lint"))
grpc-async.opam: OK
grpc-bench.opam: OK
grpc-eio.opam: OK
grpc-examples.opam: OK
grpc-lwt.opam: OK
grpc.opam: OK

[mbarbin]

I just noticed that a new check is failing:

❌ EXPERIMENTAL: freebsd-5.1_opam-2.1 (failure: Failed: Build failed)

I don't have experience with ocaml-ci nor FreeBSD builds. I haven't managed to access the log for the failure (500 error page on ocaml-ci). I don't know how the changes in this PR could cause this check to fail. There's a possibility this is a spurious or flickering failure. Putting this out there for your consideration.

Note that opam-dune-lint is also experimental, but I think it is important to try and validate this check, because it may indicate missing dependencies that could cause build failure on any platform, and also because I believe there's an ongoing effort to make build lint eventually required in ocaml-ci. I don't know how important is the Freebsd check for ocaml-grpc, and if it is reasonable to prioritize opam-dune-lint over it.

[mbarbin]

The check appears to be successful again.

https://github.com/dialohq/ocaml-grpc/pull/50/checks

✅ freebsd-5.1_opam-2.1 (passed)

If this isn't a transient issue, one potential cause could be the unintentional
initial increase of the uri dependency bound from >=4.0.0 to >= 4.4.0,
which I've since reverted. However, I'm skeptical that this was the root cause,
as I vaguely remember seeing version 4.4.0 being fetched anyways by the CI at
some point (though I can't recall the exact details, and I'm currently unable to
access the log for the check due to a 500 error).

Regardless, I think that this PR doesn't make the situation worse regarding this
check.

[quernd]

I think this makes sense. We didn't have e.g. h2 as a dependency of grpc-eio because it depends on grpc which depends on h2 but since it also uses h2 directly it should arguably be made explicit. The canonical order is also appreciated. Thanks!

[mbarbin]

Thank you @quernd for reviewing the PR. I've observed that the CI checks have re-run and some are now showing an internal error status. I'm unsure why they re-ran as they were all passing previously. Could you possibly request a re-run? I'm hopeful this is a transient issue that will resolve itself.

[quernd]

Thank you @quernd for reviewing the PR. I've observed that the CI checks have re-run and some are now showing an internal error status. I'm unsure why they re-ran as they were all passing previously. Could you possibly request a re-run? I'm hopeful this is a transient issue that will resolve itself.

I triggered a re-run but I'm seeing many "Internal Error" failures like this:

# Run eval $(opam env) to update the current shell environment
2024-01-12 15:30.25 ---> saved as "325465377ee1d011b19968dfebc53b21e0af2792e4207fc22a135c090b7376b6"


/src: (copy (src .) (dst /src))
Uncaught exception: Failure("\"rm\" \"-rf\" \"/var/cache/obuilder/merged/ece1ab8e3d253142eeecd688f1483cca1180f7e651edb0539f927b21490b21f9\" \"/var/cache/obuilder/work/ece1ab8e3d253142eeecd688f1483cca1180f7e651edb0539f927b21490b21f9\" failed with exit status 1")
2024-01-12 15:30.26: Job failed: Failed: Internal error

I don't know how rm -rf can fail with exit status 1, to be honest. @tmcgilchrist you understand OCaml-CI better, do you know what's going on?

[quernd]

All green now, likely a transient issue. Thanks @mbarbin

[tmcgilchrist]

It was an issue on the Mac workers with removing cached results. In future if you see issues please report them on https://github.com/ocurrent/ocaml-ci/issues

…

On Tue, 16 Jan 2024 at 4:39 am, Daniel Quernheim ***@***.***> wrote: Merged #50 <#50> into main. — Reply to this email directly, view it on GitHub <#50 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABJXOLKGSBMDOPJD3H7FILYOVSWVAVCNFSM6AAAAABBPOCHKGVHI2DSMVQWIX3LMV45UABCJFZXG5LFIV3GK3TUJZXXI2LGNFRWC5DJN5XDWMJRGQ4TAOBWGQ4DEOI> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[quernd]

It was an issue on the Mac workers with removing cached results. In future if you see issues please report them on https://github.com/ocurrent/ocaml-ci/issues
…

Thanks, will keep it in mind!