Fix grpc_cli missing a library for execution

[jan-xyz]

This glob pattern doesn't match the actual file on macOS anymore.
On MacOS the file is called libgrpc++_test_config.1.dylib. Trying to run grpc_cli results in an error:

$ grpc_cli ls localhost:8080
dyld: Library not loaded: @rpath/libgrpc++_test_config.1.dylib
  Referenced from: /usr/local/bin/grpc_cli
  Reason: image not found
[1]    40338 abort      grpc_cli ls localhost:8080

I am not sure if the new pattern also matches in linux. Is there a way I can provide a test which makes sure we do not break it again? It broke during #67474

• Have you followed the guidelines for contributing?
• Have you checked that there aren't other open pull requests for the same formula update/change?
• Have you built your formula locally with brew install --build-from-source <formula>, where <formula> is the name of the formula you're submitting?
• Is your test running fine brew test <formula>, where <formula> is the name of the formula you're submitting?
• Does your build pass brew audit --strict <formula> (after doing brew install <formula>)?

---

[SMillerDev]

Is there a way I can provide a test which makes sure we do not break it again?

You can invoke the cli in the test block.

[jan-xyz]

@SMillerDev, thanks for the hint. It's my first patch to homebrew-core and I have 0 experience, so I really appreciate this "obvious" hint!

I added the test:
Without the patch on macOS:

$ brew test grpc
==> Testing grpc
==> /usr/bin/clang test.cpp -I/usr/local/Cellar/grpc/1.33.2_2/include -L/usr/local/Cellar/grpc/1.33.2_2/lib -lgrpc -o test
==> ./test
==> grpc_cli help ls
Last 15 lines from /Users/jan/Library/Logs/Homebrew/grpc/test.03.grpc_cli:
2021-01-06 13:42:33 +0100

grpc_cli
help
ls

dyld: Library not loaded: @rpath/libgrpc++_test_config.1.dylib
  Referenced from: /usr/local/bin/grpc_cli
  Reason: image not found
Error: grpc: failed
An exception occurred within a child process:
  BuildError: Failed executing: grpc_cli help ls
/usr/local/Homebrew/Library/Homebrew/formula.rb:2096:in `block in system'
/usr/local/Homebrew/Library/Homebrew/formula.rb:2032:in `open'
/usr/local/Homebrew/Library/Homebrew/formula.rb:2032:in `system'
/usr/local/Homebrew/Library/Taps/homebrew/homebrew-core/Formula/grpc.rb:83:in `block in <class:Grpc>'
/usr/local/Homebrew/Library/Homebrew/formula.rb:1900:in `block (3 levels) in run_test'
/usr/local/Homebrew/Library/Homebrew/utils.rb:530:in `with_env'
/usr/local/Homebrew/Library/Homebrew/formula.rb:1899:in `block (2 levels) in run_test'
/usr/local/Homebrew/Library/Homebrew/formula.rb:904:in `with_logging'
/usr/local/Homebrew/Library/Homebrew/formula.rb:1898:in `block in run_test'
/usr/local/Homebrew/Library/Homebrew/mktemp.rb:63:in `block in run'
/usr/local/Homebrew/Library/Homebrew/mktemp.rb:63:in `chdir'
/usr/local/Homebrew/Library/Homebrew/mktemp.rb:63:in `run'
/usr/local/Homebrew/Library/Homebrew/formula.rb:2145:in `mktemp'
/usr/local/Homebrew/Library/Homebrew/formula.rb:1892:in `run_test'
/usr/local/Homebrew/Library/Homebrew/test.rb:43:in `block in <main>'
/usr/local/Homebrew/Library/Homebrew/vendor/portable-ruby/2.6.3_2/lib/ruby/2.6.0/timeout.rb:93:in `block in timeout'
/usr/local/Homebrew/Library/Homebrew/vendor/portable-ruby/2.6.3_2/lib/ruby/2.6.0/timeout.rb:33:in `block in catch'
/usr/local/Homebrew/Library/Homebrew/vendor/portable-ruby/2.6.3_2/lib/ruby/2.6.0/timeout.rb:33:in `catch'
/usr/local/Homebrew/Library/Homebrew/vendor/portable-ruby/2.6.3_2/lib/ruby/2.6.0/timeout.rb:33:in `catch'
/usr/local/Homebrew/Library/Homebrew/vendor/portable-ruby/2.6.3_2/lib/ruby/2.6.0/timeout.rb:108:in `timeout'
/usr/local/Homebrew/Library/Homebrew/test.rb:42:in `<main>'

with the patch on macOS:

$ brew test grpc
==> Testing grpc
==> /usr/bin/clang test.cpp -I/usr/local/Cellar/grpc/1.33.2_2/include -L/usr/local/Cellar/grpc/1.33.2_2/lib -lgrpc -o test
==> ./test
==> grpc_cli help ls

[SMillerDev]

This should work then. And it'll make sure it fails as we expect it to fail.

[jan-xyz]

Hmm the new test fails. I am trying to figure out locally why it doesn't match.

[iMichka]

The .* at the end needs to remain, as the filename is something like .so.1 on linux if I remember well.

[jan-xyz]

The .* at the end needs to remain, as the filename is something like .so.1 on linux if I remember well.

I will re-add that :) Thanks for checking in. There is no CI test for Linux, right?

EDIT: @iMichka what do you think about libgrpc++_test_config.{1.dylib,so.1}? Or would you prefer libgrpc++_test_config*.{dylib,so}*

[iMichka]

There is no CI test for Linux, right?

It's tested once it is merged into linuxbrew-core. The test is a little bit ... delayed :) But I will come back to this PR if it fails.

I would not hardcode the .1, that value might be different. Ideally there should just have been a single "make install" step so that we do not have to go through so many hoops to ship this.

Changing from libgrpc++_test_config*.{dylib,so}.* to libgrpc++_test_config*.{dylib,so}*, as the only change you make is to remove the mandatory dot at the end, which is not needed on Mac, but keep the * to tell it to include everything at the end on linux (I hope I got my ruby-regex-fu right).

[jan-xyz]

(I hope I got my ruby-regex-fu right)

It's actually a shell glob pattern :D

Changing from libgrpc++_test_config*.{dylib,so}.* to libgrpc++_test_config*.{dylib,so}*

Uhh yeah, that's a nice minimal change. I adjusted the PR accordingly, thanks!

[iMichka]

We need to bump the revision to 3 to ship the updated content to everyone.

missed need for revision

[jan-xyz]

I bumped the revision. Sorry for that 😬

Done

[BrewTestBot]

🤖 A scheduled task has triggered a merge.

[jan-xyz]

Thank you everyone! This was super pleasant to work on and I really appreciate all the quick replies and feedback. It was a small fix but super motivating and rewarding. Have a great day ❤️