Removing uneeding branchName argument for github packages

[b-rodrigues]

After our discussions and some thought, I think that indeed, we should simply get rid of this argument, and better do it now before submitting to CRAN. I have also changed the code on the server, but not deployed yet; locally, it works!

[b-rodrigues]

Once we merge this, I'll also merge on the server.

[b-rodrigues]

If you want to test the api locally: https://github.com/b-rodrigues/git2nixsha/tree/main

[b-rodrigues]

Tests pass locally, but complain here because when nix is not available, they try to reach the api, however the new version without branchName is not deployed it. I'd like your second opinion before pulling the trigger on the api and this PR.

[philipp-baumann]

After our discussions and some thought, I think that indeed, we should simply get rid of this argument, and better do it now before submitting to CRAN. I have also changed the code on the server, but not deployed yet; locally, it works!

I agree its a good idea to do this now!

[philipp-baumann]

If you want to test the api locally: https://github.com/b-rodrigues/git2nixsha/tree/main

Can we replace this:

https://github.com/b-rodrigues/git2nixsha/blob/6ad61c8be5f343c6f8851f9617d218a7c013d8ee/git2nixsha.R#L15

by the new approach like we do locally in hash_url() and hash_git?:
https://github.com/b-rodrigues/rix/blob/699372eaadef9a24b7f23121641ce97754fc6659/R/nix_hash.R#L135
https://github.com/b-rodrigues/rix/blob/699372eaadef9a24b7f23121641ce97754fc6659/R/nix_hash.R#L22

Then we really have consistent behavior. To be sure, I think we should also add a new test (CI with nix installed), and compare equality of get_sri_hash_deps() test output after options(rix.hash_nix="api_server") and tets whether we get identical results compared to calling it after options(rix.hash_nix="locally") (only run if nix is installed).

Overall, not using git2r maybe we get also rid of some hashing issues found before (if its not related to an other issue). Anyway, I think it's good that we have the mechanisms match 1 to 1 locally and on the hashing API server.

[b-rodrigues]

done with b-rodrigues/git2nixsha@01151eb