git_repository is not reproducible

[cameron-martin]

Description of the bug:

The git_repository repository rule leaves the .git directory, either in the root or in .tmp_git_root. This directory contains files that are not a function of the input attributes, for example .git/logs contains files that contain timestamps of when the remote branch was fetched.

This causes a real issue in rules_rust (bazelbuild/rules_rust#1927), since it generates targets that glob the entire external repository.

What's the simplest, easiest way to reproduce this bug? Please provide a minimal example if possible.

Create a git_repository repository rule, inspect the .git/logs directory and see that it contains timestamps.

Which operating system are you running Bazel on?

Ubuntu 22.04

What is the output of bazel info release?

release 6.1.0

If bazel info release returns development version or (@non-git), tell us how you built Bazel.

No response

What's the output of git remote get-url origin; git rev-parse master; git rev-parse HEAD ?

No response

Have you found anything relevant by searching the web?

No response

Any other information, logs, or outputs that you want to share?

No response

[sluongng]

Does it not remove it here

bazel/tools/build_defs/repo/git.bzl

Line 176 in 0d672f3

ctx.delete(ctx.path(".git"))

🤔

[cameron-martin]

I guess this bug only happens when strip_prefix is specified, and the git repository is in .tmp_git_root. I didn't observe the other case, and just assumed. I think the line above needs to be changed to:

if ctx.attr.strip_prefix:
    ctx.delete(ctx.path(".tmp_git_root"))
else:
    ctx.delete(ctx.path(".git"))

[fmeum]

@cameron-martin That looks very reasonable. Do you want to submit that as a PR?