Re: [PATCH 1/1] commit-graph: fix writing first commit-graph during fetch

["SZEDER Gábor" <szeder.dev@gmail.com>]

On Tue, Oct 22, 2019 at 05:45:54PM -0400, Jeff King wrote:
> On Tue, Oct 22, 2019 at 04:33:16PM -0400, Jeff King wrote:
> 
> > > I have failed to produce a test using the file:// protocol that
> > > demonstrates this bug.
> > 
> > Hmm, from the description, it sounds like it should be easy. I might
> > poke at it a bit.
> 
> Hmph. I can reproduce it here, but it seems to depend on the repository.
> If I do this:
> 
> diff --git a/t/t5510-fetch.sh b/t/t5510-fetch.sh
> index ecabbe1616..8d473a456f 100755
> --- a/t/t5510-fetch.sh
> +++ b/t/t5510-fetch.sh
> @@ -583,6 +583,14 @@ test_expect_success 'fetch.writeCommitGraph' '
>  	)
>  '
>  
> +test_expect_success 'fetch.writeCommitGraph with a bigger repo' '
> +	git clone "$TEST_DIRECTORY/.." repo &&
> +	(
> +		cd repo &&
> +		git -c fetch.writeCommitGraph fetch origin
> +	)
> +'
> +
>  # configured prune tests
>  
>  set_config_tristate () {
> 
> it reliably triggers the bug. But if I make a synthetic repo, even it
> has a lot of commits (thousands or more), it doesn't trigger. I thought
> maybe it had to do with having commits that were not at tips (since the
> tip ones presumably _are_ fed into the graph generation process). But
> that doesn't seem to help.
> 
> Puzzling...

Submodules?

  $ cd ~/src/git/
  $ git quotelog 86cfd61e6b
  86cfd61e6b (sha1dc: optionally use sha1collisiondetection as a submodule, 2017-07-01)
  $ git init --bare good.git
  Initialized empty Git repository in /home/szeder/src/git/good.git/
  $ git push -q good.git 86cfd61e6b^:refs/heads/master
  $ git clone good.git good-clone
  Cloning into 'good-clone'...
  done.
  $ git -c fetch.writeCommitGraph -C good-clone fetch origin
  Computing commit graph generation numbers: 100% (46958/46958), done.
  $ git init --bare bad.git
  Initialized empty Git repository in /home/szeder/src/git/bad.git/
  $ git push -q bad.git 86cfd61e6b:refs/heads/master
  $ git clone bad.git bad-clone
  Cloning into 'bad-clone'...
  done.
  $ git -c fetch.writeCommitGraph -C bad-clone fetch origin
  Computing commit graph generation numbers: 100% (1/1), done.
  BUG: commit-graph.c:886: missing parent 9936c1b52a39fa14fca04f937df3e75f7498ac66 for commit 86cfd61e6bc12745751c43b4f69886b290cd85cb
  Aborted

In the cover letter Derrick mentioned that he used
https://github.com/derrickstolee/numbers for testing, and that repo
has a submodule as well.