[PATCH 0/2] test bundle --stdin, fix objects_array_remove_duplicates()

From: Jonathan Nieder <jrnieder@gmail.com>
Date: Apr 19, 2010, 07:14 UTC
Message-ID: <20100419071449.GA23166@progeny.tock>
In-Reply-To: <20100119002641.GA31434@gnu.kitenet.net>

This series does not make ‘git bundle --stdin’ work. In fact, the bug that it does fix would not be visible if bundle --stdin worked as it was supposed to.

Instead, this fixes a segfault that the bug triggers. The patches have been sitting in my git tree for a while. They fix a real problem regardless, and maybe they would help someone to get bundle --stdin working properly.

Thoughts welcome, as always.
Jonathan Nieder (2):
  t5704 (bundle): add tests for bundle --stdin
  fix "bundle --stdin" segfault

 object.c          |    4 ++--
 t/t5704-bundle.sh |   16 ++++++++++++++++
 2 files changed, 18 insertions(+), 2 deletions(-)

As for how to fix bundle --stdin, Joey Hess wrote:

> I also tried going back to 22568f0a336ac37ae7329c917857b455839d1d09, but
> still see a bug with Adam Brewster's initial code to add --stdin to
> git-bundle. That code still tries to read stdin twice. If it sees
> "master" both times, it does create a bundle.

and Johannes Schindelin suggested slurping up the input and explicitly using it twice, or:

> Alternatively, you can try to implement the rev-list --boundary by
> hand (the --pretty=oneline is only needed to get a boundary marker
> IIRC), taking care to reset the commit flags that were set in the
> process.

[...]

> If you want to go that route (which is arguably more elegant anyway),
> I suggest having a look at the merge_bases() and get_merge_bases()
> functions in commit.c

Dscho’s full message is pretty helpful: http://thread.gmane.org/gmane.comp.version-control.git/137414/focus=137503