[Teammetrics-discuss] Gitstat problems.

[David Bremner]

On Fri, 19 Aug 2011 01:42:57 +0530, Sukhbir Singh <sukhbir.in at gmail.com> wrote:

>     $ git show ff1876eb0123a91c9e83e3c8379f936f0ac6c398
>     commit ff1876eb0123a91c9e83e3c8379f936f0ac6c398
>     Merge: 94d08ad dcc5c4b
>     Author: Steffen Möller <steffen_moeller at gmx.de>
>     Date:   Mon Jul 4 15:13:08 2011 +0200
> 
>     Merge branch 'debian'
> 
> Can the difference be due to the fact that this commit was a merge? Anyways,

Now I see it, after cloning via git:// rather than http:// . Not sure
what happened there.

It is indeed skipped because it is a merge commit that does not modify
any files.  You have to be a bit careful reading git log -z; that
debian/control was actual attached to the previous sha1.  I hope that is
the desired behaviour from your point of view; I'm not sure how to
easily analyse a merge commit which doesn't modify any files.

In this particular workflow, where everything is committed to the debian
branch and then merged to master, I think merges are noise. Actually,
that is part of why I don't like a "debian only" branch so much with
git.

> > BTW, did you try "git log --oneline debian" as a way to get commits
> > modifying files under debian? it seems like it might be simpler.
> 
> This seems to work wonderfully. I will just add '--no-merges' to
> filter out the merges.

BTW, (and to answer Andreas's question in another email), I should
really have written

git log --oneline -- debian 

to ensure we get commits on the current branch touching the path debian, 
rather than commits on a branch debian.

d