Re: [PATCH] contrib/ciabot: git-describe commit instead of HEAD

[Junio C Hamano <gitster@pobox.com>]

Sven Eckelmann <sven.eckelmann@gmx.de> writes:

> For each commit a shorter version of the name will be generated. This is
> either the truncated hash or the output of git-describe. The
> call to git-describe was only made with an empty shell variable instead
> of an actual commit hash. Thus it only described the current HEAD and
> not each commit we want to submit to cia.vc.
>
> Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
> ---
> I send this patch already two and four weeks ago and got no reaction.
> This is a resent in case it was forgotton.

Although the computation of "rev" indeed looks wrong, I do not see it used
anywhere, so it might be a better patch to remove it ... wait, the code
does something funny to slurp everything into xml via ... 

	context = locals()
        context.update(globals())
        out = xml % context

Yuck.

Even though it has been my policy not to complain too loudly on stuff in
contrib/ part of the tree, I have to say that this is a bit too _cute_ for
my taste.

Anyway, Ok, so it _is_ used.  And it needs to be fixed.

I was hoping that Eric who took over the maintainership of contrib/ciabot/
directory in April would give his blessing sooner, without me having to
look at the script.

Ack?

>  contrib/ciabot/ciabot.py |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/contrib/ciabot/ciabot.py b/contrib/ciabot/ciabot.py
> index d0627e0..9775dff 100755
> --- a/contrib/ciabot/ciabot.py
> +++ b/contrib/ciabot/ciabot.py
> @@ -122,7 +122,7 @@ def report(refname, merged):
>      branch = os.path.basename(refname)
>  
>      # Compute a shortnane for the revision
> -    rev = do("git describe ${merged} 2>/dev/null") or merged[:12]
> +    rev = do("git describe '"+ merged +"' 2>/dev/null") or merged[:12]
>  
>      # Extract the neta-information for the commit
>      rawcommit = do("git cat-file commit " + merged)
> -- 
> 1.7.2.3