On Mon, 23 May 2011 15:39:07 +0200, Matthias Guedemann <Matthias.Guedemann@ovgu.de> wrote:
> 
> > If you did then I'm in the same boat. Notmuch/emacs used to display both
> 
> welcome to the boat!

My fault. I'm sorry about this breakage.

> Now a quick-and-dirty way around this:
> 
> reset to c51d5b3cdb5ca0816816e88ca6f7136a24e74eee
> 
> and apply this patch:
..
> -       *part_count = *part_count + 1;
> -       (*show_part) (part, part_count);
> +       //*part_count = *part_count + 1;
> +       //(*show_part) (part, part_count);

Your patch here removes the change to the part numbering for "notmuch
show".

I've just pushed a change which instead adds this same part numbering to
"notmuch part". Please try it out and let me know if you have any
further trouble.

Clearly, we've got a hole in our test suite since it didn't notice this
problem at all.

Meanwhile, Jameson has proposed (in IRC) a patch series to further
rework the way that "notmuch part" works. That series also adds various
tests for "notmuch part" (which will instead be called "notmuch show
--part") to the test suite.

So hopefully we won't have this same regression appear in the future.

Thanks, everyone for reporting this bug.

-Carl