Re: [PATCH] gpg-interface.c: detect and reject multiple signatures on commits

From: Jonathan Nieder <jrnieder@gmail.com>
Date: Aug 17, 2018, 06:54 UTC
Message-ID: <20180817065444.GC131749@aiede.svl.corp.google.com>
In-Reply-To: <1534488137.1262.2.camel@gentoo.org>

Michał Górny wrote:

> On Wed, 2018-08-15 at 14:31 -0700, Jonathan Nieder wrote:

>> It's not clear to me that the name is_status or SIGNATURE_STATUS
>> captures what this field represents.  Aren't these all sigcheck
>> statuses?  Can you describe briefly what distinguishes the cases where
>> this should be 0 versus 1?

[...]

>                                  Maybe it should be EXCLUSIVE_STATUS
> or something like that, to distinguish from things that can occur
> simultaneously to them.

Thanks.  Makes sense.

[...]

Show 7 quoted lines

>> Can we have a test to make sure this behavior doesn't regress?  See
>> t/README for an overview of the test framework and "git grep -e gpg t/"
>> for some examples.
>
> Will try.  Do I presume correctly that I should include the commit
> object with the double signature instead of hacking git to construct it?
> ;-)

Good question. You can hack away with a new program in t/helper/, or you can make your test do object manipulation with "git cat-file commit <object>" and "git hash-object -t commit -w --stdin". If you run into trouble, just let the list know and I'm happy to try to help. (Or if you would like real-time help, I'm usually in #git-devel on freenode.)

Jonathan