From: Junio C Hamano <gitster@pobox.com>
Date: Wed, 26 Mar 2008 16:45:01 GMT
Subject: Re: [PATCH] mktag.c: improve verification of tagger field and tests
Message-ID: <7vhcete8oi.fsf@gitster.siamese.dyndns.org>
In-Reply-To: <47EA7927.9060201@nrlssc.navy.mil>

Brandon Casey <casey@nrlssc.navy.mil> writes:

> Eventually filter-branch along with the rest of git will be rewritten
> in c and the need for many low-level git tools will vanish.

That holds true only for the tools shipped with git.git itself.  However,
you never should forget that people _script_ around git.  I do not see a
reason to stop supporting mktag in this discussion.

I would suggest:

 * make mktag.c a built-in first;

 * rename verify_tag() in mktag.c to verify_tag_buffer(), and update its
   implementation to tighten the format validation, perhaps along the
   lines you propose in your patch, and move it to tag.c.  By the way, I
   think tagger information should get the same validation as committer
   and author information gets elsewhere in the system;

 * add a call to verify_tag_buffer() you introduce above immediately
   before write_sha1_file() in builtin-tag.c, to make sure both programs
   produce valid tags, with the same definition of validity.