From: Jeff King <peff@github.com>
Date: Thu, 16 Jun 2011 17:46:53 GMT
Subject: Re: [PATCH 7/7] archive: provide builtin .tar.gz filter
Message-ID: <20110616174653.GD6584@sigill.intra.peff.net>
In-Reply-To: <20110616075621.GA12413@arachsys.com>

On Thu, Jun 16, 2011 at 08:56:21AM +0100, Chris Webb wrote:

> Jeff King <peff@github.com> writes:
> 
> > No, I mean that people may _want_ the timestamp in day to day use. Using
> > "-n" all the time suppresses it. And there is no reason to suppress it,
> > except that our test does not account for it properly. So your patch is
> > hurting people who don't want "-n" (i.e., want the timestamp) just to
> > make our test happy.
> 
> It's useful to omit the timestamp outside of git too. Source-based package
> management systems generally store a URL from which to fetch a source
> tarball, and a hash of that source tarball to ensure it hasn't been tampered
> with. It's nice to be able to use a gitweb URL like
> 
>   http://git.kernel.org/?p=git/git.git;a=snapshot;h=e5af0de202e885b793482d416b8ce9d50dd2b8bc;sf=tgz
> 
> as the tarball source, and still be able to verify its integrity against a
> prestored hash.

OK. I'm totally willing to accept that people actually prefer the "-n"
behavior. I don't care either way myself. I just don't want the reason
to default to "-n" to be "because our test scripts need it" and not
"because this is what people actually want".

-Peff