[71] in athena10
Re: Athena 10 source packages
daemon@ATHENA.MIT.EDU (Evan Broder)
Wed Jan 30 13:26:19 2008
Message-ID: <47A0C138.80607@mit.edu>
Date: Wed, 30 Jan 2008 13:26:00 -0500
From: Evan Broder <broder@MIT.EDU>
MIME-Version: 1.0
To: Greg Hudson <ghudson@mit.edu>
CC: athena10@mit.edu
In-Reply-To: <200801301813.m0UIDsA2010230@equal-rites.mit.edu>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Greg Hudson wrote:
> Actually, I have one new concern about this approach. I'm rebuilding
> the orig tarfile each time I create or update a build directory, but
> the autogenerated and boilerplate portions of the tarball are going to
> depend on the machine used to run dasource.
>
> How cranky will the Debian package tools get if the orig tarball
> changes from release to release without the upstream version number
> changing? This may be an issue even without the transform step, since
> the MD5 of a .tar.gz file is dependent on more than just the contents
> of the files contained within (it also depends on file owners, mod
> times, the phase of the moon affecting the order of readdir(), etc.).
> I'd rather not have to create a "golden" orig tarfile as a separate
> step each time the base sources change, but I will adapt to that if
> necessary.
>
Heh. I solved this problem by writing a script to build my tarballs and
setting various flags and properties manually. I set the file owner to
root/wheel and the mod time to the timestamp from the most recent commit
on the package, and the tarballs seem to be coming out mostly the same.
You can see my source at /mit/macathena/bin/macathenaDist.py
- Evan
