Re: Tarball builds in the new world order
Tom Lane <tgl@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Alvaro Herrera <alvherre@alvh.no-ip.org>
Cc: Peter Eisentraut <peter.eisentraut@enterprisedb.com>,
Greg Sabino Mullane <htamfids@gmail.com>,
pgsql-hackers@lists.postgresql.org,
Magnus Hagander <magnus@hagander.net>
Date: 2024-04-28T19:44:10Z
Lists: pgsql-hackers
Alvaro Herrera <alvherre@alvh.no-ip.org> writes: > Why is it that the .gitrevision file is only created here, instead of > being added to the tarball that "git archive" produces? Adding an > argument like > --add-virtual-file $(distdir)/.gitrevision:$(GIT_REFSPEC) > to the git archive call should suffice. I think we don't want to do that. In the first place, it's redundant because "git archive" includes the commit hash in the tar header, and in the second place it gets away from the concept that the tarball contains exactly what is in our git tree. Now admittedly, if anyone's built tooling that relies on the presence of the .gitrevision file, they might prefer that we keep on including it. But I'm not sure anyone has, and in any case I think switching to the git-approved way of incorporating the hash is the best thing in the long run. What I'm thinking of doing, as soon as we've sorted the tarball creation process, is to make a test tarball available to the packagers group so that anyone interested can start working on updating their packaging process for the new approach. Hopefully, if anyone's especially unhappy about omitting .gitrevision, they'll speak up. regards, tom lane
Commits
-
Allow selecting the git revision to be packaged by "make dist".
- 01df14763431 17.0 landed