linux-kernel - Re: [ANNOUNCE] Git v2.8.1 (and call-for-help to "make rpmbuild" users)

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <57038A24.2030808@drmicha.warpmail.net>
Date:	Tue, 5 Apr 2016 11:49:24 +0200
From:	Michael J Gruber <git@...icha.warpmail.net>
To:	Junio C Hamano <gitster@...ox.com>, git@...r.kernel.org
Cc:	Linux Kernel <linux-kernel@...r.kernel.org>
Subject: Re: [ANNOUNCE] Git v2.8.1 (and call-for-help to "make rpmbuild"
 users)

Junio C Hamano venit, vidit, dixit 03.04.2016 21:21:
> If you do not build RPM binary packages from our pristine source,
> you can safely ignore this release and stop reading this message.
> 
> Now that the audience of this message has been limited to a narrow
> target, before I make an announcement, here is a call-for-help to
> you.
> 
> Git v2.8 removed README file and added a corresponding README.md
> file.  The change however did not adjust git.spec.in that still
> referred to README, causing "make rpmbuild" to fail.  The breakage
> was not noticed by anybody who tested v2.8.0-rc0 and later release
> candidates, and ended up in the final v2.8 release, and we saw a
> handful of bug reports on the list after the release happened.
> 
> This maintenance release is to correct this bug for those who run
> "make rpmbuild". It has no other changes.
> 
> It is clear that nobody who relies on being able to "make rpmbuild"
> ever tried any of the 5 release candidate snapshots that happened
> during Feb 26-Mar 28.  We had a whole month and nobody noticed?
> 
> This incident clearly shows that something needs to happen, if
> people want "make rpmbuild" to keep working.  Even though this
> maintenance release may fix this single bug, breakages similar to it
> that only affect "make rpmbuild" users are guaranteed to appear in
> future releases, unless those who can prevent them from happening
> start helping to test at least release candidate snapshots.

In RedHat/Fedora land, we use our own spec files, "make" and "make
install" into a special purpose tree and package that.

E.g. in fedora, the packager noticed the README file name change in
2.8.0, adjusted the corresponding "%doc" line, and that was basically
all besides updating the sources.

As a fedora user, I either use rpms targetting fedora (possibly patched
and rebuilt by myself), or I build from sources (--prefix=$HOME). With
"generic" rpms you never know which guidelines they follow or where they
install to. Also, naming schemes for dependencies may differ between
RedHat land, SuSe land and other rpm based distros.

So, just like in the deb case, I'm wondering if there's a use case for
make rpmbuild in git.git.

Michael