another semi-mass-rebuild (Was: Re: rebuilt for unwind info generation, broken in gcc-4.1.1-21)

[Robert Scheck]

Hello folks,

On Mon, 02 Oct 2006, Thorsten Leemhuis wrote:
> Well, here they are. The current plan is:
> 
> - do another mass rebuild for all those packages that might be affected.
> That means: all arch packages that were build from Sept. 8 until Sept.
> 26 -/+ some hours for safety
> - let a script (
> https://www.redhat.com/archives/fedora-extras-list/2006-July/msg00777.html
> ?) increase release, commit, tag and queue the build of all effected
> packages. Someone from FESCo (probably c4chris) will handle that.
> - we start on Wednesdays evening (CEST)
> - we hereby encourage maintainers to queue the rebuilds of their stuff
> of their own before Wednesdays evening

IMHO your current plan is broken or even wrong. Please do something like 

for i in *debuginfo*; do
  rpm2cpio $i | strings | grep -E "4.1.1-2[123456]" | sort -u;
done

to check whether your packages were built using the buggy GCC. Because
using the timeframe, there would be (many?) false-positives, e.g.

  mimedefang-2.57-4.fc6  17-Sep-2006  GNU C 4.1.1 20060828 (Red Hat 4.1.1-20)
  tcpick-0.2.1-10.fc6    10-Sep-2006  GNU C 4.1.1 20060828 (Red Hat 4.1.1-20)

My solution also saves time and ressources of the build system, because two
false-positives out of my seven packages seem to be very much - at least for
me.


Greetings,
  Robert