On 09/01/18 16:16 +0100, Tomasz Torcz 👁️ wrote: > On Tue, Jan 09, 2018 at 12:11:31PM +0100, Jan Kurik wrote: >> = System Wide Change: Binutils version 2.29.1 = >> https://fedoraproject.org/wiki/Changes/BINUTILS2291 >> >> Change owner(s): >> * Nick Clifton <nickc AT redhat DOT com> >> >> Rebase the binutils package from version 2.29 to version 2.29.1. This >> will bring in the bug-fixes from the 2.29.1 point release, but not add >> any new features. >> > >> Change the source parameter in the binutils.spec rpm and adjust the >> local patches to take account of the bugs that are now already fixed. > > I'm a bit perplexed by this change. It looks like minor version > update, in such case it need no to be announced so widely. > On the other hand, you are changing the source. According to the > guidelines, changing source requires re-review. > So why this is a system-wide change? I think I have something to add here as if no other package, libqb was quite affected with the former change of binutils 2.28 -> 2.29 coming to Fedora 27: https://bugzilla.redhat.com/show_bug.cgi?id=1478089 + binutils bug: https://bugzilla.redhat.com/show_bug.cgi?id=1477354 (TL;DR eventually, libqb adopted measures to overcome changed visibility of some symbols, binutils didn't go back in its behaviour) In parallel, there was a separate discussion directly with upstream: https://lists.gnu.org/archive/html/bug-binutils/2017-08/msg00195.html which resulted in https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=f399679112df997c1416f7993eaac0f5fd76c144 that is part of 2.29.1 and which actually forced an existing prototype of the libqb fix to be refined once more because, suddenly, some new configure-time checks were to loose to detect the necessity to apply said measures (while they were working fine with 2.29 alone). So I can attest this change has a merit (compare with unannounced 2.29 change that hit libqb hard), even though it's not expected the number of packages relying on some rather obscure linker features (that are hence not under constant coverage) will be notable. But keep in mind that a single broken package can spoil some other, dependent packages, as was the case with libqb. Thanks for playing it considerately now, Nick. -- Jan (Poki)
Attachment:
pgpI9KbtflQIo.pgp
Description: PGP signature
_______________________________________________ devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx
