Matthias Klose wrote:
> On 12.04.2010 05:48, Jonathan Nieder wrote:
>> Unfortunately, patch 1 adds the current version number to some of the
>> maintainer scripts; keeping that up to date is a maintainance burden I
>> do not want to impose. So patch 2 teaches debian/rules to take care
>> of that.
>
> I added this for shlibs file generation in the meantime. please use
> the DEB_VERSION var.
Nice.
> Also the this_version stuff won't handle
> binNMUs, maybe it's enough to drop the +bN suffix from this_version.
Hmm, keeping the +bN would make the version compare later than versions
without that suffix, which would already make it do the right thing.
This probably deserves a comment, at least.
>> Patch 3 removes /usr/bin/ld.bfd from the binutils-multiarch package.
>> As far as I can tell, the linker was excluded from binutils-multiarch
>> because it had regressions relative to the single-target linker
>> (Bug#61719, #51625). Except once the linker was renamed to ld.bfd,
>> the exclusion no longer applied, creating file conflicts --- oops.
>
> No, ld.bfd is introduced again to differentiate between the gnu
> linker and the new gold linker. or do I misunderstand something?
Renamed was the wrong word, sorry. What I meant is that the rules for
binutils-multiarch do
rm -f [...] \
$(d_mul)/usr/bin/ld
but gold and ld.bfd didn’t exist when that was written, so it doesn’t
remove ld.bfd. Since there is no diversion for ld.bfd, either, the
result is that both binutils and binutils-multiarch include
/usr/bin/ld.bfd binaries, a file conflict.
> didn't integrate your patches yet, here is my current status:
> http://people.debian.org/~doko/tmp/
Thanks, will take a look.
> won't continue before the weekend
I’ll try to find time to update the patches before then.
Thanks for the feedback.
Jonathan
--
To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
