[MIR] needed (b-d of bzr)

Hi Barry, Matthias,

On Mon, 2011-05-23 at 23:04 +0000, Launchpad Bug Tracker wrote:
> You have been subscribed to a public bug by Matthias Klose (doko):
>
> Binary package hint: cython
>
> [MIR] needed (b-d of bzr)
Alternatively, and considering we can't sync from Debian directly
anyway, perhaps we should just flip cython and pyrex in the build
dependency on "cython | python-pyrex" in the uploads to Ubuntu?

Cheers,

Jelmer

14:08 <mterry> doko, why does cython show up in component-mismatches? it's part of a "cython | python-pyrex" build-depend, and python-pyrex is in main
14:08 <doko> germinate issue
14:08 <mterry> Is that just a nuance the mismatch script doesn't handle, or is that something we should change in the packagin?
14:09 <mterry> germinate is how we generate this report?
14:09 <cjwatson> not a germinate issue
14:09 <cjwatson> may be a seed issue
14:09 <cjwatson> yes, we germinate all the seeds and then compare against main
14:10 <cjwatson> well, I don't *think* it's a germinate issue anyway, it isn't normally. I suppose I ought to check
14:11 <mterry> cjwatson, cython doesn't seem to be in platform or ubuntu seeds
14:12 <cjwatson> sure
14:12 <cjwatson> didn't say it was :)
14:12 <mterry> cjwatson, thought that was what you meant by seed issue
14:12 <cjwatson> what I mean is that normally the data should be blamed not the processing tool
14:13 <cjwatson> in much the same way that one doesn't automatically say that a bug in the output of a C program is a gcc bug
14:13 <mterry> I thought most programmers blamed gcc for their bugs ;)
14:18 <cjwatson> from what I can tell, germinate simply ends up trying to resolve build-deps from bzr before it runs across anything that would cause it to put python-pyrex in main
14:18 <cjwatson> it's important to remember that it does not care what is *currently* in main when making this judgement; it's selecting everything from scratch
14:19 <cjwatson> the simplest workaround is probably to put python-pyrex in the supported-development-common seed along with bzr (with a suitable comment)
14:19 <mterry> cjwatson, apparently, part of our delta for bzr is already dropping other build-depends that aren't in main
14:19 <mterry> cjwatson, it might make sense just to drop cython too
14:20 <cjwatson> if you do that then it will know that python-pyrex is supported before trying to resolve build-dependencies, so it won't need to make a decision about the disjunctive build-dep
14:20 <cjwatson> *shrug* either is reasonable, personally I'd probably seed it but whichever

This seed change should fix it:

revno: 1650
fixes bug(s): https://launchpad.net/bugs/785706
committer: Colin Watson <email address hidden>
branch nick: platform.oneiric
timestamp: Tue 2011-09-27 13:35:58 +0100
message:
  explicitly seed python-pyrex to work around bzr build-dependency ordering (LP: #785706)