I know a month or so ago we voted to add a bit to the guidelines reinforcing the necessity of -devel packages with .pc files requiring pkgconfig. I went to file a bug about a core package that wasn't doing this (which resulted in bustage of a package I was reviewing):
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=202280
It seems Matthias Clasen (mclasen@redhat.com) wants to debate this and I'm not sure what to do. I don't have any arguments other than directory ownership and "that's what the committee decided".
- J<
Jason L Tibbitts III wrote:
I know a month or so ago we voted to add a bit to the guidelines reinforcing the necessity of -devel packages with .pc files requiring pkgconfig. I went to file a bug about a core package that wasn't doing this (which resulted in bustage of a package I was reviewing):
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=202280
It seems Matthias Clasen (mclasen@redhat.com) wants to debate this and I'm not sure what to do. I don't have any arguments other than directory ownership and "that's what the committee decided".
The other reason is that .pc files are pretty much useless if pkgconfig isn't there.
IMO, the policy/guideline is in place, and it's our (packaging comittee) resposibility to lay the smack on anyone who refuses to play ball.
Now, if Matthias disagrees with the policy, it is certainly within his right to raise the issue for more debate (which folks *should* do if they disagree with the policies), but until policy is changed, he's expected to follow that policy just like everyone else.
(bugzilla updated)
-- Rex
On Thursday 17 August 2006 00:34, Jason L Tibbitts III wrote:
It seems Matthias Clasen (mclasen@redhat.com) wants to debate this and I'm not sure what to do. I don't have any arguments other than directory ownership and "that's what the committee decided".
My comments (put in bugzilla and discussed on IRC last night)
Until such time as the filesystem package owns the %{_libdir}/pkgconfig directory, this guideline will stand. If filesystem owns the directory and we remove the fact that many packages could end up owning it, we can revisit this issue and possible remove the necessity on pkgconfig for any file that drops a .pc file. However the former needs to happen before we discuss the latter.
On 8/17/06, Jesse Keating jkeating@redhat.com wrote:
On Thursday 17 August 2006 00:34, Jason L Tibbitts III wrote:
It seems Matthias Clasen (mclasen@redhat.com) wants to debate this and I'm not sure what to do. I don't have any arguments other than directory ownership and "that's what the committee decided".
My comments (put in bugzilla and discussed on IRC last night)
Until such time as the filesystem package owns the %{_libdir}/pkgconfig directory, this guideline will stand. If filesystem owns the directory and we remove the fact that many packages could end up owning it, we can revisit this issue and possible remove the necessity on pkgconfig for any file that drops a .pc file. However the former needs to happen before we discuss the latter.
This doesn't make any sense at all.
It doesn't make any sense to have /usr/lib[64]/pkgconfig owned by filesystem.
And even, for the sake of argument, that it did.
You will still be requireing that every package that include a devel package with a .pc also Require pkgconfig in order to parse that .pc file. So by changing the directory ownership from one package to another (which makes absolutely no sense whatsoever) still gains you absolutely nothing, or actually gains you more problems than what we originally had.
QED. End of stupid discussion.
On Thursday 17 August 2006 15:07, Christopher Stone wrote:
It doesn't make any sense to have /usr/lib[64]/pkgconfig owned by filesystem.
And even, for the sake of argument, that it did.
You will still be requireing that every package that include a devel package with a .pc also Require pkgconfig in order to parse that .pc file. So by changing the directory ownership from one package to another (which makes absolutely no sense whatsoever) still gains you absolutely nothing, or actually gains you more problems than what we originally had.
You'll note that I said 'discuss'. We have things like filesystem owning /usr/libexec/ which isn't part of the FHS (yet) but used by a lot of our packages. Some folks have asked that filesystem own %{_libdir}/pkgconfig/ too. I haven't stated whether or not this is a good idea, just that it has been requested. I also didn't say we'd automatically remove the need for Requires: pkgconfig, I said we'd discuss it.
As a preemptive strike, we discussed it in our meeting this week and decided that we wouldn't change the guidelines at all, things that have a .pc file should have a Require: pkgconfig, end of story.
packaging@lists.fedoraproject.org