Hi folks,
I'm seeing something odd, and I'm looking to know if anyone else has had the same experience, and then, if so, how to work around it.
I build my RPMs to work on a LOT of systems, and it necessitates some pretty hairy stuff. One of them is the eternal struggle to lock down a dependency on the apache-de-uh, I mean httpd-devel package in more-than-2 distro releases. I think they need to give a degree for that.
So I've got this (pseudo, ignore spurious line-breaks):
%define distro rh62
%define buildreq_rh62 apache-devel
%define buildreq %{expand:%%{?buildreq_%{distro}}%%{?!buildreq_%{distro}:apache-devel}}
BuildPreReq: %buildreq
But, then again, everyone has that. The goal is that the RPM will fail happy with:
error: Failed build dependencies:
apache-devel is needed by cure-cancer-1-1It doesn't under rpm402. For some reason, it charges on ahead like it didn't even see the build dep. Does rpm402 ignore builddep lines with macros in them? Does it ignore BuildDep lines with macros that fail for whatever reason? My macro returns the expected data on an echo, but I'm wondering if some there/not-there issue affects it 'live'.
BuildPreReq: %{_sbindir}/apxsis also no good. Ideally, I'd like to avoid anything not approved for rpm306, too. I'm also a one-SRPM guy, or I'd just have 6 million different SRPMs to suit every point-release-candidate out there. I'll overlook the massive amount of errors resulting from the lack of a -devel package to query ('-q: command not found' or so) if I can just make it bail from build-deps like it should (and also it looks better in apt if it does the build deps 'Right').
For all my griping about This Week's GCC, I know a solution must be out there. We're not that short-sighted and provincial as a community, are we? Nah, I'm just not seeing something. Any ideas?
-- - bish
_______________________________________________ Rpm-list mailing list Rpm-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/rpm-list
