I’m very slowly bootstrapping Fedora to run on RISC-V, and today I managed to get rpmbuild to work, so that’s a sort of milestone:
... Provides: config(setup) = 2.10.4-1.fc24 setup = 2.10.4-1.fc24 Requires(rpmlib): rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1 Conflicts: bash <= 2.0.4-21 filesystem < 3 initscripts < 4.26 Checking for unpackaged file(s): /usr/lib/rpm/check-files /rpmbuild/BUILDROOT/setup-2.10.4-1.fc24.%{_arch} warning: Could not canonicalize hostname: ucbvax Wrote: /rpmbuild/RPMS/noarch/setup-2.10.4-1.fc24.noarch.rpm Executing(%clean): /bin/sh -e /usr/var/tmp/rpm-tmp.0iJnms + umask 022 + cd //rpmbuild/BUILD + cd setup-2.10.4 + rm -rf '/rpmbuild/BUILDROOT/setup-2.10.4-1.fc24.%{_arch}' + exit 0 Executing(--clean): /bin/sh -e /usr/var/tmp/rpm-tmp.Vdj45n + umask 022 + cd //rpmbuild/BUILD + rm -rf setup-2.10.4 + exit 0
Unfortunately because I haven’t got GCC working in the bootstrap environment, I’m a bit limited in the packages that I can build, so I’m starting off with some low-dependency noarch packages. In reality we won’t need to recompile noarch packages at all, they can be copied off other arch builders, but it’s a good test of rpmbuild.
“In reality we won’t need to recompile noarch packages at all”
I’ve long wondered why noarch isn’t promoted to a top-level arch for this very reason. Seems like there’s lots of extra work going on that’s entirely unnecessary. E.g., the voodoo of replacing duplicates with hard-links for the benefit of mirrors, not to mention the actual build efforts in the first place.
I realize it would probably be a major undertaking, but that’s never stopped us before. Is there something glaringly wrong with my theory?
Oh and BTW, good work on the RISC-V. This will be interesting to see how it plays out in the long run.
I’m amazed with this series of RISC-V on a FPGA, congratulations.
Which linux distro are you booting?