On Mon, Jan 30, 2023 at 9:00 PM Miguel Ojeda <miguel.ojeda.sandonis@xxxxxxxxx> wrote: > > On Mon, Jan 30, 2023 at 2:29 AM Masahiro Yamada <masahiroy@xxxxxxxxxx> wrote: > > > > I guess you are talking about kernel-devel-*.rpm > > (and linux-headers-.deb). > > > > They are not useful for building external modules > > written in Rust since they do not contain *.rmeta etc. > > I am not caring about that because Rust support is not > > mature enough yet. > > Yeah, that is what I meant, i.e. since the Rust ML was Cc'd, I checked > and wanted to say removing `rust` from there was OK (an `Acked-by` > seemed too much for just that line :). > > > I stopped hard-coding the top-level directories. > > The resulting source package still contains all check-in files > > under rust/, so it is good from the source package perspective. > > Sounds good to me. > > > 5/5 changed the behavior because rpm-pkg re-uses the > > *.src.rpm generated by srcrpm-pkg. > > (3/5?) Yes. > > > Having *.src.rpm in the kernel tree seems Redhat's preference. > > Commit 8818039f959b2efc0d6f2cb101f8061332f0c77e > > added --define='_srcrpmdir $(srctree)'. > > Thanks for the details! I just noticed it, so I thought I would let > you know just in case. > > (Perhaps it could be useful to mention this change in the output in > the commit message.) Fair enough. I updated the commit description in v3. > > > In contrast, binary rpm files are generated under rpmbuild/RPMS/. > > I want to fix this inconsistency, though. > > That would be nice. > > Cheers, > Miguel -- Best Regards Masahiro Yamada
