Folks, rpm 4.20 alpha landed in rawhide today, and with the sheer amount of change that went into the bowels of the build code, this process is being rougher than usual. Apologies for the disruption and the late heads-up.
Please file bugs with low bar to entry if you suspect an rpmbuild regression. It's better to have false positives than silent bugs or people working around the wrong things etc.
So far we're aware of:
- GenericError: srpm mismatch on subpackage debuginfo (so not any debuginfo, only subpackage debuginfo), is an rpm regression and we're working on a fix (see the other thread for more details)
- java/mvn package fail on test-related paths https://bugzilla.redhat.com/show_bug.cgi?id=2283795, this is not a regresion but an intended change in the rpm build paths layout where all builds now have a their own specific build-directory whether they use %setup or not.
The build directory layout change is something the vast majority of packages will never notice, but packages manipulating paths relative to "raw" %{_builddir} *may* need to be adjusted.
So again, if in doubt at all, ask or just file a bug.
- Panu -
Dne 29. 05. 24 v 2:31 odp. Panu Matilainen napsal(a):
Folks, rpm 4.20 alpha landed in rawhide today
4.19.91 == 4.20 alpha?
Because I do not see 4.20 in Koji:
https://koji.fedoraproject.org/koji/packageinfo?packageID=319
On 5/29/24 18:09, Miroslav Suchý wrote:
Dne 29. 05. 24 v 2:31 odp. Panu Matilainen napsal(a):
Folks, rpm 4.20 alpha landed in rawhide today
4.19.91 == 4.20 alpha?
Because I do not see 4.20 in Koji:
https://koji.fedoraproject.org/koji/packageinfo?packageID=319
Yes, that's the one. cmake doesn't support the kind of creative versions autotools did.
- Panu -