Both the regression I reported on the 0.6.2 update and https://github.com/fedora-infra/rpmautospec/issues/71 / https://github.com/fedora-infra/rpmautospec/issues/80 appear to be fixed in 0.6.3. Thanks!
$ fedpkg co cramjam-cli
$ cd cramjam-cli
$ fedpkg mockbuild -- --dnf
Downloading cramjam_cli-0.1.1.tar.gz
######################################################################## 100.0%
Could not execute mockbuild: Couldn’t parse spec file cramjam-cli.spec:
Spec file is missing.
The above works fine with 0.6.1.
There appears to be a regression with rpmautospec
packages in Koji that started yesterday. I don’t know what version of rpmautospec
is running there, and I don’t know if it’s associated with an rpmautospec update or something else, so I’m not giving negative karma – but I am leaving this comment as a heads-up.
Superseded by FEDORA-2023-427f7d3cbb.
This update has been unpushed.
The new build should have identical contents, but it was built with the test suite enabled.
One user did report problems in the Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=2240134#c10
This seems to work fine for me on a GNOME/Wayland session, using either the Wayland or the non-Wayland launcher.
I haven’t tested this interactively, but it installs, and that’s infinitely better than the previous status quo!
The versioning is correct now, this doesn’t appear to cause any new FTI bugs, and the pylint CLI passes a quick interactive “smoke test.”
I’m not confident that this is perfect, but until upstream catches up, it’s a lot better than not having uvloop and uvicorn at all! With this update, I am able to restore the uvloop and uvicorn dependencies in python-asyncpg and python-fastapi, respectively.
I guess packaging an alpha of pylint 3 was the only way to get acceptable Python 3.12 support? It’s unfortunate that according to https://pylint.pycqa.org/en/latest/whatsnew/3/3.0/index.html there will be further breaking changes before 3.0.0 final, and they can be expected to break people’s configuration files.
I think I will proceed by removing the PySide2 support just as I have done in F39 and F40/Rawhide. That configuration should be supportable for the remaining life of EPEL9.
Updated as described.
@tis, thanks for your feedback. I was aware that PySide2 is incompatible with Python 3.12, which is why qtsass
’s PySide2 support is disabled in F39+. The package is still useful there for supporting Spyder and a couple of other packages that use QDarkStyleSheet with PyQt5.
I was not aware that PySide2 was about to be broken in EPEL9. That’s unpleasant, and I’m glad you told me!
I think I will proceed by removing the PySide2 support just as I have done in F39 and F40/Rawhide. That configuration should be supportable for the remaining life of EPEL9.
Still fixes FTBFS in python-Rtree
due to Python 3.12 ctypes regression.
Since FEDORA-2023-4543b659d6 reached F39 stable first with a rebuilt fastnetmon
for capnproto
, I needed to rebuild fastnetmon
again. Unfortunately, that means karma and time in testing for this update have been reset.
Tested by manual download and installation of RPMs since this is not yet in
updates-testing
.I confirmed that a trivial crate library package (
rust2rpm drop-tracker
) has the same spec file except for therust2rpm
version comment.I confirmed that
rust2rpm -p b3sum
correctly flags the missing license file in the crate, and that I can proceed usingrust2rpm -p --ignore-missing-license-files b3sum
. Furthermore, I was able to write arust2rpm.toml
that almost eliminates hand-editing the generated spec file, including generating and installing a man page withhelp2man
and patching in the missing license file from upstream. All I had to do by hand was replace# FIXME: no license files detected
with%license LICENSE
, and paste in the output of%cargo_license_summary
in the right place.This is a really exciting update!