Comments

58 Comments
User Icon music commented & provided feedback on jrnl-2.8.4-2.fc36 4 weeks ago

Thanks for your testing!

Updated with a new build; see comments in FEDORA-2022-afbc5a5a95.

Updated with a new build to fix the “*” symlink regression mentioned above, and to fix the self-reported version number.

Thanks! I’ll patch the version number in the source code.

I’m not sure what to do about the trailing BUILD (here “2”). It’s really designed to count upstream binary release builds, I think. It’s exposed to plugins but I’m not sure if any use it. Still, that keeps me from wanting to stuff the entire RPM “release” string in there, e.g., 1.fc36, lest some plugin expect it to be a decimal representation. Maybe I’ll just use the RPM packages numeric pkgrel field.

I also discovered that I introduced a regression in which broken symlinks named * could be created by the launcher script in ~/.imagej/luts and ~/.imagej/plugins. That was relatively harmless, but I’ll update this with a fix for that, as well.

I had to manually update my F35 workstation with the corresponding pyproject-rpm-macros update in order to be able to do anything, even just fedpkg build, with a spec file that uses the new -w option. I guess that isn’t surprising.

Once I did that, this worked beautifully with python-hatchling in F37 and F36. I can fully test it on F35 once the update for that release moves from pending to testing.

This is great!

BZ#2076994 RFE: Support generating runtime BuildRequires for build backends without prepare_metadata_for_build_wheel()

Works to run the async tests for python-aiosignal.

BZ#2063769 Please provide python3-pytest-asyncio for EPEL-9

Replaced the usd build in this update with one that contains the fix for fix for CVE-2022-28041. Changed update type from bugfix to security.

Thanks for testing!

So, this update is ABI-compatible with the existing blender package, but it would cause it to FTBFS until cmake(pxr) is changed to usd-devel.

Note that blender has its own CMake scripts for finding usd, which it was always using instead of the ones installed with usd-devel. However, this update no longer provides cmake(pxr) since it does not install CMake scripts.

Although rebuilding Blender is not strictly necessary (just pushing a commit to the branch in dist-git could be enough to prevent an FTBFS bug from being filed), I’ll nevertheless rebuild Blender and add it to this update.

So, this update is ABI-compatible with the existing blender package, but it would cause it to FTBFS until cmake(pxr) is changed to usd-devel.

Note that blender has its own CMake scripts for finding usd, which it was always using instead of the ones installed with usd-devel. However, this update no longer provides cmake(pxr) since it does not install CMake scripts.

Although rebuilding Blender is not strictly necessary (just pushing a commit to the branch in dist-git could be enough to prevent an FTBFS bug from being filed), I’ll nevertheless rebuild Blender and add it to this update.

karma

This does, in fact, fix the regression in building luminance-hdr with LTO enabled (scratch-build in side tag).

This update was combined with and now obsoleted by the older update FEDORA-2022-262b2da36d, which was also still in testing.

This update has been unpushed.

For testing with python-fastapi, requires: FEDORA-2022-dbf6e00ba8

Fails to install:

Error: 
 Problem: cannot install the best candidate for the job
  - nothing provides python3.9dist(scikit-learn) >= 1.0.1 needed by python3-imbalanced-learn-0.9.0-3.fc34.noarch

FEDORA-2022-117a79c82d has the same issue but is already stable.

Awesome! Glad it’s working as expected.