New version 1.6.1 is released.
Updates may require up to 24 hours to propagate to mirrors. If the following command doesn't work, please retry later:
sudo dnf upgrade --refresh --advisory=FEDORA-2022-2f75a06546
Please login to add feedback.
This update has been submitted for testing by mtasaka.
This update's test gating status has been changed to 'ignored'.
This update has been pushed to testing.
This update has been submitted for stable by bodhi.
Works great! LGTM! =)
This update has been pushed to stable.
This update breaks in some non-obvious way test suite of rubygem-activerecord [1]:
The fact is that even if rubygem-racc-1.6.0-205.fc38 is installed, when I change the line of "# stub: racc 1.6.0 ruby lib" line in racc-1.6.0.gemspec to 1.6.1, the above error is reproduced, on the other hand, even if rubygem-racc-1.6.1-200.fc38 is insalled, changing the line "# stub: racc 1.6.1 ruby lib" to 1.6.0 removes the above error.
So I think there is some path search mechanism confusion when default gem and "external" racc gem are both installed: I guess that when rubygem-racc-1.6.1-200.fc38 is installed, activerecord (or some dependency gem or so) tries to search path for 1.6.1 is "default gem" and it is failing, perhaps. Anyway not a fault for this update, so far.
You are right that update of racc is probably just red herring. The problems are:
1) The test cases are heavily stubbing
Pathname.new
, which is used by RubyGems. 2) The issue is that whilepathname
is already required and re-evaluated while it should be no-op.I have filled this issue along the way:
https://github.com/rails/rails/issues/46741
The idea is that if
Pathname
was not used at all, there would not be need forrequire
. However, it is still unclear why it is attempted to be loaded. There might be some problem in Zeitwerk or elsewhere. I think I have also seen at some point, that no gem binary extensions are discovered for some reason. So there is something fishy.Anyway, I think that if needed, we can skip this test.