... and make use of the new pywrap feature from our test.
Use the same eslint and stylelint plugin configuration as the cockpit
main repo.
Co-authored-by: Katerina Koukiou <kkoukiou@redhat.com>
(cherry picked from commit 2215aa3bf8)
Bump Cockpit commit to pick up the necessary adjustments for latest PF,
and adjust test for the new "-v5" namespace prefix.
Closes#654
(cherry picked from commit 092cefab3b)
Fedora 37's rpmspec changed behaviour: `-q` now shows the source RPM
name instead of the binary one. `--srpm` and `--rpms` don't influence
this behaviour any more. So get along with both variants.
The tlog UID was being set explicitly to ensure journalctl matches
used during tests would find pre-recorded journals. This is no longer
needed as we removed the TLOG_UID filter from the journalctl match
string.
For consistency with the already xz'ed node tarball.
Rename the oddly named `dist-gzip` target to the standard `dist`.
Cherry-picked from starter-kit commit 4ca75f143c
node-sass is a compiled ELF module, which is problematic for
distributions that want to rebuild everything from source. The sassc CLI
program is packaged everywhere, and both use the same libsass library.
So drop node-sass and replace it with sass which is also what cockpit
and other external plugins are using.
Port alerts in PF-react-core and fix missing icons in alerts
The previous fix (commit 6e05f5b483) only cleaned
test/images/$(TEST_OS), which is just a symlink to $(TEST_OS).qcow.
Clean the actual image as well.
Closes#46
Remove the entire previous overlay instead of just removing the built
package. This ensures that there are no leftovers from previous
interactive debugging sessions.
This has already shown to lead to confusion in practice in
cockpit-podman.
Closes#44
When the topmost commit isn't tagged, rpmbuild otherwise fails with
error: line 2: Illegal char '-' (0x2d) in: Version: 176-1-g9101a30a
Replace the dashes from `git describe` with periods, to get a valid RPM
upstream version number.
Closes#45
This is only being used for extracting the package name from package
json, to avoid having to duplicate it in `Makefile`.
But for only this purpose, Python 3 is rather heavyweight. It's also not
available on RHEL/CentOS 7.
Our release tarballs only contain files, not their containing
directories, due to `git ls-files`. tar creates these directories on the
fly on unpack, with a current timestamp instead of an archived one. This
causes directories in src/ to be newer than dist/ and thus a `make` will
try to rebuild the webpack. This breaks RPM builds.
Release tarballs already have dist/. We can't rebuild the webpack in an
RPM as that requires npm and network access for `npm install`. But we
also don't want to ship the entire `node_modules/` as that is huge and
would require lots of careful license review.
So ship a stub node_modules/ that satisfies the Makefile dependency.
Make sure that this is not newer than package.json or anything in dist/,
so that none of the Makefile's rebuild rules trigger.
Closes#36
When building in production mode, there is no index.js, just an
index.min.js.gz. This previously caused unnecessary rebuilds, or even
RPM build failures in environments where npm was not available. Test for
index.html instead, which is reliable.
Also factor this out into a constant to make it easier to adjust in
forked projects.
This is easier to handle downstream, as the spec file can be used as-is,
instead of having to get rebuilt.
Adjust the clean rule to only remove the .spec if the .spec.in exists,
so that it gets removed in the upstream git, but not in unpacked release
tarball trees.
Building the *.po files requires the `po2json` module, so add a
dependency to it.
Put the example node module into a variable to avoid repeating it. Also
test for `po2json` instead of `react-lite`, as the latter is more likely
to get replaced by actual projects in favor of the full React.
Closes#33
