Apparently, the &status[]= format doesn't work, but specifying &status=
three times appears to work.
Signed-off-by: Thomas Bracht Laumann Jespersen <laumann@gentoo.org>
When looking for the status of the "last" run, we only want to consider
completed runs. Without this filtering, we risk seeing runs that are
still in progress (status is "running") which isn't of interest and only
results in spamming the #guru channel.
Signed-off-by: Thomas Bracht Laumann Jespersen <laumann@gentoo.org>
This commit replicates the CI/CD behavior we currently have on GitHub
by making the following changes:
- Jobs that used Actions were replaced with in-repo scripts. This
removes the dependency on GitHub and makes the overall pipeline more
forge-agnostic.
- rectalogic/notify-irc replaced with ./scripts/irc_notify.py
- pkgcore/pkgcheck-action replaced with ./scripts/pkgcheck.sh
- Actions that used `gh` (the GitHub CLI) were replaced with in-repo
scripts that perform the same action but with the Forgejo API.
- ./scripts/last_status.py to get the last status of a completed job
(e.g. succeeded, failed, canceled)
- ./scripts/last_successful_commit.py to get the last commit that had
a successful job
- Forgejo-specific changes
- Variables renamed from "github" to "forgejo" where appropriate
(though Forgejo lets you get away with keeping them as "github" to
make migration easier)
- Use ./.forgejo/ instead of ./.github/ -- this will simultaneously
keep the GitHub actions intact while also disabling the GitHub
actions from running in Forgejo
- Codeberg-specific changes
- Use codeberg-tiny-lazy shared runner for {duplicates,emails}.yml
- Use codeberg-medium-lazy shared runner for pkgcheck.yml
Signed-off-by: Dustin Smith <d.usty360@gmail.com>
Part-of: https://codeberg.org/gentoo/guru/pulls/30
Merges: https://codeberg.org/gentoo/guru/pulls/30
Signed-off-by: Thomas Bracht Laumann Jespersen <laumann@gentoo.org>
A new check that checks that all new emails written in metadata.xml
correspond to existing user in bugzilla.
Signed-off-by: Arthur Zamarin <arthurzam@gentoo.org>
checking out any current repoman version tag puts portage
into a version where EAPI 8 support is not enabled
Signed-off-by: Theo Anderson <telans@posteo.de>
Without the -V flag to enable version sort, 'sort' can falsely identify
a version string like 3.0.9 to be greater than 3.0.10. This can be
shown by running 'git tag | grep portage | sort -u' in the Portage Git
tree. This issue is not so prominent for repoman yet because the latest
version of repoman as of now is 3.0.3, but it will surface when repoman
3.0.10 is released.
Signed-off-by: Yuan Liao <liaoyuan@gmail.com>
Closes: https://github.com/gentoo/sci/pull/1094
Signed-off-by: Andrew Ammerlaan <andrewammerlaan@gentoo.org>
Reduce code duplication by moving the ::gentoo setup part
into a seperate script. This way only one file has to be
edited when e.g. updating the profile to a new version
Signed-off-by: Andrew Ammerlaan <andrewammerlaan@riseup.net>
By using a seperate script instead of running everything in the yml
file, we gain the ability to use and manipulate variables. This
means that we can now automatically run the latest version of
repoman, which should make this more hands-off by removing the
need to update this every now and then.
Signed-off-by: Andrew Ammerlaan <andrewammerlaan@riseup.net>
- add self as maintainer
- update variable names
- un-nest for loops
- check for matching package name before starting loops
- speed improvements: ~3s vs 5m
Package-Manager: Portage-3.0.14, Repoman-3.0.2
Signed-off-by: Theo Anderson <telans@posteo.de>
Repository mirror & CI