Introducing new commit conventions in system roles
by Sergei Petrosian
Hello team,
We introduce commit conventions in our organizations that will help us
automate many release tasks - automatically collect commits into changelog,
determine semantic version of new releases, and collect commits into
release notes.
We will follow Conventional Commits
<https://www.conventionalcommits.org/en/v1.0.0/> format, it requires you to
provide commit type and optional `!` after the commit type to indicate that
your commit introduces a breaking API change. You can find more information
at Conventional
<https://linux-system-roles.github.io/contribute.html#conventional-commits...>
Commits format
<https://linux-system-roles.github.io/contribute.html#conventional-commits...>
section on our site. It also generally improves commits hygiene in our org.
A correct commit format will be verified by a CommitLint GitHub Action, it
will verify each commit in the PR to follow conventional commits format,
and also the PR title to follow the format for consistency. The action
won't be able to verify if we provide a correct commit type and if we
indicate a breaking change correctly, so we must pay special attention to
it.
One concern is with the *Squash and merge *strategy that PRs currently use.
When you click the *Squash and merge* button in GitHub UI, it lets you
manually modify the resulting commit and push it to main right away. This
is a subject to human error because the resulting commit goes straight to
main, and in the case of a typo we would need to force push to main to fix
it. It is best to move to a *Rebase and merge *strategy instead to ensure
that all commits pushed to main have passed CommitLint checks. This is not
ideal but the pay off is huge.
Let me know if you have any concerns or questions.
Thank you
--
Sergei Petrosian, RHCE
He / Him / His
Software Engineer, RHEL System Roles
Red Hat Prague, Czech Republic <https://www.redhat.com>
spetrosi(a)redhat.com
M: +420792444506 IRC: spetrosi
<https://www.redhat.com>
3 months, 3 weeks
issues with recent ansible-lint and var naming
by Richard Megginson
You might have noticed that the latest ansible-lint is showing issues with
variable names:
roles/rsyslog/tasks/main_core.yml:358: var-naming[no-role-prefix]:
Variables names from within roles should use role_name_ as a prefix. (vars:
__file__)
roles/rsyslog/tasks/main_core.yml:358: var-naming[no-role-prefix]:
Variables names from within roles should use role_name_ as a prefix. (vars:
__line__)
roles/rsyslog/tasks/main_core.yml:120: var-naming[no-role-prefix]:
Variables names from within roles should use role_name_ as a prefix. (vars:
__rsyslog_global_common_rule)
Or perhaps it is flagging a variable that begins with `ansible_`
These are fixed by https://github.com/ansible/ansible-lint/pull/3458
roles/rsyslog/tasks/deploy.yml:13: var-naming[no-role-prefix]: Variables
names from within roles should use role_name_ as a prefix. (register:
__rsyslog_deploy_templates)
roles/rsyslog/tasks/deploy_nolog.yml:17: var-naming[no-role-prefix]:
Variables names from within roles should use role_name_ as a prefix.
(register: __rsyslog_deploy_nolog_templates)
This has been reported as
https://github.com/ansible/ansible-lint/issues/3452
Some of the others we will need to fix e.g.
var-naming[no-role-prefix]: Variables names from within roles should use
role_name_ as a prefix. (vars: tasks)
https://github.com/linux-system-roles/logging/blob/main/roles/rsyslog/tas...
In this case, the `tasks` variable should be renamed to `__rsyslog_tasks`
4 months, 2 weeks