Repository navigation
(MODULES-11718) Add puppet 9 support in puppetlabs-docker - #1065
Conversation
76fe34f to
2e593d9
Compare
|
Interested in getting this in, as part of CD4PE support for Puppet 9. Our |
65c2b5a to
a529356
Compare
1) ci, nightly & mend update for puppet 9 support 2) Gemfile udpate for new gem compatible with puppet 9 3) rakefile and metadata change for puppet 9 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
voxpupuli-puppet-lint-plugins 7.0, needed for the Puppet 9 lane, bundles
exec_idempotency and params_not_optional_with_undef, which 5.0 never
shipped. With fail_on_warnings = true already set, their six warnings fail
`rake lint` outright and short-circuit metadata_lint, check:* and rubocop.
All four findings are false positives against deliberate code:
- image.pp:153 exec is `~>`-chained off a notify, so it only fires
on refresh; an `unless` guard would be dead code.
- services.pp:168 `docker service update` and `docker service scale`
- services.pp:187 are non-idempotent by design, guarded by $update /
$scale.
- run.pp:209 `Variant[String,Array[String[1],1],Undef] $net =
undef` is equivalent to Optional[Variant[...]];
rewriting it risks shifting type-checking for
consumers of $net.
Disabling keeps this PR scoped to Puppet 9 enablement and follows the
existing disable_strict_indent precedent directly above, which was added
for the same class of lane-dependent plugin behaviour.
Co-Authored-By: Claude <noreply@anthropic.com>
a529356 to
35931df
Compare
|
@imaqsood : Are there any other steps needed here to move this forward? |
CI failures are pre-existing infra, not this PRAll 6 red jobs fail before any module code runs — 5 in 1. Dead GCE image family alias (5 jobs) — CentOS-8 (pc8 + pc9), Ubuntu-20.04 (pc8), Debian-11 (pc8 + pc9): Once a distro goes EOL, Google marks every image in its family Same for 2. Debian 10 apt mirrors archived (1 job) — provisions fine, then fails agent install with exit 100: Buster has moved to Follow-up tracked under MODULES-11955: |
Checklist
puppet apply)