Skip to content

Install PSScriptAnalyzer directly instead of via psmodulecache - #3035

Closed
nohwnd wants to merge 1 commit into
rel/5.x.xfrom
ci/drop-psmodulecache-rel5
Closed

Install PSScriptAnalyzer directly instead of via psmodulecache#3035
nohwnd wants to merge 1 commit into
rel/5.x.xfrom
ci/drop-psmodulecache-rel5

Conversation

@nohwnd

Copy link
Copy Markdown
Member

Same change as #3034 on main.

sha_pinning_required applies to actions referenced inside composite actions. potatoqualitee/psmodulecache does not pin its own actions/cache reference - v6.2.1 uses @v4.2.0, v6.3 uses @v5.0.5 - so the job fails at Set up job before running anything.

This branch matters because its workflows trigger on rel/*, so they actually run and would break exactly the way main did. rel/6.0.x has the same line but its code-analysis only triggers on main, so it never fires there.

The step only installed PSScriptAnalyzer and ConvertToSARIF, which take seconds from PSGallery.

Upstream fix sent as potatoqualitee/psmodulecache#68, but we should not depend on it landing.

🤖

Same change as #3034 on main. sha_pinning_required applies to actions referenced
inside composite actions, and potatoqualitee/psmodulecache does not pin its own
actions/cache reference (v6.2.1 uses @v4.2.0, v6.3 uses @v5.0.5), so the job fails
at "Set up job" while we depend on it.
This branch triggers on rel/*, so it runs and would break the same way main did.
🤖
@nohwnd
nohwnd marked this pull request as draft September 5, 2026 09:53
@nohwnd

Copy link
Copy Markdown
MemberAuthor

Parking this as a draft. sha_pinning_required is off again so nothing is blocked, and the module cache step predates this work, so I want to discuss it before removing it.

The reason enforcement and psmodulecache cannot coexist is in the description above, and the upstream fix is potatoqualitee/psmodulecache#68.

🤖

@nohwnd

Copy link
Copy Markdown
MemberAuthor

Closing in favour of #3036, where the reasoning and the options are written up.

Removing the module cache is only one of four ways to handle this, and it is not obviously the right one until we know what the cache was solving. The policy is off, so nothing is blocked.

🤖

@nohwndnohwnd closed this Sep 5, 2026
Sign up for freeto join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant

@nohwnd