fix(tests): run smoke tests with partial metadata#2445
Conversation
|
tested here: github.com/rostalan/pull/133 |
|
[12:26 PM]Nick Boldt would it be better to prevent merging changes that cause incomplete metadata to be included? Shouldn't this be a failure that causes the plugin owner to go back and FIX the problem rather than hiding it for a future person to discover? Eventually we might have 3 catalogs so there's one for DP content in an incubator catalog |
nickboldt
left a comment
There was a problem hiding this comment.
I would prefer if we don't allow partial metadata. Instead we should prevent adding plugins which don't include their associated metadata.
Unless there's a use case for plugins being included in this repo for export, but NOT being in the extensions catalog for marketing them?
876a099 to
8cceb3d
Compare
|
/publish |
|
PR action ( |
8cceb3d to
8bf7b35
Compare
|
waiting for redhat-developer/rhdh-plugin-export-utils#148. |
|
/publish |
|
PR action ( |
Allow smoke tests to continue when a workspace has runnable metadata for only some published plugins, and surface skipped plugin metadata gaps in workflow comments. Co-authored-by: Cursor <cursoragent@cursor.com>
Point workflow calls to the forked export-utils workflows and include metadata coverage warnings in the auto-publish PR completion comment so warning-only missing metadata is visible.
Stop passing the `force` input into the forked update-plugins-repo-refs reusable workflow since that input is not declared in the current forked contract.
Restore workflow references from the temporary forked export-utils repo back to redhat-developer now that the utils changes are merged upstream.
9dde3e6 to
36d1261
Compare
|
/publish |
|
PR action ( |
…kflow Pass the force input again now that update-plugins-repo-refs uses upstream utils workflow contract supporting this input.
|
|
/publish |
|
PR action ( |



Summary
Test plan
.github/workflows/workspace-tests.yamlwith Python YAMLsmoke-tests/test.envstill triggers the skip pathMade with Cursor