Mark auto-generated files for GitHub linguist#5269
Conversation
Extends linguist-generated coverage beyond the genkit CLI stubs in cmd/account and cmd/workspace to include direct engine YAML, schema artifacts, mocks, pydabs codegen, acceptance test golden outputs, and the genkit-managed tagging workflow. Entries land in .gitattributes.manual so they survive `./task generate-genkit`, which overwrites .gitattributes and then prepends the manual file. Co-authored-by: Isaac
Approval status: pending
|
|
|
||
| # Acceptance test golden outputs (./task test-update, ./task test-update-templates) | ||
| acceptance/**/out* linguist-generated=true | ||
| acceptance/**/output/** linguist-generated=true |
There was a problem hiding this comment.
I like seeing these on all PRs because it forces you to acknowledge any side effects of the change.
Sometimes the unexpected behavior is captured only inside the output files.
| python/databricks/bundles/volumes/** linguist-generated=true | ||
|
|
||
| # Acceptance test golden outputs (./task test-update, ./task test-update-templates) | ||
| acceptance/**/out* linguist-generated=true |
There was a problem hiding this comment.
I think all the output of the acceptance test is important to see explicitly as it helps to detect the changes in behaviour
| internal/mocks/**/*.go linguist-generated=true | ||
|
|
||
| # Genkit-generated tagging artifacts (./task generate-genkit) | ||
| internal/genkit/tagging.py linguist-generated=true |
There was a problem hiding this comment.
I'd maybe keep this unhidden because if there's something changes here it can affect the release process
Follow the existing pattern of local .gitattributes files (bundle/schema/,
bundle/internal/tf/, bundle/internal/validation/generated/, acceptance/,
integration/) instead of one large block in the root .gitattributes.
Also drop entries the reviewers asked to leave visible:
- internal/genkit/tagging.py{,.lock} and .github/workflows/tagging.yml
(release-relevant)
- acceptance/**/out* and acceptance/**/output/** (acceptance output changes
are signal in code review)
Changes
Extend
linguist-generated=truecoverage so generated files don't pollute language stats or show up as hand-written code in review. Follows the existing convention of one local.gitattributesper generated folder (alongsidebundle/schema/,bundle/internal/tf/,bundle/internal/validation/generated/,acceptance/,integration/).New / updated local files:
.gitattributeslocation./task generate-directbundle/direct/dresources/*.generated.yml./task generate-schema-docsbundle/schema/(extended)jsonschema_for_docs.json./task generate-schemabundle/internal/schema/annotations_openapi.yml./task generate-docsbundle/docsgen/output/*.mdmockeryinternal/mocks/*.go./task pydabs-codegenpython/databricks/bundles/{catalogs,jobs,pipelines,schemas,volumes}/**Excluded by request from review:
internal/genkit/tagging.py{,.lock}and.github/workflows/tagging.yml— kept visible because changes there can affect the release process (@andrewnester).acceptance/**/out*andacceptance/**/output/**— kept visible because acceptance-output diffs are how reviewers detect behavior changes (@pietern, @andrewnester).Why
bundle/direct/dresources/apitypes.generated.yml(andresources.generated.yml) were showing up as human-authored code on GitHub. The genkit-managed section of the root.gitattributesonly coverscmd/account/**andcmd/workspace/**, so the rest of the auto-generated tree was unmarked. This brings linguist's view in line with.agent/rules/auto-generated-files.md.Tests
Verified each pattern with
git check-attr linguist-generated <path>:true.unspecified:bundle/direct/dresources/apitypes.yml,bundle/internal/schema/annotations.yml,bundle/internal/schema/annotations_openapi_overrides.yml,bundle/docsgen/{main.go,README.md},bundle/schema/schema.go,internal/mocks/README.md,python/databricks/bundles/{__init__.py,build.py,core/**}..gitattributesfiles (TF schema, validation generated,bundle/schema/jsonschema.json, acceptanceout.test.toml) still resolve correctly../task wspasses.This PR was written by Claude Code.