-
[OPIK-7711] [QA][FE] test: add E2E coverage for multi-trace deletion (#7707)
frostbyte_neo 发布于
2026-08-04 13:25:03 +00:00 - [OPIK-7711] [QA][FE] test: add E2E coverage for multi-trace deletion
Trace deletion was untested end to end despite being destructive and
irreversible. Adds two mirrored tests: one deletes through the UI and
verifies the traces are gone from the API, the other deletes through the
API and verifies the table reflects it.The traces bulk-delete button is icon-only with no accessible name, so it
gets a data-testid matching the existing dataset-items convention.Co-Authored-By: Claude Opus 5 noreply@anthropic.com
- fix(e2e): use taxonomy-registered capability tags for trace deletion
tag-lint rejected the invented @cap: names. The taxonomy already reserved
traces.delete-traces for this work; use it for the UI flow and register a
delete-traces-api sibling for the API-side one, marking both covered.Co-Authored-By: Claude Opus 5 noreply@anthropic.com
- docs(skills): require taxonomy-registered tags in the e2e skills
Both skills documented a stale feature-tag vocabulary (@datasets,
@trace-explore) that predates the @area:/@cap: grammar the tag-lint CI job
enforces, so following them produced invented tag names and a red build.Document the real rule: @area:/@cap: come from coverage/taxonomy.yaml, the
capability is usually already reserved there as covered:false, and the
taxonomy is updated in the same change. Add tag_lint.py plus the
feature-directory and tsc runs to the verification steps of both.Co-Authored-By: Claude Opus 5 noreply@anthropic.com
- refactor(e2e): use the @e2e alias for the SdkClient type import
Addresses review feedback. Also expands the traceRow comment to record why
data-row-id is the right hook here: it is set from the row model by the
shared DataTable and there is no text alternative, since the trace id is a
filter field rather than a rendered column.Co-Authored-By: Claude Opus 5 noreply@anthropic.com
Co-authored-by: Claude Opus 5 noreply@anthropic.com
下载附件