Skip to content

feat(report): fold long tables after ten rows under a "Show N more" label - #67

Merged
babaliauskas merged 2 commits into
mainfrom
feat/show-more-rows
Oct 22, 2026
Merged

babaliauskas merged 2 commits into
mainfrom
feat/show-more-rows

Conversation

@babaliauskas

Copy link
Copy Markdown
Collaborator

Stacked on #66; merge that first. Until then this diff also shows #66's commit (9db862e). This PR's own change is 5cb8fe2.

What this changes

The Per-example breakdown and Trace rules broken tables listed every row: 21 and 16 on a real planner run, 100+ on a full captured suite.

  • Folding. Each table now shows its first 10 rows (reports.html.TABLE_ROWS_SHOWN) and folds the rest, in the same table, under a "Show N more examples" / "Show N more broken rules" label. The label then reads "Show fewer".
  • Order. The breakdown is already sorted regressed-first and worst-first, so the rows that matter stay visible.
  • No script. The report is JavaScript-free by design. A visually hidden checkbox before the table drives a sibling-selector rule that hides .row-more rows. It works offline and from an email attachment, and the checkbox keeps keyboard focus, with a focus ring on the label.
  • Short tables. A table of 10 rows or fewer gets no toggle.
  • Print. A printed report shows every row: @media print forces .row-more visible and hides the label.

Template, CSS and one constant only; no data or report.json change. CHANGELOG, DOCS.md and llms-full.txt are updated.

Test plan

  • New tests in TestLongTablesShowTheirFirstRows, each seen failing first:
    • 12 rows → 2 folded plus "Show 2 more examples" and "Show fewer";
    • 11 rows → the singular "Show 1 more example";
    • 10 rows → no toggle;
    • the trace-rules table folds the same way ("Show 3 more broken rules");
    • the print stylesheet shows .row-more.
  • make ci: ruff, ruff format, mypy --strict, and pytest (3401 passed, 95.3% coverage)
  • pre-commit run --all-files
  • Rendered the real planner run (21 examples, 16 broken rules) and checked headless-Chromium screenshots of the folded and unfolded states

🤖 Generated with Claude Code

babaliauskas and others added 2 commits October 22, 2026 23:46
A card flagged for its wording alone showed both sides making the same tool call with the target's in a red box, which read as a tool failure. The box and a target's "(no tool calls)" are now red only when the pair's tool diff is non-empty, and a within-threshold finding sits in a neutral notice instead of the warning tint.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…abel

The per-example breakdown and the Trace rules broken table listed every row, 21 and 16 on a real planner run. They now show the first 10 (the breakdown is sorted regressed and worst first) and fold the rest in the same table under a label driven by a hidden checkbox: no script, keyboard-focusable, and a printed report shows every row.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@babaliauskas
babaliauskas merged commit fa08fcf into main Oct 22, 2026
4 checks passed
Sign up for free to 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