Skip to content

fix(filters): use tail_lines for just/mise/task to preserve failure summaries (#1065)#2481

Open
ousamabenyounes wants to merge 2 commits into
rtk-ai:developfrom
ousamabenyounes:fix/issue-1065
Open

fix(filters): use tail_lines for just/mise/task to preserve failure summaries (#1065)#2481
ousamabenyounes wants to merge 2 commits into
rtk-ai:developfrom
ousamabenyounes:fix/issue-1065

Conversation

@ousamabenyounes

Copy link
Copy Markdown
Contributor

Fixes #1065.

The just, mise and task TOML delegator filters used max_lines = 50, which truncated wrapped tool output at 50 lines. When the wrapped command printed more than that, the pass/fail summary at the tail (e.g. FAILED: test_x, result: N failures) was dropped — hiding test failures.

Fix

Replace max_lines = 50 with tail_lines = 200 in all three filters so the end of the output (where test runners print their summary) is always preserved. Adds an inline [[tests.<filter>]] regression case per filter (input deliberately exceeds the old 50-line cap and ends with a failure summary).

Test verification (RED → GREEN)

Run via rtk verify --filter <name> (the project's TOML inline-test runner).

RED — config reverted to max_lines = 50:

--- just ---  3/4 tests passed   (tail summary truncated: "... (11 lines truncated)")
--- mise ---  3/4 tests passed
--- task ---  3/4 tests passed
rtk: 1 test(s) failed

GREEN — with tail_lines = 200:

--- just ---  4/4 tests passed
--- mise ---  4/4 tests passed
--- task ---  4/4 tests passed

(Re-proposes the accidentally-closed #1191, rebased onto current develop.)

ousamabenyounes and others added 2 commits June 17, 2026 22:38
max_lines = 50 was truncating verbose wrapped-tool output (pytest,
cargo test) from the front, hiding failure summaries that appear at
the end. Replace with tail_lines = 200 so the last 200 lines of
output — where failures are reported — are always visible.

Closes rtk-ai#1065

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…l_lines

Address @YOMXXX review on rtk-ai#1191. Prior fixture (~9 lines) did not
exceed the old `max_lines = 50` cap, so the test passed even under
the regressed behavior. New fixtures use 58 progress lines + 3-line
failure summary at the tail (61 total). Verified locally: temporarily
reverting `tail_lines = 200` back to `max_lines = 50` makes all three
tests FAIL (summary truncated at line 50), restoring the new setting
makes them PASS — proving the assertion actually protects the change.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
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.

TOML delegator filters (just/mise/task) truncate wrapped tool output, hiding test failures

1 participant