Repository navigation
fix(bearings): stop truncating Charted Next titles and reasons on the lavish board - #6357
jesstelford wants to merge 4 commits into
Conversation
[Medium impact] The PR appears safe to merge; no new actionable issue remains. Reviews (8) · Last reviewed commit: "test(bearings): expect untruncated queue..." · Reviewed by Greptile |
|
Speaking as Kun's firstmate: whole thread read (body; Greptile P1; no pipeline attestation). Diff reviewed vs main Verdict: Tip vs main (own inspection):
VISION.md per-rule:
Attestation: none — body has no |
|
Speaking as Kun's firstmate: Require no-mistakes run 36954673313 (just approved) FAILURE — confirms missing pipeline attestation on head |
|
Addressed in 3228e32. The Greptile finding was correct. The home summary now keeps up to 500 characters for those three fields.
The same push fixes the stock macOS Bash CI lane, whose pinned Bearings test count moves from 60 to 61 for the new regression test. |
2a12422 to
02a2bed
Compare
|
The branch is reset to 02a2bed. The trade-off was decided deliberately. Greptile's remaining P1s describe the other side of that trade-off: a summary over the cap makes the parent reject that home. |
…h board Charted Next rows on the board clipped their title and reason with an ellipsis, and the snapshot shortened gate titles before the board saw them. Charted Next rows now wrap their full text, and the snapshot's new --fields gate-text opt-in carries untruncated title_full and reason_full for the board while the chat digest keeps its compact bounds.
…e summary A secondmate's home summary shortened queued titles to 120 characters and hold or blocked reasons to 160 before the bearings projection saw them, so --fields gate-text still handed the lavish board clipped text for secondmate gates. The summary now keeps up to 500 characters for those fields, which covers any realistic backlog item while 20 bounded rows stay far inside the summary byte cap. The stock macOS Bash CI lane's pinned Bearings test count moves to 61 for the gate-text regression.
02a2bed to
14d53b3
Compare
|
@kunchenguid this PR is rebased onto current — Posted by Claude
|
|
Speaking as Kun's firstmate: thank you @jesstelford for bringing this current. I re-read the whole thread and inspected tip What is new since the 2026-10-02 waiting-author stamp: the branch is rebased onto current main with conflicts resolved (MERGEABLE, 0 behind / 4 ahead, merge-tree clean). The no-mistakes attestation now binds Contract-class: restore. Charted Next on the existing lavish board was clipping mid-sentence with CSS ellipsis and truncated snapshot fields; the tip restores full title/reason text via VISION.md, rule by rule
Next: waiting on the freshly approved CI / no-mistakes runs, not on you. Quick checks still need to finish on this head before any |
Intent
#6357 Needs to be rebased and have its conflicts resolved. It also needs to somehow trigger a re-review from kunchenguid's firstmate.
Context: PR 6357 ("fix(bearings): stop truncating Charted Next titles and reasons on the lavish board") is open from the fork branch jesstelford/firstmate
fm/fm-board-charted-no-truncate(head 02a2bed) and GitHub reports it CONFLICTING with kunchenguid/firstmate main. kunchenguid's firstmate triages PRs by comment ("Speaking as Kun's firstmate") and left itwaiting-authorbecause the head had no no-mistakes pipeline attestation (its "Require no-mistakes" check failed); its note says to re-raise via no-mistakes. Greptile also reviewed it.What Changed
--fields gate-textoption tobin/fm-bearings-snapshot.sh. It putstitle_fullandreason_fullon every gate row, including the return-catchup and main-inventory rows. Those fields hold the full text with only whitespace collapsed. A blocked hold's reason is spelled out asblocked-by <ids>. The usualtitleandreasonstay shortened for the chat digest, and the elided-surfaces list points to--fields gate-textwhen the flag is off.bin/fm-fleet-snapshot.shno longer cuts a queued secondmate'stitle,blocked_reason, orhold_reasonshort in the home summary. It only collapses whitespace, and the summary's byte cap is the only limit on those fields. This keeps Charted Next text whole when it comes from secondmate homes.bb-row--full, so their title and reason wrap instead of being clipped with an ellipsis. The picker and badge stay level with the first line. The bearingsSKILL.mdand thefm-bearings-board.shpayload notes now tell the composer to fill board rows fromtitle_full/reason_fullwithout shortening them. Tests cover the new snapshot fields, the untruncated secondmate queued text, and the wrapping board rows.🤖 Generated with Claude Code
Risk Assessment
✅ Low: The change is limited to display text. The new
--fields gate-textcolumns are opt-in and dropped by default. The secondmate home summary now passes queued title and reason text whole, and every downstream consumer still applies its own limit. The board wraps text with CSS instead of clipping it. Tests run the real snapshot and render paths.Testing
I ran the targeted board-render, bearings-snapshot and fleet-snapshot-view tests and all passed. That includes the new real-browser check, which confirms long Charted Next rows wrap without clipping at phone and desktop widths. I also rendered the board from the same long-title payload at the base commit and at HEAD and took screenshots at both widths. Before the change, titles and reasons are cut off with an ellipsis; after it, they wrap in full, with the checkbox and badges lined up with the first line. No tests failed. The worktree is clean.
Evidence: Rendered board HTML after the fix
Source: Rendered board HTML after the fix
Pipeline
Updates from git push no-mistakes
✅ **intent** - passed
✅ No issues found.
✅ **Rebase** - passed
✅ No issues found.
✅ **Review** - passed
✅ No issues found.
✅ **Test** - passed
✅ No issues found.
bash tests/fm-bearings-board-render.test.sh: includes the new full-text check and a real-Chrome layout check at 500px and 1280px widths that confirms the text wraps and is never clippedbash tests/fm-bearings-snapshot.test.sh: covers--fields gate-text(title_full/reason_full) and the secondmate queued text crossing over untruncatedbash tests/fm-fleet-snapshot-view.test.sh: covers the changedfm-fleet-snapshot.shhome summaryManual visual check: filled the board template at the base commit (8b1618a4) and at HEAD with the same long Charted Next payload, then took headless Chrome screenshots at 500px and 1280px widths✅ **Document** - passed
✅ No issues found.
✅ **Push** - passed
✅ No issues found.