solidjs / solid / 41673859564

79%

main: 89%

push

github

fix(signals): optimistic list push after filter, keyed re-base settles by key (#4642) (#4656)

* fix(signals): optimistic list push after filter, and keyed re-base settling (#4642)

- optimisticView: compare the length guess against the clone, not the
  source, so a removal guess past the source's end no longer leaves a hole
  that the next push lands after (`<For>` got an undefined row).
- laneSet: a leaf the keyed re-base makes the lane's drops the frame's truth
  staging on it, so the lane's reveal no longer promotes it over the
  composed row (duplicated rows).
- dissolveLane (correction): a store leaf the lane composed lands the
  backing's truth beneath it, as a guess does, instead of committing the
  composed row by position (`_slotTruth`, optimistic.ts).
- dissolveLane (held, shown lane): an effect born in the parked pass keeps
  its first run, at the parent's landing (a row the keyed list built never
  rendered its title).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWNSSgtemsSYsaX2hXE

* test(signals): pin the first-load frame for #4642 (guess survives, re-based by key)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWNSSgtemsSYsaX2hXE

* docs(signals): D-34 clarification — an in-flight load is an older question (#4642)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWNSSgtemsSYsaX2hXE

* perf(signals): gate the dissolveLane slot-truth and first-run branches on guessed (#4642 review trim)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWNSSgtemsSYsaX2hXE

* fix(signals): keyed re-base frame bugs; OPTIMISTIC_WRITE_BEFORE_LOAD (#4642, #4656 review)

- An earlier action's refresh() no longer blocks the lanes later guesses
  compose in: a refresh re-asks the guesses' own question (D-34 ¶3), so a
  l... (continued)

807 of 1108 branches covered (72.83%)

Branch coverage included in aggregate %.

1132 of 1370 relevant lines covered (82.63%)

30.0 hits per line