solidjs / solid / 34888890288

78%

main: 89%

push

github

fix(signals): optimistic lanes release when nothing authoritative remains; shared render effects don't entangle unrelated writes (#3426, #3427, #3407) (#3434)

Three of gabbev's optimistic/transaction reports, one theme: the pure graph
does not wait where it is not forced to, and a hold is a fact about what a
pass observed — never about which reader happened to exist.

- #3426: the last async reader of an optimistic value unmounting mid-action
  releases the frame. `reporterBlocksSource` is now `sourceObserved`, shared
  by the settle verdict, the lane hold (`waitingTransition`) and the landing:
  the hold check prunes dead reporters itself. A live action parks its
  transaction without a verdict, so this is the only prune such a frame ever
  gets — the lane used to wait for a flight nobody observed.

- #3427: the action body ending starts the correction (`endOptimism`, called
  from `flush` after the heap and before the verdict). With the bodies over
  (`_acted`, no live `_actions`) and nothing authoritative up — no override
  node's own source in flight, no held plain load the action asked for —
  each override's truth supersedes it as an arriving differing truth does
  (A18): the graph re-derives now and the transaction settles when THAT
  lands. Before: wait for the obsolete lane-derived flight, flash the
  obsolete optimistic frame, revert, re-ask — a waterfall. Display is
  unchanged (override on screen until commit). A co-written pending flag
  stays through the action's own load (attribution-holds pin); optimistic
  store edits keep settle-then-revert; companions snap at settlement.

- #3407: a render effect's pass belongs to whatever dirtied it. `recompute`
  re-entered a stamped effect's transaction whenever another was active, so
  a sync `action` write to a signal that merely shared a hole with a held
  async merged into the async's transaction and waited (0:0 → 2:1) while the
  same plain write passed through. The stamp re-entry is memo-o... (continued)

519 of 790 branches covered (65.7%)

Branch coverage included in aggregate %.

772 of 1007 relevant lines covered (76.66%)

15.1 hits per line