solidjs / solid / 38882946233

78%

main: 89%

push

github

fix(solid): judge SSR waterfall waits by when their work started (#4108)

* fix(solid): don't count lazy() code waits as an SSR data waterfall (#4105)

On a cold module cache, nested lazy() route levels resolve one level per
<Loading> render pass even when the router kicked every matched level's
import in the first pass, so SSR_BOUNDARY_WATERFALL reported parallel chunk
loads as sequential data waits and pointed at queries. lazy()/lazyModule()
now mark their import and asset-lookup promises (dev only), and the boundary
leaves passes that waited only on those out of the waterfall count. The
record's `passes` is unchanged; the finding gains `codePasses` when any were
left out.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018LY7F9czoRqdMEvk2QwPr1

* fix(solid): judge SSR waterfall waits by when their work started (#4105)

Replaces the lazy()-only exclusion with the client's origin rule. A
<Loading> boundary meets a nested route's levels one render pass at a time,
but the router starts every matched level's lazy() import and preload
together; SSR_BOUNDARY_WATERFALL counted each pass as a sequential wait. A
wait now counts only if something it waited on started at or after the
previous wait landed (`data.parallelWaits` names the ones left out).

The start stamps move from the attribution engine onto the core's
`OBSERVE.attribution` (`markFlight`, `flightOrigin`), so a router or cache
can stamp what it hands out without loading the engine, and the server can
read them. `attribution.markFlight` writes to the same map. lazy() and
lazyModule() stamp their import (and lazy()'s client read wrapper); the
server's async memo carries a stamp from its source promise onto the
deferred the boundary waits on; lazy asset lookups and their gates carry
their start. Prod artifacts are unchanged; the observe tier pays ~56 B br.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude... (continued)

756 of 1050 branches covered (72.0%)

Branch coverage included in aggregate %.

9 of 9 new or added lines in 2 files covered. (100.0%)

1096 of 1338 relevant lines covered (81.91%)

27.79 hits per line