Five findings (F6-F10) and a methodology correction. The twelve test tickets the same
read produced are #192-#203, with #204 for four candidates whose cost was not obviously
worth paying; nothing here is work, by this document's standing rule.
The correction is the part worth carrying forward. Wave 3 filtered candidates on `mi > 0`
and CLAUDE.md recommended it. That filter fails in both directions. It over-reports on
Compose: JoinScreen.kt:222 reads mi=10 and also ci=38, and JoinStateAffordancesTest
already clicks that Save button and asserts save:joined.mp4 -- the missed instructions are
the synthesized $changed/$dirty recomposition-skip path, the same codegen this repo
already knew inflated the branch count, showing up in the instruction count too. Every
onClick lambda flagged that way turned out to be covered at method level.
It under-reports on the case that mattered more. ConversionViewModel.cancel() and
JoinViewModel.cancel() miss no line at all, so no line-level filter can see them -- yet
only the null arm of activeWorkId?.let(workManager::cancelWorkById) had ever been entered,
and nothing in 584 tests connected the Cancel button to WorkManager. That is #192, and it
needs `ci > 0 && mb > 0` at method level to surface. Use both filters; `ci == 0` alone is
JaCoCo's own missed-line definition and needs no judgement, which is why it is the first.
The five findings are what a test would not fix. F6: four more unreachable arms, each
traced to the upstream guard that makes it so, one of which (ConversionRouter:214-217)
carries a KDoc describing a hazard :117 already removed. F7: probeWithExtractor's catch is
unreachable for the same reason probeForConcat's is -- the measurement was on record for
one site and not the other, three lines apart in the same file. F8: three more dead
members and six unused defaults. F9: both getForegroundInfo overrides are dead because
getForegroundInfoAsync is only called for expedited work and nothing sets it -- which
sharpens #88's close rather than reopening it. F10: three arms that ARE reachable and
still cannot be made to bite, recorded because all three were picked up as candidates and
put down again.
Six of the ten findings are now "no action" or "not a test gap", and that shape is the
honest summary of what is left: arms nothing can reach, members nothing calls, and arms a
test can reach but not pin. A coverage number tells none of them apart.
One close is qualified rather than overturned. #86 and #133 ruled AndroidDeviceCodecs.probe()
out through ShadowMediaCodecList, on the grounds that MediaCodecInfoBuilder cannot set
isAlias or canonicalName. A pure seam does not have that constraint and #133 did not
evaluate one, so #194 is a different mechanism, not a third run of the same spike -- and
its argument is not coverage but that the runCatching fallback logs "assuming permissive"
while returning empty sets, which makes canEncode and canDecode answer no for everything.
Documentation only: no Kotlin, Gradle or shell file is touched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
JMR-devandClaude Opus 5
223fe6deea