PackageBoundaryTests: fix pack hang; trigger+repeat composition tests #38
Loading…
Reference in a new issue
No description provided.
Delete branch "maint-pkgboundary-trigger-repeat-20261111"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
PackageBoundaryTests hang. Root cause: the test shells out to
dotnet packwith stdout/stderr redirected.packleaves VBCSCompiler and MSBuild node-reuse processes alive; they inherit the redirected pipes, soReadToEndAsyncnever sees EOF even thoughpackexited and the nupkg was written (observed: nupkg present in the temp dir, test silent for 5+ min). Fix: pack withUseSharedCompilation=false,-nodeReuse:false,--disable-build-serversand the matching env vars;RunAsyncnow has a 5 min process timeout (kills the tree) and a 10 s pipe-drain grace, each failing with a message naming the cause. Both tests now pass in ~11 s.Trigger + repeat.
TriggerRepeatCompositionTestsrenders a voice withtrigger=0.5underpatch repeat=0.05(shorter than sustain+decay) through native Faust and asserts: the envelope releases inside each trigger window and restarts at each trigger; pitch falls back at each repeat boundary (with a no-repeat control that climbs); the repeat grid restarts from the trigger. Plus one lowering-text pin of the two ages.No sound or timbre changed; tests only.
Recommendation (not applied), explosion repeat. Do not restore
patch repeaton the explosion patch on lowering grounds: its only repeat-restarted parameters arepitch_ramp=0.032on a noise voice at freq 20 (no arpeggio), so restarting the slide every 0.175 s is near-inaudible and would change the render. Leave it until Meta's ears (listen-sfxr-repeat-clips-20261111) say whether the original stutter mattered.🤖 Generated with Claude Code