0.1.5-beta1 KNOWN ISSUES AND ACCEPTANCE LIMITS - 2026-10-19

1 AUDIO PROCESSING HEADROOM
Heavy active-strip/effect workloads may glitch. No safe universal strip-count
limit is established. Short125BPM clapOFF/ON and requestedfive-minuteclapON
checks added0dropouts; quiet listening was normal. Guard remained active.
These are bounded tests on one device, not all-workload or overnight acceptance.
The rejected mono0.1.4candidate is not included.

2 SAVING DURING EXTERNAL MIDI
A user heard a glitch near autosave and observed one new dropout during an
editing interval. Causation is not instrumented proof, but source explicitly
documents audio interruption during flash erase. Existing AUTOSAVE OFF is
available; switch it while the DAW is stopped and save manually between takes.
Changing OFF schedules one last save to retain the preference.

3 TIMING UNDER INVESTIGATION
Perceived whole-kit hesitation improved with strips bypassed. Isolatedkick
recordings did not show clear extra fixed attack delay with its stripON.
167matchedattacks/take had about5.6..5.9ms offset standard deviation across
the MIDI/device/analog capture chain. Changing Reasonbuffer1024to256 did not
materially reduce it. Digital loopback reproduced all167attacks with one fixed
27sample shift and no added variation detected. Cause is unresolved; do not
present it as proven FM1-only jitter or apply a global30ms correction.

4 INDEPENDENT-LANE DISPLAY / COMPATIBILITY
Drum grid shading and playhead use the selected drum's length/position; they do
not show every row's independent clock accurately. EXPAND is a one-way timing
format conversion, not a reversible toggle. Older firmware may reject expanded
projects. The upstream one-knob BREAK filter remains deferred due to storage
conflict; existing BREAK stripHP/LP controls remain.

5 BACKUP AND COVERAGE
Current project export does not include the reusable strip-default bank, private
sample loops or calibration. Reusable defaults and sample assets need separate
preservation. No intentional physical power-loss testing or exhaustive restore
qualification is claimed. FullTRS/allCC/channel remap/thermal/maximumIRQ/overnight
soak acceptance remains open. Current measuredIRQhigh-water1052/7936bytes is a
bounded observation, not maximum possible stack usage.

6 FEATURES STILL PROPOSED
Per-drum input-note mapping, MIDI-file import with groove preservation, automatic
latency calibration, selectable compression/FX routing, visible overload warning
and expanded on-device About/limitations copy are not implemented by this build.