Repository navigation
Conversation
…tWallclock's answer CAP-079 showed three defects in 1.4.0 (ai-sessions/0099, CAP-079-FINDINGS.md §3.3, §5.2, §11): - the ring hold ended at 20 s while an untouched ring lasts ~63 s, so the Buds' stop status had no recipient and "Stopped on the bud" never appeared. RING_HOLD_MAX_MS is now 65 s (DECISIONS.md ADR-069 Update; no new bytes); - the red AncNotAllowed line outlived the Buds' push of a settable mode; applyAncState now clears it for 13 = 1-4; - SetWallclock's RESPONSE was not an RpcResult, so the experiment's F1 timed out; routeMaestro now classifies it (the release never sends it). Tests use CAP-079 bytes (Cap079Fixtures.kt: H1 2501, 2742, 3848-3862; H3 2559/2563); each change was mutation-checked (killed). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Three texts still described 1.3.x (CAP-079-FINDINGS.md §3.3): the ANC line's "Tapping a mode checks again first" (1.4 writes without a check), the battery (i)'s "read ... Message Stream" (L/R come from runtime info since ADR-068) and Find's "briefly claims" (a ring now holds the channel). Wording chosen by the maintainer in chat 2026-10-18; literal screen tests pin each one. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The fixes after CAP-079 ship as 1.4.1 on a new branch; 1.4.0 stays as built (the maintainer's choice in chat 2026-10-18). The experiment build becomes 1.4.1-exp.1 through its existing suffix. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…dose note - ADR-069 Update: the ring hold's limit is 65 s (CAP-079: an untouched ring lasts ~63 s), no new bytes. - ADR-060 Update (still a draft): the maintainer has no sound-level meter, so condition (i) becomes a meter-free test (10-dB steps in a test file). - PROTOCOL.md §6: a 🟡 item; in CAP-079 the day dose grew by the live value's integral in minutes (ratio 1/60 in six intervals). The tool behind the count, scripts/dosimeter_dose_vs_live.py, is committed (rule 4a). All approved by the maintainer in chat 2026-10-18. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A short run of what CAP-079 could not show: the ring stops inside the 65-s hold, the red line cleared by a push, the new texts, a Disconnect during a Connect (step-by-step: a slow Connect after Android's own Disconnect, then "Use different Buds"), F1's answer, and WELL-002 without a sound-level meter (four 10-dB pink-noise steps; the day dose against the live integral). Registered as planned; TalkBack stays in TODO.md (the maintainer's choice). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Records the fixed 1.4.0 known issues, the 65-s RING_HOLD_MAX_MS (§3.2, §5a, §6.0b), test section AC for CAP-080, TalkBack as a later TODO item and WELL-002's meter-free form. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The session record: the checkpoint answers verbatim, the mutation table, the gate, the meter-free WELL-002 follow-up and the maintainer's approvals. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Resolves the CHANGELOG and README conflicts with the 1.4.0 after-release docs (PR #72): the [1.4.1] block above the dated [1.4.0], and one 1.4.1 status line. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Conflicts in PROTOCOL.md (§6 notes of 0100 and 0101 both kept, §8 both rows) and ai-sessions/INDEX.md (rows 0100 and 0101 both 'complete'). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
6 tasks done
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stacked on #69 (
feature/0098-round-2, release 1.4.0). Do not merge before 1.4.0 is released and the hardware runCAP-080(Group BP) is analysed — the maintainer's order (chat 2026-10-18): 1.4.0 release →CAP-080+ 1.4.1 release →ai-sessions/0101. When #69 is merged this PR can be retargeted tomain; it then shows only its own commits if #69 is merged with a merge commit (after a squash or rebase merge this branch needs a rebase first).What changes (1.4.1, versionCode 10401) — nothing new on the wire
RING_HOLD_MAX_MS, DECISIONS.md ADR-069 Update): an untouched ring lasts ≈ 63 s (CAP-079§5.2), so the Buds' stop status now reaches the app and "Stopped on the bud at …" can appear.CAP-079§3.3).SetWallclock's answer is classified as anRpcResult, so F1 no longer reports a time-out (CAP-079§11). The release never sends it.Decisions in chat 2026-10-18: the version and branch, the 65-s hold (ADR-069 Update), the red line, the texts, F1's classification; later the same day an ADR-060 Update (still a draft):
WELL-002without a sound-level meter.PROTOCOL.md§6: one 🟡 item (the Dosimeter day dose = the live value integrated in minutes, ratio 1/60 in sixCAP-079intervals,scripts/dosimeter_dose_vs_live.py). No FACT promoted.Verification
CAP-079bytes (Cap079Fixtures.kt: H1 2501, 2742, 3848–3862; H3 2559/2563); literal screen tests for the three texts../gradlew --no-daemon --max-workers=1 clean assembleDebug testDebugUnitTest test lint :app:compileReleaseKotlin :app:processReleaseMainManifest lintRelease assembleRelease assembleExperiment— success, 5,407 tests, 0 failures; the experiment F1 test added afterwards passes in:data:testExperimentUnitTest.ExperimentRunner/ExperimentFrames, 0 Dosimeter strings), permissions unchanged (noINTERNET), no dependency changed.scripts/lint_docs.pyexit 0.CAP-080(skeleton incaptures/CAP-080-yyyy-MM-dd_HH-mm-ss_HH-mm-ss-Group_BP/: the ring stops, the red line, the texts, a Disconnect during a Connect, F1,WELL-002without a meter).Details:
ai-sessions/0100_FEATURE_RESULT_2026_10_18.md.🤖 Generated with Claude Code