Skip to content

StreamVault is a TV-first free IPTV player for Android TV built with Kotlin, Jetpack Compose, Room, Hilt, and Media3.

docs(live): record incremental performance samples

David Nashash69dcb64ede04
diff --git a/docs/COMPOSE_REDUCTION_PHASE5_LIVE_REPORT.md b/docs/COMPOSE_REDUCTION_PHASE5_LIVE_REPORT.md
index 341907c7..a105c186 100644
--- a/docs/COMPOSE_REDUCTION_PHASE5_LIVE_REPORT.md
+++ b/docs/COMPOSE_REDUCTION_PHASE5_LIVE_REPORT.md
@@ -94,8 +94,12 @@ hierarchies are under `build/adb-validation/guide-journey-*` and
- Locale-by-locale resource parity is open. The feature has 245 default keys;
translated app locales remain in the app until values and placeholders are
compared. No app resources were deleted.
-- Golden baseline review, five-pair post-extraction performance isolation,
- clean/warm guardrails, and baseline/startup profile regeneration are open.
+- Golden baseline review, the formal paired performance target, clean/warm
+ guardrails, and baseline/startup profile regeneration are open. Five
+ post-extraction Live source-edit and five moved-test samples are recorded in
+ `validation/phase5_live/performance-after.md`; they establish the
+ post-extraction envelope but are not a comparable pre/post pair because the
+ available pre record is no-change rather than source-edit.
`verifyBaselineProfileSources` passes (46,249 baseline / 30,093 startup
rules), but `:app:generateBaselineProfile` hit a JVM native-memory OOM during
release Java compilation; generated files still contain stale app Home/EPG
diff --git a/docs/superpowers/plans/2026-08-28-live-feature-extraction.md b/docs/superpowers/plans/2026-08-28-live-feature-extraction.md
index 0e2c0bfc..bd7bd954 100644
--- a/docs/superpowers/plans/2026-08-28-live-feature-extraction.md
+++ b/docs/superpowers/plans/2026-08-28-live-feature-extraction.md
@@ -1224,7 +1224,7 @@ git commit -m "test(live): record feature runtime acceptance"
- Consumes: committed post-extraction ref and all Tasks 0-10 evidence.
- Produces: five-pair build comparison, task-isolation proof, regenerated profile evidence, final ownership report, and explicit remaining gates without a Phase 5 completion claim.
-- [ ] **Step 1: Capture five post-extraction live source-edit samples**
+- [x] **Step 1: Capture five post-extraction live source-edit samples**
In an isolated snapshot of the committed post-extraction ref, touch only
`feature/live/src/main/java/com/streamvault/feature/live/home/HomeScreen.kt`
@@ -1238,7 +1238,13 @@ Record wall time, profile path, task states, median, average, and p95. Prove no
`:feature:playback`, `:feature:provider`, or `:feature:settings` Kotlin compile
task executes.
-- [ ] **Step 2: Capture five post-extraction moved-test samples**
+Recorded in `validation/phase5_live/performance-after.md`: five successful
+post-extraction Live source-edit samples (median 12.828s; setup outlier
+42.404s). The formal paired pre/post target remains open because the available
+pre record is a no-change app baseline rather than an identical source-edit
+snapshot.
+
+- [x] **Step 2: Capture five post-extraction moved-test samples**
Touch only `feature/live/src/test/java/com/streamvault/feature/live/home/HomeViewModelTest.kt` and run five times:
@@ -1248,6 +1254,11 @@ Touch only `feature/live/src/test/java/com/streamvault/feature/live/home/HomeVie
Compute the same statistics and compare against Task 0. Report outliers and cache/setup state; do not claim the 20% target from incomparable samples.
+Recorded in `validation/phase5_live/performance-after.md`: five successful
+post-extraction Live test-edit samples (median 12.583s; setup outlier
+49.658s). No performance target is claimed without a comparable pre-edit
+snapshot.
+
- [ ] **Step 3: Capture clean and warm guardrails**
In isolated pre/post snapshots with equivalent Gradle/SDK/temp storage, run clean debug and warm no-change `:app:assembleDebug`. Record executed/from-cache/up-to-date task counts, median where repeated, and comparability limitations.