StreamVault is a TV-first free IPTV player for Android TV built with Kotlin, Jetpack Compose, Room, Hilt, and Media3.
Phase 5 System extraction — post-extraction performance
Captured: 2026-09-04 on feature/improveCompose, after the System
presentation extraction. The paired pre-extraction baseline is recorded in
performance-before.md.
Protocol
- Gradle 8.12; Oracle JVM 21.0.5; Windows 11 amd64.
- Each sample used
--profile --no-daemon --console=plain --warning-mode=none. - The source and test measurements used reversible one-blank-line edits and restored the files byte-for-byte after each series.
- Source hash after restoration:
WelcomeScreen.kt = ca1c597ed3c8a3c4bfa47ce9749bd1c4e9e21ad5. - Test hash after restoration:
WelcomeViewModelTest.kt = 928b17738c55a0035b1adb8aacf6b031b4800ce8.
:feature:system:compileDebugKotlin source-edit samples
| Sample | Duration | Result |
|---|---|---|
| source-1 | 22.9892s | success |
| source-2 | 7.2796s | success |
| source-3 | 7.3984s | success |
| source-4 | 7.3602s | success |
| source-5 | 7.2687s | success |
Raw durations: 22.9892s, 7.2796s, 7.3984s, 7.3602s, 7.2687s. All-sample min/median/max: 7.2687s / 7.3602s / 22.9892s. The first sample is the configuration/cache warm-up outlier. Warm samples 2–5 have a median of 7.3199s.
:feature:system:compileDebugUnitTestKotlin test-edit samples
| Sample | Duration | Result |
|---|---|---|
| test-1 | 25.1138s | success |
| test-2 | 7.6541s | success |
| test-3 | 7.5883s | success |
| test-4 | 7.7130s | success |
| test-5 | 7.6373s | success |
Raw durations: 25.1138s, 7.6541s, 7.5883s, 7.7130s, 7.6373s. All-sample min/median/max: 7.5883s / 7.6373s / 25.1138s. The first sample is the configuration/cache warm-up outlier. Warm samples 2–5 have a median of 7.6457s.
Paired value gate
Compared with the pre-extraction warm medians in performance-before.md:
| Measurement | Before | After | Change | Gate |
|---|---|---|---|---|
| System source edit | 20.6311s | 7.3199s | 64.56% faster | pass (≥25%) |
| System unit-test compile | 18.1523s | 7.6457s | 57.88% faster | pass (≥20%) |
The results are comparable warm paired measurements. The first sample in each series is reported, but excluded from the paired median because it includes configuration/cache initialization.
Build guardrails
clean :app:assembleDebug: passed;BUILD SUCCESSFUL in 1m 9s, 255 actionable tasks, 113 executed, 142 from cache. This was a clean invocation with remote build-cache reuse.- Warm
:app:assembleDebug: passed;BUILD SUCCESSFUL in 14s, 242 actionable tasks, 1 executed, 241 up-to-date. :app:assembleBeta :app:assembleRelease: passed;BUILD SUCCESSFUL in 18m35s, 464 actionable tasks, 276 executed, 149 from cache, 39 up-to-date.- The app lint task remains red on pre-existing app-wide debt and is tracked separately in the final System report; feature lint passes.
Paired clean/warm guardrail
The same clean :app:assembleDebug and warm :app:assembleDebug commands were
run in a disposable rollback worktree at
cddef1526924634819c7fb27d15c8a49b3945a1d and in the current checkout. The
rollback worktree received only a copy of the active local SDK configuration;
the active checkout was not changed.
| Scenario | Rollback | Current | Change | Gate |
|---|---|---|---|---|
| clean assemble | 161.907s | 66.502s | 58.95% faster | pass (<=10% slower) |
| warm repeat 1 | 11.768s | 11.955s | 1.59% slower | pass |
| warm repeat 2 | 11.707s | 11.935s | 1.95% slower | pass |
| warm repeat 3 | 11.072s | 12.691s | 14.62% slower | noise sample |
Steady-state warm medians are 11.707s rollback and 11.955s current, a 2.12% slowdown. The first warm invocation after the clean build included task-set configuration-cache setup (22.862s rollback and 26.527s current), so it is reported separately and excluded from the steady-state median. The clean runs reused the remote build cache with 232 versus 255 actionable tasks; the current run executed 113 tasks versus 102 at rollback.
The incremental source/test thresholds, sibling Kotlin isolation, and the clean/warm build-overhead guardrail pass. Startup/profile status and the remaining device/production journeys are recorded in the System report.