Fix Mac native screenshots to capture the genuine Metal display pipeline by shai-almog · Pull Request #5159 · codenameone/CodenameOne

shai-almog · 2026-06-03T13:24:28Z

Problem

The cn1ss screenshot helper special-cased Mac native (Mac Catalyst) to bypass Display.screenshot() and instead paint the current form into an off-screen mutable image. This hid a real framework bug and silently degraded coverage:

The off-screen re-paint can't capture native peers, so the committed BrowserComponent mac-native baseline was a blank page (vs the real rendered HTML on the iOS-Metal baseline).
Any screenshot() staleness/correctness issue on Catalyst went completely untested, because the test no longer exercised that path.

Root cause

CodenameOne_GLViewController.drawFrame early-returns whenever the app is not UIApplicationStateActive, doing no render at all:

if ([UIApplication sharedApplication].applicationState != UIApplicationStateActive) return;

That gate is correct on iOS (a backgrounded app must not touch the GPU) but wrong on Mac Catalyst, where an app keeps drawing its window while it isn't the focused application. During a long headless screenshot run the Catalyst app loses "active" status, after which every drawFrame is a no-op and the Metal screenTexture (which the screenshot reads back) freezes on the last active frame for the rest of the suite. The paint hack sidestepped this by never touching the screen pipeline at all.

Fix

All changes are gated to Mac Catalyst / desktop, so iOS device + simulator paths are untouched.

CodenameOne_GLViewController.m (root cause) — on Catalyst, only skip rendering when the app is truly backgrounded, not merely inactive, so screenTexture stays current.
IOSNative.m — capture by reading the Metal screenTexture back into a CGImage (the exact pixels presentFramebuffer blits to the drawable), via a blit on the renderer's own command queue (FIFO-ordered, then waitUntilCompleted). Genuine on-screen content, no CADisplayLink dependency.
IOSImplementation.screenshot() — before the native capture, force the current form through the real EDT screen-render path (paintComponent to the screen graphics + flushGraphics, exactly what paintDirty() runs) so a static form's texture is fresh. Genuine Metal draw path, not an off-screen re-paint.
Cn1ssDeviceRunnerHelper — remove the off-screen paint hack; Mac native now uses Display.screenshot() like every other port.

Re-baselined scripts/mac-native/screenshots/ from the genuine pipeline (now full 2× display resolution; BrowserComponent shows the real title bar with a blocked/black peer area, which is expected on Catalyst).

Verification (local Mac Catalyst build)

Stage	distinct frames / 122
Hack removed, readback only	~13 (mass freeze)
+ forced genuine render	111 (tail still froze)
+ `drawFrame` gate fix	122 / 122

Independent strict-comparison run against the new baseline (CN1SS_FAIL_ON_MISMATCH=1): 122/122 equal — deterministic. Spot-checked visually: graphics-fill-rect shows its real rects (was a MorphTransition frame), tail theme screens render their own content (were frozen on FAB-dark), BrowserComponent shows the genuine title bar with a black peer area.

🤖 Generated with Claude Code

The cn1ss screenshot helper special-cased Mac native (Mac Catalyst) to bypass Display.screenshot() and instead paint the current form into an off-screen mutable image. That hid a real framework bug and silently degraded coverage: because the off-screen re-paint cannot capture native peers, the committed BrowserComponent mac-native baseline was a blank page, and any screenshot() staleness on Catalyst went untested. Root cause: CodenameOne_GLViewController.drawFrame early-returns whenever the app is not UIApplicationStateActive, doing no render at all. That gate is correct on iOS (a backgrounded app must not touch the GPU) but wrong on Mac Catalyst, where an app keeps drawing its window while it is not the focused application. During a long headless screenshot run the Catalyst app loses "active" status, after which every drawFrame is a no-op and the Metal screenTexture (which the screenshot reads back) freezes on the last active frame for the rest of the suite. Fix, all gated to Mac Catalyst / desktop so iOS device + simulator paths are untouched: - CodenameOne_GLViewController.m: on Catalyst only skip rendering when the app is truly backgrounded, not merely inactive, so screenTexture stays current. (root cause) - IOSNative.m: capture by reading the Metal screenTexture back into a CGImage -- the exact pixels presentFramebuffer blits to the drawable -- via a blit on the renderer's own command queue (FIFO-ordered, then waitUntilCompleted). This is the genuine on-screen content and does not depend on a CADisplayLink present cycle (which never fires headless). - IOSImplementation.screenshot(): before the native capture, force the current form through the real EDT screen-render path (paintComponent to the screen graphics + flushGraphics, exactly what paintDirty() runs) so a static form's texture is fresh. Genuine Metal draw path, not an off-screen re-paint. - Cn1ssDeviceRunnerHelper: remove the off-screen paintComponent hack; Mac native now uses Display.screenshot() like every other port. Re-baselined scripts/mac-native/screenshots from the genuine pipeline (now full 2x display resolution; BrowserComponent shows the real title bar with a blocked/black peer area, which is expected on Catalyst). Verified on a local Mac Catalyst build: 122/122 screenshots render as distinct, genuinely-rendered frames (was ~13 distinct due to the freeze), and an independent strict-comparison run reports 122/122 equal. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

shai-almog · 2026-06-03T13:49:12Z

Compared 122 screenshots: 122 matched.

Native Android coverage

📊 Line coverage: 12.93% (7708/59617 lines covered) [HTML preview] (artifact android-coverage-report, jacocoAndroidReport/html/index.html)
- Other counters: instruction 10.48% (38248/364885), branch 4.50% (1547/34396), complexity 5.50% (1815/32971), method 9.65% (1488/15413), class 15.79% (341/2159)
- Lowest covered classes
  - kotlin.collections.kotlin.collections.ArraysKt___ArraysKt – 0.00% (0/6327 lines covered)
  - kotlin.collections.unsigned.kotlin.collections.unsigned.UArraysKt___UArraysKt – 0.00% (0/2384 lines covered)
  - org.jacoco.agent.rt.internal_b6258fc.asm.org.jacoco.agent.rt.internal_b6258fc.asm.ClassReader – 0.00% (0/1519 lines covered)
  - kotlin.collections.kotlin.collections.CollectionsKt___CollectionsKt – 0.00% (0/1148 lines covered)
  - org.jacoco.agent.rt.internal_b6258fc.asm.org.jacoco.agent.rt.internal_b6258fc.asm.MethodWriter – 0.00% (0/923 lines covered)
  - kotlin.sequences.kotlin.sequences.SequencesKt___SequencesKt – 0.00% (0/730 lines covered)
  - kotlin.text.kotlin.text.StringsKt___StringsKt – 0.00% (0/623 lines covered)
  - org.jacoco.agent.rt.internal_b6258fc.asm.org.jacoco.agent.rt.internal_b6258fc.asm.Frame – 0.00% (0/564 lines covered)
  - kotlin.collections.kotlin.collections.ArraysKt___ArraysJvmKt – 0.00% (0/495 lines covered)
  - kotlinx.coroutines.kotlinx.coroutines.JobSupport – 0.00% (0/423 lines covered)

✅ Native Android screenshot tests passed.

Native Android coverage

📊 Line coverage: 12.93% (7708/59617 lines covered) [HTML preview] (artifact android-coverage-report, jacocoAndroidReport/html/index.html)
- Other counters: instruction 10.48% (38248/364885), branch 4.50% (1547/34396), complexity 5.50% (1815/32971), method 9.65% (1488/15413), class 15.79% (341/2159)
- Lowest covered classes
  - kotlin.collections.kotlin.collections.ArraysKt___ArraysKt – 0.00% (0/6327 lines covered)
  - kotlin.collections.unsigned.kotlin.collections.unsigned.UArraysKt___UArraysKt – 0.00% (0/2384 lines covered)
  - org.jacoco.agent.rt.internal_b6258fc.asm.org.jacoco.agent.rt.internal_b6258fc.asm.ClassReader – 0.00% (0/1519 lines covered)
  - kotlin.collections.kotlin.collections.CollectionsKt___CollectionsKt – 0.00% (0/1148 lines covered)
  - org.jacoco.agent.rt.internal_b6258fc.asm.org.jacoco.agent.rt.internal_b6258fc.asm.MethodWriter – 0.00% (0/923 lines covered)
  - kotlin.sequences.kotlin.sequences.SequencesKt___SequencesKt – 0.00% (0/730 lines covered)
  - kotlin.text.kotlin.text.StringsKt___StringsKt – 0.00% (0/623 lines covered)
  - org.jacoco.agent.rt.internal_b6258fc.asm.org.jacoco.agent.rt.internal_b6258fc.asm.Frame – 0.00% (0/564 lines covered)
  - kotlin.collections.kotlin.collections.ArraysKt___ArraysJvmKt – 0.00% (0/495 lines covered)
  - kotlinx.coroutines.kotlinx.coroutines.JobSupport – 0.00% (0/423 lines covered)

Benchmark Results

Detailed Performance Metrics

Metric	Duration
Base64 payload size	8192 bytes
Base64 benchmark iterations	6000
Base64 native encode	812.000 ms
Base64 CN1 encode	224.000 ms
Base64 encode ratio (CN1/native)	0.276x (72.4% faster)
Base64 native decode	760.000 ms
Base64 CN1 decode	182.000 ms
Base64 decode ratio (CN1/native)	0.239x (76.1% faster)
Image encode benchmark status	skipped (SIMD unsupported)

github-actions · 2026-06-03T13:49:39Z

✅ Continuous Quality Report

Test & Coverage

✅ Tests: 2895 total, 0 failed, 1 skipped
📊 Line coverage: 52.41% [HTML preview] [Download]
- Lowest covered classes
  - com.codename1.io.oidc.OidcClient – 0.00%
  - com.codename1.ai.OpenAiClient – 0.00%
  - com.codename1.social.Auth0Connect – 0.00%
  - com.codename1.io.grpc.ProtoWriter – 0.00%
  - com.codename1.social.AppleSignIn – 0.00%
  - com.codename1.io.graphql.GraphQLSubscription – 0.00%
  - com.codename1.io.grpc.GrpcWeb – 0.00%
  - com.codename1.nfc.NdefMessage – 0.00%
  - com.codename1.io.grpc.ProtoReader – 0.00%
  - com.codename1.nfc.NdefRecord – 0.00%

Static Analysis

SpotBugs [Report archive]
- ✅ ByteCodeTranslator: 0 findings (no issues)
- ✅ android: 0 findings (no issues)
- ✅ codenameone-maven-plugin: 0 findings (no issues)
- ✅ core-unittests: 0 findings (no issues)
- ✅ ios: 0 findings (no issues)
✅ PMD: 0 findings (no issues) [Report archive]
✅ Checkstyle: 0 findings (no issues) [Report archive]

Generated automatically by the PR CI workflow.

shai-almog · 2026-06-03T14:11:07Z

Compared 61 screenshots: 61 matched.
✅ Native Mac screenshot tests passed.

Benchmark Results

VM Translation Time: 0 seconds
Compilation Time: 97 seconds

shai-almog · 2026-06-03T14:20:24Z

Compared 94 screenshots: 94 matched.
✅ JavaScript-port screenshot tests passed.

shai-almog · 2026-06-03T14:33:25Z

Compared 122 screenshots: 122 matched.
✅ Native iOS screenshot tests passed.

Benchmark Results

VM Translation Time: 0 seconds
Compilation Time: 279 seconds

Build and Run Timing

Metric	Duration
Simulator Boot	72000 ms
Simulator Boot (Run)	1000 ms
App Install	14000 ms
App Launch	8000 ms
Test Execution	340000 ms

Detailed Performance Metrics

Metric	Duration
Base64 payload size	8192 bytes
Base64 benchmark iterations	6000
Base64 native encode	583.000 ms
Base64 CN1 encode	1449.000 ms
Base64 encode ratio (CN1/native)	2.485x (148.5% slower)
Base64 native decode	350.000 ms
Base64 CN1 decode	1075.000 ms
Base64 decode ratio (CN1/native)	3.071x (207.1% slower)
Base64 SIMD encode	507.000 ms
Base64 encode ratio (SIMD/native)	0.870x (13.0% faster)
Base64 encode ratio (SIMD/CN1)	0.350x (65.0% faster)
Base64 SIMD decode	503.000 ms
Base64 decode ratio (SIMD/native)	1.437x (43.7% slower)
Base64 decode ratio (SIMD/CN1)	0.468x (53.2% faster)
Image encode benchmark iterations	100
Image createMask (SIMD off)	240.000 ms
Image createMask (SIMD on)	13.000 ms
Image createMask ratio (SIMD on/off)	0.054x (94.6% faster)
Image applyMask (SIMD off)	288.000 ms
Image applyMask (SIMD on)	434.000 ms
Image applyMask ratio (SIMD on/off)	1.507x (50.7% slower)
Image modifyAlpha (SIMD off)	407.000 ms
Image modifyAlpha (SIMD on)	66.000 ms
Image modifyAlpha ratio (SIMD on/off)	0.162x (83.8% faster)
Image modifyAlpha removeColor (SIMD off)	198.000 ms
Image modifyAlpha removeColor (SIMD on)	104.000 ms
Image modifyAlpha removeColor ratio (SIMD on/off)	0.525x (47.5% faster)
Image PNG encode (SIMD off)	1099.000 ms
Image PNG encode (SIMD on)	2425.000 ms
Image PNG encode ratio (SIMD on/off)	2.207x (120.7% slower)
Image JPEG encode	837.000 ms

shai-almog · 2026-06-03T14:42:33Z

Compared 52 screenshots: 52 matched.
✅ Native iOS Metal screenshot tests passed.

Benchmark Results

VM Translation Time: 0 seconds
Compilation Time: 204 seconds

Build and Run Timing

Metric	Duration
Simulator Boot	57000 ms
Simulator Boot (Run)	1000 ms
App Install	11000 ms
App Launch	14000 ms
Test Execution	1504000 ms

…ine-screenshot-capture # Conflicts: # scripts/mac-native/screenshots/LightweightPickerButtons.png # scripts/mac-native/screenshots/LightweightPickerButtons_above_center.png # scripts/mac-native/screenshots/LightweightPickerButtons_below_right.png # scripts/mac-native/screenshots/LightweightPickerButtons_between_mixed.png # scripts/mac-native/screenshots/ValidatorLightweightPicker.png

The genuine screenshot pipeline captures at the host's native scale; my local Retina Mac produced 2x (2048x1370) images while the macos CI runner renders at 1x (1024x685), and GPU rasterization differs between them, so locally-generated goldens cannot pixel-match CI. Adopt the CI runner's own 1024x685 output as the golden set (the authoritative reference for the comparison), restoring the established 1x baseline resolution. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

shai-almog and others added 2 commits June 3, 2026 18:06

shai-almog merged commit 9d9c95a into master Jun 3, 2026
29 of 32 checks passed

shai-almog deleted the fix/mac-native-genuine-screenshot-capture branch June 3, 2026 17:35

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix Mac native screenshots to capture the genuine Metal display pipeline#5159

Fix Mac native screenshots to capture the genuine Metal display pipeline#5159
shai-almog merged 3 commits into
masterfrom
fix/mac-native-genuine-screenshot-capture

shai-almog commented Jun 3, 2026

Uh oh!

shai-almog commented Jun 3, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Jun 3, 2026 •

edited

Loading

Uh oh!

shai-almog commented Jun 3, 2026 •

edited

Loading

Uh oh!

shai-almog commented Jun 3, 2026 •

edited

Loading

Uh oh!

shai-almog commented Jun 3, 2026 •

edited

Loading

Uh oh!

shai-almog commented Jun 3, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

shai-almog commented Jun 3, 2026

Problem

Root cause

Fix

Verification (local Mac Catalyst build)

Uh oh!

shai-almog commented Jun 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Native Android coverage

Native Android coverage

Benchmark Results

Detailed Performance Metrics

Uh oh!

github-actions Bot commented Jun 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

✅ Continuous Quality Report

Test & Coverage

Static Analysis

Uh oh!

shai-almog commented Jun 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Benchmark Results

Uh oh!

shai-almog commented Jun 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

shai-almog commented Jun 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Benchmark Results

Build and Run Timing

Detailed Performance Metrics

Uh oh!

shai-almog commented Jun 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Benchmark Results

Build and Run Timing

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

shai-almog commented Jun 3, 2026 •

edited

Loading

github-actions Bot commented Jun 3, 2026 •

edited

Loading

shai-almog commented Jun 3, 2026 •

edited

Loading

shai-almog commented Jun 3, 2026 •

edited

Loading

shai-almog commented Jun 3, 2026 •

edited

Loading

shai-almog commented Jun 3, 2026 •

edited

Loading