Configure GitHub Actions for release builds by qiarie · Pull Request #44 · ohs-foundation/ohs-player-reference-client-app

qiarie · 2026-05-12T07:08:03Z

Fixes #43

All CI build checks pass e.g. Tests
Appropriate changes to documentation are included in the PR

ellykits · 2026-05-13T11:07:30Z

How do we intend to deploy the web version for the Kotlin Multiplatform @qiarie? Can you trigger a sample release for purposes of testing we can delete the tag afterwards.

Co-authored-by: Elly Kitoto <junkmailstoelly@gmail.com>

…anch is merged

Let callers pass values explicitly

- Closed shell-injection sink in preflight - Hardened env-var handling (VERSION_NAME/VERSION_CODE blank-safe - composePackageVersion fails loudly on malformed input) - Widened tag regex to Semantic Versioning-legal hyphen suffixes - Added count-check to the desktop matrix output - Updated spotless job fetch-depth: 0 so ratchetFrom = "origin/main" resolves

actions/checkout only populates the ref it checks out — refs/pull/<N>/merge for PRs - so refs/remotes/origin/main is never created and spotless' ratchetFrom = "origin/main" fails at task-graph construction. Switch the spotless job to fetch-depth: 0 and add an explicit fetch that creates the missing ref

- Upgrade Kotlin to 2.3.21 - Disable unstable JS and Android-host test tasks

…piles

qiarie · 2026-05-29T09:45:55Z

How do we intend to deploy the web version for the Kotlin Multiplatform @qiarie? Can you trigger a sample release for purposes of testing we can delete the tag afterwards.

Refer to this test pre-release Release v1.0.0-test.4

ellykits · 2026-06-02T08:30:34Z

+      # Each leg gets its own 16 GB runner, so the JS and Wasm test-executable
+      # compiles don't compete with each other for Kotlin daemon heap.


Delete comment

ellykits · 2026-06-02T08:32:01Z

+            # compileTestDevelopmentExecutableKotlinJs has historically OOMed
+            # at 6g; give the Kotlin daemon more room only for this leg.


Delete comment. This can be inferred from the setting.

ellykits · 2026-06-02T08:37:33Z

+            tasks: >-
+              :ohs-player-library:assembleAndroidMain
+              :ohs-player-reference-app:lintDebug


What's the intention for these two tasks? They serve different purposes. The assembleAndroidMain should be used to compile, process resources and package apk artifacts, whereas lintDebug is used for analyzing code, for bugs, performance issues, security and styling deviations. Those two cases differ I think you should decide on which one to keep (build and package or debug variant analysis).

Check out this comment:

ohs-player-library uses the androidKotlinMultiplatformLibrary plugin, which doesn't register lintDebug - assembleAndroidMain is the closest "compile + verify the Android variant" task it exposes. ohs-player-reference-app uses classic androidTarget and keeps lintDebug.

ohs-player-library uses the com.android.kotlin.multiplatform.library plugin, which doesn't register a lintDebug task, so lint isn't available there. assembleAndroidMain is the closest task it exposes that compiles and verifies the Android variant. ohs-player-reference-app is still a classic com.android.application + androidTarget, which does have lintDebug.

The goal of this leg isn't strictly "build" or strictly "analyze" - it's "verify the Android variant of both modules is healthy using whatever each module's plugin exposes": a compile/package check on the library and lint analysis on the app.

This asymmetry resolves itself during the AGP9 migration. The app will be split into a thin com.android.application module (androidApp) on top of KMP library modules, at which point ohs-player-reference-app becomes a com.android.kotlin.multiplatform.library and loses lintDebug - same as the library today. It will change to

tasks: >- :ohs-player-library:assembleAndroidMain :ohs-player-reference-app:assembleAndroidMain # both KMP modules: compile + verify :androidApp:lintDebug # lint on the one real Android app module

I'd lean towards leaving this leg as-is until the AGP9 migration is ready.

ellykits · 2026-06-02T08:38:34Z

+            # Same reason as the js leg: compileTestDevelopmentExecutableKotlinWasmJs
+            # exhausts 6g; needs more daemon heap, but only on this runner.


This can be inferred from the setting below, no need to add this comment, the alternative make it brief.

ellykits · 2026-06-02T08:40:01Z

+    # iOS targets can't be compiled on Linux. macos-latest billing is 10× so
+    # keep each leg focused on a single arch so the K/N compiler isn't
+    # fighting itself for the 14 GB runner. Debug link is sufficient to
+    # prove iOS code compiles and links; release link belongs in release.yml
+    # whenever iOS distribution is added.


Paraphrase this comment and make it brief.

ellykits · 2026-06-02T09:05:46Z

    sourceCompatibility = JavaVersion.VERSION_11
    targetCompatibility = JavaVersion.VERSION_11


Is it possible to upgrade this to at least Java 17. I'm okay doing that in a followup PR.

We can leave this to the AGP9 migration PR. We can also agree on the runtime floor to support then.

ellykits · 2026-06-02T09:07:52Z

+// WiX/MSI requires a strict MAJOR.MINOR.PATCH numeric version, so strip any
+// Semantic Version pre-release suffix (e.g. -alpha.1) from VERSION_NAME for the desktop
+// installers. Android's versionName keeps the suffix; this drift is intentional.
+// Uses the providers API so the configuration cache tracks VERSION_NAME as a
+// declared input.


Improve this comment and provide example how the new version name will be on the respective platforms with before and after (stripping) versions.

ellykits · 2026-06-02T09:11:24Z

+// Same as in ohs-player-library: Compose UI tests in commonTest don't run
+// cleanly on Android-host JVM (needs Robolectric + per-class @RunWith) nor in


I do not think we need Robolectric with Compose multiplatform to run the UI tests. Refer to Compose Multiplatform UI test. We should adopt that.

Co-authored-by: Elly Kitoto <junkmailstoelly@gmail.com>

- Update comments

- Disable failing JS/Wasm/Android-host test tasks only on CI - Rename helpers for clarity: envOrAbsent -> nonBlankEnv, envOrFile -> envOrKeystore - Expand build-script comments - Update release.yml workflow_dispatch dry-run description

Configure GitHub CI for release builds

3bec492

qiarie requested review from ellykits and ndegwamartin May 12, 2026 07:08

ellykits requested changes May 13, 2026

View reviewed changes

Comment thread .github/workflows/release.yml Outdated

Comment thread .github/workflows/release.yml

Comment thread .github/workflows/release.yml Outdated

Comment thread .github/workflows/release.yml

Comment thread ohs-player-reference-app/build.gradle.kts Outdated

qiarie and others added 3 commits May 13, 2026 14:33

Update .github/workflows/release.yml

76692db

Co-authored-by: Elly Kitoto <junkmailstoelly@gmail.com>

Update .github/workflows/release.yml

139528c

Co-authored-by: Elly Kitoto <junkmailstoelly@gmail.com>

Update ohs-player-reference-app/build.gradle.kts

36b8b85

Co-authored-by: Elly Kitoto <junkmailstoelly@gmail.com>

qiarie changed the title ~~Configure GitHub CI for release builds~~ Configure GitHub Actions for release builds May 13, 2026

Add preview-pages workflow to showcase Wasm build before github-ci br…

da2bb4a

…anch is merged

qiarie temporarily deployed to github-pages May 13, 2026 12:31 — with GitHub Actions Inactive

qiarie added 19 commits May 13, 2026 15:43

Merge remote-tracking branch 'origin/main' into github-ci

4a6ea1f

Bump up heap memory

89e5cc9

Update preview-pages.yml to use development distribution

2c01821

Increase Kotlin daemon heap memory to 10GB

9df5f0f

Run spotlessApply

27e0932

Fix syntax issue for preview-pages.yml

43e52a6

Add spotless CI check

716a061

Try G1GC for the OOM failures

d76caf0

Remove preview-pages task

6d03d52

Disable compile for Wasm

8dc2e12

Strip defaults off the interface declaration for Render()

a77ab4d

Let callers pass values explicitly

Merge remote-tracking branch 'origin/main' into github-ci

266d9b8

Retrigger CI

23f610d

Retrigger CI

5b20fed

Run spotlessApply

321f6c7

Fix spotless ratchet "No such reference 'origin/main'"

e519042

Stabilize build pipeline

6ad517d

- Upgrade Kotlin to 2.3.21 - Disable unstable JS and Android-host test tasks

qiarie added 14 commits May 28, 2026 13:23

Disable parallel execution for JS/Wasm on CI

11b1f49

Increase Kotlin daemon heap space

8b7bb34

Remove spotless checks in the check task

af52f34

Remove Spotless from check

9743eac

Increase gradle daemon heap size

9fd2bad

Bump CI Java version to 21

6dc58d0

Separate check tasks

109a9a6

Rename :ohs-player-library:lintDebug to assembleAndroidMain

7eb3a62

Increase heap size for Wasm/JS task

d9284c4

Switch iOS compile to use debug instead of release

c87128b

Remove AAB release option

e3b06b5

Set --max-workers=1 to serialize the library + reference-app Wasm com…

78b0ff1

…piles

Split iOS jobs by architecture

8e7d3da

Increase timeout for iOS compile (iosSimulatorArm64) to 45 minutes

aacb169

qiarie enabled auto-merge (squash) May 29, 2026 16:11

ellykits requested changes Jun 2, 2026

View reviewed changes

qiarie and others added 5 commits June 2, 2026 12:14

Update ohs-player-reference-app/build.gradle.kts

00a67ba

Co-authored-by: Elly Kitoto <junkmailstoelly@gmail.com>

Update ohs-player-reference-app/build.gradle.kts

d921aae

Co-authored-by: Elly Kitoto <junkmailstoelly@gmail.com>

Update TODOs

bfd8a38

- Update comments

Skip unsupported test tasks on CI

84f7914

		# Each leg gets its own 16 GB runner, so the JS and Wasm test-executable
		# compiles don't compete with each other for Kotlin daemon heap.

		# compileTestDevelopmentExecutableKotlinJs has historically OOMed
		# at 6g; give the Kotlin daemon more room only for this leg.

		# Same reason as the js leg: compileTestDevelopmentExecutableKotlinWasmJs
		# exhausts 6g; needs more daemon heap, but only on this runner.

		sourceCompatibility = JavaVersion.VERSION_11
		targetCompatibility = JavaVersion.VERSION_11

		// Same as in ohs-player-library: Compose UI tests in commonTest don't run
		// cleanly on Android-host JVM (needs Robolectric + per-class @RunWith) nor in

Conversation

qiarie commented May 12, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ellykits commented May 13, 2026

Uh oh!

qiarie commented May 29, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

qiarie Jun 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

qiarie Jun 2, 2026 •

edited

Loading