chore: fix ios e2e workflow; upgrade to Xcode 16.4 #915
Conversation
WalkthroughWorkflows updated to centralize XCODE_VERSION, add NDK and Android toolchain setup and caching, and introduce per-project iOS E2E wiring (change detection, workspace/scheme validation, pod install, simulator creation/selection, and exported simulator IDs). Added devDependency @types/node to common/package.json. Changes
Sequence Diagram(s)sequenceDiagram
participant GH as GitHub Actions (e2e-ios)
participant Repo as Repository FS
participant Simctl as simctl / runtimes
participant Xcode as xcodebuild
participant E2E as E2E Test Runner
GH->>Repo: Read IOS_PROJECT_NAME & IOS_PROJECT_SCHEME
GH->>Repo: Validate workspace exists (app/ios/${IOS_PROJECT_NAME}.xcworkspace)
GH->>Repo: List schemes -> validate IOS_PROJECT_SCHEME
GH->>Simctl: List runtimes (Verify iOS Runtime)
alt runtime usable
GH->>Simctl: Find or create simulator -> boot -> wait for readiness
Simctl-->>GH: Return IOS_SIMULATOR_ID / IOS_SIMULATOR_NAME
GH->>GH: Export simulator env vars
end
GH->>Xcode: xcodebuild -workspace WORKSPACE_PATH -scheme IOS_PROJECT_SCHEME -destination id=IOS_SIMULATOR_ID
Xcode-->>Repo: Produce build artifacts
GH->>E2E: Run tests on simulator
E2E-->>GH: Return results
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes Possibly related PRs
Suggested reviewers
Poem
📜 Recent review detailsConfiguration used: .coderabbit.yaml 💡 Knowledge Base configuration:
You can enable these sources in your CodeRabbit configuration. 📒 Files selected for processing (1)
⏰ Context from checks skipped due to timeout of 300000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (2)
🔇 Additional comments (1)
✨ Finishing Touches🧪 Generate unit tests
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. CodeRabbit Commands (Invoked using PR/Issue comments)Type Other keywords and placeholders
Status, Documentation and Community
|
Giga Summary• Streamline iOS E2E workflow by removing redundant caches and setup steps.
Quality Assessment• Streamlined workflow and cache cleanup enhance efficiency. Quality Score: 7/10 (Threshold: 7/10) 💬 Detailed comments have been added to specific code changes. |
![giga-agent[bot]](https://avatars.githubusercontent.com/in/1202121?s=60&v=4){kind=small}
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 0
🔭 Outside diff range comments (2)
.github/workflows/mobile-e2e.yml (2)
197-199: Avoid curl | bash supply-chain risk; pin and verify Maestro installer.Piping remote scripts to shell is a medium+ risk in CI. Prefer a pinned installer with checksum verification or a package manager.
Options:
- Preferred: Install via Homebrew (version-pinned):
- - name: Install Maestro - if: steps.cache-maestro.outputs.cache-hit != 'true' - run: curl -Ls "https://get.maestro.mobile.dev" | bash + - name: Install Maestro + if: steps.cache-maestro.outputs.cache-hit != 'true' + run: | + brew update + brew install mobile-dev-inc/tap/[email protected]
- Or, download a specific release asset and verify checksum before executing (store/update the expected sha256 securely in the repo):
- - name: Install Maestro - if: steps.cache-maestro.outputs.cache-hit != 'true' - run: curl -Ls "https://get.maestro.mobile.dev" | bash + - name: Install Maestro + if: steps.cache-maestro.outputs.cache-hit != 'true' + run: | + VERSION="1.41.0" + curl -Lso maestro-install.sh "https://get.maestro.mobile.dev" + echo "${MAESTRO_INSTALL_SHA256} maestro-install.sh" | shasum -a 256 -c - + bash maestro-install.sh
286-293: Hard-coded iOS runtime string makes simulator creation brittle; use the detected RUNTIME_ID.
simctl create ... "iOS17.5"will break as macOS runners update runtimes. You already computeRUNTIME_ID; use it with a fallback.Apply this diff:
- # Create a new iPhone SE (3rd generation) simulator with the latest iOS version - xcrun simctl create "iPhone SE (3rd generation)" "iPhone SE (3rd generation)" "iOS17.5" || { + # Create a new iPhone SE (3rd generation) simulator with the detected latest iOS runtime + if [ -n "$RUNTIME_ID" ]; then + xcrun simctl create "iPhone SE (3rd generation)" "iPhone SE (3rd generation)" "$RUNTIME_ID" || { + echo "❌ Failed to create iPhone SE (3rd generation) simulator with $RUNTIME_ID" + echo "Trying to create any iPhone SE simulator..." + xcrun simctl create "iPhone SE" "iPhone SE" "$RUNTIME_ID" || { + echo "❌ Failed to create simulator" + exit 1 + } + } + else + echo "⚠️ No RUNTIME_ID detected; attempting best-effort create with default identifier" + xcrun simctl create "iPhone SE (3rd generation)" "iPhone SE (3rd generation)" || true + fi - echo "❌ Failed to create iPhone SE (3rd generation) simulator" - echo "Trying to create any iPhone SE simulator..." - xcrun simctl create "iPhone SE" "iPhone SE" "iOS17.5" || { - echo "❌ Failed to create simulator" - exit 1 - } - }
🧹 Nitpick comments (2)
.github/workflows/mobile-e2e.yml (2)
189-189: Workspace focus can prune deps needed by native iOS builds — verify coverage or add fallback.
yarn workspaces focus @selfxyz/mobile-app --mode=skip-buildwill prune to the minimal set for that workspace. Ensure all native modules and scripts invoked by Pods/Xcode are reachable (e.g., react-native CLI, codegen, Metro config). If any are dev-only in other workspaces, they may be excluded.Suggested guard to fail fast if key deps are missing:
- - run: yarn workspaces focus @selfxyz/mobile-app --mode=skip-build --silent + - run: | + yarn workspaces focus @selfxyz/mobile-app --mode=skip-build --silent + # Sanity checks for commonly needed deps in native builds + yarn -s why react-native || { echo "❌ react-native not resolved"; exit 1; } + yarn -s why @react-native/codegen || true
205-207: Update Node modules cache key to use the root lockfileThe workflow is hashing
app/yarn.lock, which doesn’t exist in this repo—only the rootyarn.lockis present. As a result, the cache key never changes when dependencies update, causing unnecessary cache misses (and slower CI runs).Please update the cache key in
.github/workflows/mobile-e2e.yml(lines 205–207):diff --git a/.github/workflows/mobile-e2e.yml b/.github/workflows/mobile-e2e.yml
index abcdef1..2345678 100644
--- a/.github/workflows/mobile-e2e.yml
+++ b/.github/workflows/mobile-e2e.yml
@@ -205,7 +205,7 @@ jobs:
path: ~/.npm
- name: Cache Node modules
uses: actions/cache@v3
with:key: ${{ runner.os }}-node-${{ env.NODE_VERSION_SANITIZED }}-${{ hashFiles('app/yarn.lock') }}
with:key: ${{ runner.os }}-node-${{ env.NODE_VERSION_SANITIZED }}-${{ hashFiles('yarn.lock') }} restore-keys: | ${{ runner.os }}-node-${{ env.NODE_VERSION_SANITIZED }}- ${{ runner.os }}-node-This ensures the cache key reflects changes to your actual lockfile.
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
💡 Knowledge Base configuration:
- MCP integration is disabled by default for public repositories
- Jira integration is disabled by default for public repositories
- Linear integration is disabled by default for public repositories
You can enable these sources in your CodeRabbit configuration.
📒 Files selected for processing (1)
.github/workflows/mobile-e2e.yml(1 hunks)
🔇 Additional comments (1)
.github/workflows/mobile-e2e.yml (1)
186-189: Composite action already includes the lockfile hash—no changes neededThe
cache-yarncomposite action useshashFiles(inputs.lock-file)(defaulting toyarn.lock) in its cache key. Your workflow’scache-versioninput is correctly combined with that hash, so you don’t need to appendhashFiles('**/yarn.lock')yourself.Likely an incorrect or invalid review comment.
…nities-to-speed-up-ios-e2e-tests
.github/workflows/mobile-ci.yml
Outdated
| uses: maxim-lobanov/setup-xcode@v1 | ||
| with: | ||
| xcode-version: "16.4" | ||
| xcode-version: "16.2" |
There was a problem hiding this comment.
fixes incorrect upgrade from previous pr
There was a problem hiding this comment.
Actionable comments posted: 1
🔭 Outside diff range comments (1)
.github/workflows/mobile-e2e.yml (1)
220-247: Hardcoded simulator runtime (“iOS17.5”) will fail or flake with Xcode 16.2 imagesYou detect the latest available runtime but don’t persist or use it; then creation hardcodes “iOS17.5”, which likely isn’t present on current images (16.2 typically ships iOS 18.x). Result: simulator creation failures. Persist the detected runtime to GITHUB_ENV and use it during sim creation.
Persist the runtime (end of “Install iOS Runtime” step):
if [ -n "$RUNTIME_ID" ] && [[ "$RUNTIME_ID" == com.apple.CoreSimulator.SimRuntime.iOS-* ]]; then echo "✅ Latest iOS runtime: $RUNTIME_ID" else echo "⚠️ Could not determine an iOS runtime identifier" fi + # Persist for later steps + if [ -n "$RUNTIME_ID" ]; then + echo "IOS_RUNTIME_ID=$RUNTIME_ID" >> "$GITHUB_ENV" + fiUse the persisted runtime when creating the simulator:
- xcrun simctl create "iPhone SE (3rd generation)" "iPhone SE (3rd generation)" "iOS17.5" || { + RUNTIME_TO_USE="${IOS_RUNTIME_ID}" + if [ -z "$RUNTIME_TO_USE" ]; then + RUNTIME_TO_USE=$(xcrun simctl list -j runtimes \ + | jq -r '.runtimes[] | select(.platform=="iOS" and .isAvailable==true) | .identifier' \ + | sort -V | tail -1) + fi + xcrun simctl create "iPhone SE (3rd generation)" "iPhone SE (3rd generation)" "$RUNTIME_TO_USE" || {Also applies to: 286-299
🧹 Nitpick comments (3)
.github/workflows/mobile-deploy.yml (1)
118-123: Pin runner to a macOS image that guarantees Xcode 16.2 availabilityYou’ve downgraded Xcode to 16.2, but the job runs on macos-latest, which can drift to an image where 16.2 isn’t installed. setup-xcode can only select preinstalled versions; if 16.2 isn’t present, the job will fail. Pin to macos-14 (which has 16.2 available) to avoid flaky/broken builds.
Suggested change (outside the selected lines):
jobs: build-ios: - runs-on: macos-latest + runs-on: macos-14.github/workflows/mobile-e2e.yml (2)
186-190: Use immutable installs with Yarn focus to preserve reproducibilityReplacing “yarn install --immutable” with “yarn workspaces focus …” removes the frozen lockfile check. That can silently drift dependencies and destabilize E2E. Also ensure focus actually performs installation for the target workspace in CI.
Suggested change:
- - run: yarn workspaces focus @selfxyz/mobile-app + - run: yarn workspaces focus @selfxyz/mobile-app --all --immutableVerification checklist:
- Confirm your .yarnrc.yml uses nodeLinker: node-modules if you rely on app/node_modules cache.
- Ensure the focus command completes an install; otherwise, keep a preceding “yarn install --immutable” at repo root.
202-205: Align runner image with Xcode 16.2 to avoid version selection failuresThis job uses macos-latest (Line 161) and selects Xcode 16.2. If the latest image doesn’t include 16.2, setup-xcode will fail. Pin to macos-14 for stability or set xcode-version to “latest-stable” if you can absorb the newer toolchain.
Suggested change (outside the selected lines):
e2e-ios: - runs-on: macos-latest + runs-on: macos-14
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
💡 Knowledge Base configuration:
- MCP integration is disabled by default for public repositories
- Jira integration is disabled by default for public repositories
- Linear integration is disabled by default for public repositories
You can enable these sources in your CodeRabbit configuration.
📒 Files selected for processing (4)
.github/workflows/mobile-ci.yml(1 hunks).github/workflows/mobile-deploy.yml(1 hunks).github/workflows/mobile-e2e.yml(2 hunks)app/fastlane/Fastfile(1 hunks)
⏰ Context from checks skipped due to timeout of 300000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (3)
- GitHub Check: e2e-ios
- GitHub Check: analyze-android
- GitHub Check: analyze-ios
🔇 Additional comments (1)
.github/workflows/mobile-ci.yml (1)
124-126: Xcode 16.2 selection looks good and aligns with other workflowsThe version pin is consistent with deploy/e2e. Since this job already runs on macos-14, setup-xcode should reliably select 16.2.
There was a problem hiding this comment.
Actionable comments posted: 1
♻️ Duplicate comments (1)
app/fastlane/Fastfile (1)
206-217: Robust workspace resolution: good improvement and aligns with prior feedbackChoosing between scheme- and project-named .xcworkspace reduces CI flakiness when names diverge. This addresses the earlier review concern about hard-coding the scheme as the workspace filename.
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
💡 Knowledge Base configuration:
- MCP integration is disabled by default for public repositories
- Jira integration is disabled by default for public repositories
- Linear integration is disabled by default for public repositories
You can enable these sources in your CodeRabbit configuration.
📒 Files selected for processing (1)
app/fastlane/Fastfile(1 hunks)
⏰ Context from checks skipped due to timeout of 300000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (2)
- GitHub Check: e2e-ios
- GitHub Check: analyze-android
There was a problem hiding this comment.
Actionable comments posted: 0
🔭 Outside diff range comments (3)
.github/workflows/mobile-ci.yml (3)
11-17: Workflow will not trigger on edits to this fileThe push path filter omits this workflow file. Changes to .github/workflows/mobile-ci.yml won’t trigger the workflow, which can silently break CI.
Apply one of these to ensure triggers fire on workflow changes:
push: paths: - "common/**" - "app/**" - - ".github/workflows/app.yml" + - ".github/workflows/app.yml" + - ".github/workflows/mobile-ci.yml" - ".github/actions/**"Or more generally:
push: paths: - "common/**" - "app/**" - - ".github/workflows/app.yml" + - ".github/workflows/**" - ".github/actions/**"
161-178: Unreachable/inverted Pods cleanup logic in cache verificationInside the “cache restored successfully” branch (where Pods is known to be non-empty), you then check for an empty/nonexistent Pods dir and attempt to clear it. That condition can never be true here, so corrupted/empty caches won’t be cleaned as intended.
Simplify and make the cleanup effective:
- name: Verify CocoaPods Cache run: | echo "Checking CocoaPods cache status..." - if [ -d "app/ios/Pods" ] && [ "$(ls -A app/ios/Pods)" ]; then - echo "✅ CocoaPods cache restored successfully" - ls -la app/ios/Pods/ | head -10 - - # Check if key files exist - if [ ! -f "app/ios/Podfile.lock" ]; then - echo "⚠️ Podfile.lock is missing - pods cache may be stale" - fi - # If Pods directory exists but is incomplete, clear only Pods; keep Podfile.lock - if [ ! -d "app/ios/Pods" ] || [ -z "$(ls -A app/ios/Pods 2>/dev/null)" ]; then - echo "⚠️ Pods directory incomplete - clearing Pods (preserving Podfile.lock)" - rm -rf app/ios/Pods - fi - else - echo "⚠️ CocoaPods cache is empty or missing - will install fresh" - fi + if [ -d "app/ios/Pods" ]; then + if [ -n "$(ls -A app/ios/Pods 2>/dev/null)" ]; then + echo "✅ CocoaPods cache restored successfully" + ls -la app/ios/Pods/ | head -10 + if [ ! -f "app/ios/Podfile.lock" ]; then + echo "⚠️ Podfile.lock is missing - pods cache may be stale" + fi + else + echo "⚠️ Pods directory exists but is empty - clearing Pods" + rm -rf app/ios/Pods + fi + else + echo "⚠️ CocoaPods cache is missing - will install fresh" + fi
200-229: bundle exec pod install likely fails due to Gemfile locality; preserve Podfile.lock for reproducibilityThis step runs in ./app/ios, but bundler was installed using the Gemfile in ./app. Running bundle exec from ios without telling Bundler where the Gemfile is will error (“Could not locate Gemfile”). Also, deleting Podfile.lock on cache misses undermines reproducible builds and cache keys.
- Point Bundler to ../Gemfile while remaining in ios for CocoaPods.
- Avoid removing Podfile.lock; clear only Pods when needed.
# Clean Pods directory if it's corrupted or empty if [ ! -d "Pods" ] || [ -z "$(ls -A Pods 2>/dev/null)" ]; then echo "Cleaning empty/corrupted Pods directory..." - rm -rf Pods Podfile.lock + rm -rf Pods fi # Install pods - bundle exec pod install --silent || { echo "❌ Pod install failed"; exit 1; } + BUNDLE_GEMFILE=../Gemfile bundle exec pod install --silent || { echo "❌ Pod install failed"; exit 1; } echo "✅ Pods installed successfully"If you prefer, you can set the env at the step level instead of inline:
env: BUNDLE_GEMFILE: ../Gemfilein the same step (keeps the command cleaner).
🧹 Nitpick comments (1)
.github/workflows/mobile-ci.yml (1)
135-135: Redundant second checkout increases runtimeA second actions/checkout immediately after Ruby setup is unnecessary and adds I/O overhead.
- - uses: actions/checkout@v4
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
💡 Knowledge Base configuration:
- MCP integration is disabled by default for public repositories
- Jira integration is disabled by default for public repositories
- Linear integration is disabled by default for public repositories
You can enable these sources in your CodeRabbit configuration.
⛔ Files ignored due to path filters (1)
yarn.lockis excluded by!**/yarn.lock,!**/*.lock
📒 Files selected for processing (3)
.github/workflows/mobile-ci.yml(4 hunks).github/workflows/mobile-e2e.yml(1 hunks)common/package.json(1 hunks)
✅ Files skipped from review due to trivial changes (1)
- common/package.json
🚧 Files skipped from review as they are similar to previous changes (1)
- .github/workflows/mobile-e2e.yml
⏰ Context from checks skipped due to timeout of 300000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (2)
- GitHub Check: run_circuit_tests
- GitHub Check: build
🔇 Additional comments (2)
.github/workflows/mobile-ci.yml (2)
126-127: Good call aligning Xcode to 16.2This helps eliminate version drift across jobs and reduces cache thrash on macOS runners.
188-190: Confirm yarn install scope matches PR goal (mobile-only install)The shared ./.github/actions/yarn-install is invoked without a working-directory or focus/frozen settings. Depending on how that action is implemented, it may install all workspaces at repo root, counter to the stated objective to limit installs to the mobile app workspace.
- If the action already focuses the app workspace, ignore this.
- If not, consider one of:
- Run yarn from ./app with a workspace-focused strategy your repo supports (e.g., workspace-tools plugin: yarn workspaces focus @selfxyz/app --production).
- Or keep root install but gate non-mobile workspaces via --since/filters if you’re on Yarn 4 with constraints.
I can review ./.github/actions/yarn-install if you want to ensure it’s properly scoped.
There was a problem hiding this comment.
Actionable comments posted: 1
🔭 Outside diff range comments (8)
.github/workflows/mobile-ci.yml (3)
220-224: Do not delete Podfile.lock in CI; it breaks reproducible and secure buildsRemoving Podfile.lock undermines determinism and can introduce supply-chain drift. Keep the lockfile and only clear Pods when needed.
Apply this diff:
- if [ ! -d "Pods" ] || [ -z "$(ls -A Pods 2>/dev/null)" ]; then - echo "Cleaning empty/corrupted Pods directory..." - rm -rf Pods Podfile.lock - fi + if [ ! -d "Pods" ] || [ -z "$(ls -A Pods 2>/dev/null)" ]; then + echo "Cleaning empty/corrupted Pods directory (preserving Podfile.lock)..." + rm -rf Pods + fi
240-243: Hard-coded Pods-Self path is brittle and likely to fail on different project namesChecking for Pods/Target Support Files/Pods-Self ties the workflow to a specific iOS target name. This will break if the Pod target name differs (e.g., when using a different workspace/scheme).
Apply this diff to verify presence generically:
- if [ ! -f "Pods/Target Support Files/Pods-Self/Pods-Self.debug.xcconfig" ]; then - echo "❌ Key CocoaPods configuration file missing" - exit 1 - fi + # Verify at least one Pods xcconfig exists for any target + if ! ls "Pods/Target Support Files"/*/Pods-*.debug.xcconfig >/dev/null 2>&1; then + echo "❌ No CocoaPods xcconfig found under 'Pods/Target Support Files/*/Pods-*.debug.xcconfig'" + echo "Available target support files:" + find "Pods/Target Support Files" -maxdepth 2 -type f 2>/dev/null | head -50 || true + exit 1 + fi
251-255: Workspace name is hard-coded to OpenPassport; auto-detect or parameterize to avoid false failuresThe e2e workflow now supports dynamic project/workspace names; this step still hard-codes OpenPassport.xcworkspace and will fail for projects named differently (e.g., “Self”).
Apply this diff to auto-detect the workspace:
- if [ ! -f "OpenPassport.xcworkspace/contents.xcworkspacedata" ]; then - echo "❌ OpenPassport.xcworkspace is missing or corrupted" - exit 1 - fi + WORKSPACE_FILE="$(find . -maxdepth 1 -name '*.xcworkspace' -type d -print -quit)/contents.xcworkspacedata" + if [ -z "$WORKSPACE_FILE" ] || [ ! -f "$WORKSPACE_FILE" ]; then + echo "❌ No .xcworkspace found or workspace is corrupted" + echo "Available workspaces:" + find . -maxdepth 1 -name '*.xcworkspace' -type d + exit 1 + fi.github/workflows/mobile-e2e.yml (5)
205-208: Avoid curl | bash for Maestro; pin versioned binary to reduce supply-chain riskPiping a remote script into bash is a security risk and may ignore your MAESTRO_VERSION. Fetch a pinned release and install the binary explicitly.
[security]Apply this diff:
- - name: Install Maestro - if: steps.cache-maestro.outputs.cache-hit != 'true' - run: curl -Ls "https://get.maestro.mobile.dev" | bash + - name: Install Maestro (pinned) + if: steps.cache-maestro.outputs.cache-hit != 'true' + run: | + set -euo pipefail + VERSION="${MAESTRO_VERSION}" + OS="darwin" + ARCH="arm64" + URL="https://github.com/mobile-dev-inc/maestro/releases/download/${VERSION}/maestro-${OS}-${ARCH}.zip" + echo "Downloading Maestro ${VERSION} from ${URL}" + TMP="$(mktemp -d)" + curl -L "$URL" -o "$TMP/maestro.zip" + unzip -q "$TMP/maestro.zip" -d "$TMP" + mkdir -p "$HOME/.maestro/bin" + mv "$TMP/maestro" "$HOME/.maestro/bin/maestro" + chmod +x "$HOME/.maestro/bin/maestro"Note: If you need universal/Intel runners, detect architecture via
uname -mand map accordingly.
294-298: Pod install without Ruby/Bundle setup will be flaky on runnersThere’s no Ruby setup or bundle install. On macOS runners, CocoaPods may not be present, causing “pod: command not found” or gem resolution drift. Use ruby/setup-ruby and bundle exec.
Apply this change outside this block to set up Ruby and gems (add after Xcode setup, before “Install iOS dependencies”):
- name: Set up Ruby uses: ruby/setup-ruby@v1 with: ruby-version: '3.2' bundler-cache: true working-directory: app - name: Install Ruby dependencies run: bundle install --jobs 4 --retry 3 working-directory: appThen minimally adjust this step to leverage bundler:
- (cd app/ios && pod install --silent) || { echo "❌ Pod install failed"; exit 1; } + (cd app/ios && bundle exec pod install --silent) || { echo "❌ Pod install failed"; exit 1; }
329-336: Simulator creation uses invalid runtime identifier and ignores detected runtimeUsing "iOS17.5" is not a valid runtime identifier (expected “com.apple.CoreSimulator.SimRuntime.iOS-17-5”), and the earlier detected runtime is not used. This will cause sim creation to fail on many runners.
Apply this diff:
- echo "Creating a new iPhone SE (3rd generation) simulator..." - # Create a new iPhone SE (3rd generation) simulator with the latest iOS version - xcrun simctl create "iPhone SE (3rd generation)" "iPhone SE (3rd generation)" "iOS17.5" || { + echo "Creating a new iPhone SE (3rd generation) simulator..." + # Create a new simulator using the latest available iOS runtime + RUNTIME_ID="${RUNTIME_ID:-$(xcrun simctl list -j runtimes | jq -r '.runtimes[] | select(.platform=="iOS" and .isAvailable==true) | .identifier' | sort -V | tail -1)}" + if [ -z "$RUNTIME_ID" ]; then + echo "❌ Could not determine an available iOS runtime identifier" + exit 1 + fi + xcrun simctl create "iPhone SE (3rd generation)" "iPhone SE (3rd generation)" "$RUNTIME_ID" || { echo "❌ Failed to create iPhone SE (3rd generation) simulator" echo "Trying to create any iPhone SE simulator..." - xcrun simctl create "iPhone SE" "iPhone SE" "iOS17.5" || { + xcrun simctl create "iPhone SE" "iPhone SE" "$RUNTIME_ID" || { echo "❌ Failed to create simulator" exit 1 } }
252-260: Caching iOS Simulator directories is fragile and bloats cachesCaching CoreSimulator Devices and iOS DeviceSupport often corrupts simulators across runs and wastes space. Prefer not caching these.
[performance]Apply this diff to remove the simulator cache:
- - name: Cache iOS Simulator - uses: actions/cache@v4 - with: - path: | - ~/Library/Developer/CoreSimulator/Devices - ~/Library/Developer/Xcode/iOS DeviceSupport - key: ${{ runner.os }}-simulator-v1 - restore-keys: | - ${{ runner.os }}-simulator-
221-226: Ensure RUBY_VERSION Is Defined for the Bundler Cache KeyThe Bundler cache step at lines 221–226 uses
${{ env.RUBY_VERSION }}but noRUBY_VERSIONis set anywhere in this workflow. As a result, your cache key will always resolve to…-ruby(with an empty suffix), leading to ineffective cache differentiation and more cache misses.Please add a concrete Ruby version to your environment or remove the Ruby component from the key. For example, to pin Ruby 3.2 globally:
• File: .github/workflows/mobile-e2e.yml
Location: top-levelenv:block (around lines 1–30)Suggested diff:
env: # Build environment versions JAVA_VERSION: 17 ANDROID_API_LEVEL: 33 ANDROID_NDK_VERSION: 27.0.11718014 # Cache versions GH_CACHE_VERSION: v1 # Global cache version GH_GEMS_CACHE_VERSION: v1 # Ruby gems cache version + RUBY_VERSION: 3.2 # Specify Ruby version for Bundler cache # Performance optimizations GRADLE_OPTS: -Dorg.gradle.daemon=false …Alternatively, if you prefer not to pin Ruby here, remove
-ruby${{ env.RUBY_VERSION }}from thecache-versionkey under the Bundler cache step.
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
💡 Knowledge Base configuration:
- MCP integration is disabled by default for public repositories
- Jira integration is disabled by default for public repositories
- Linear integration is disabled by default for public repositories
You can enable these sources in your CodeRabbit configuration.
📒 Files selected for processing (2)
.github/workflows/mobile-ci.yml(5 hunks).github/workflows/mobile-e2e.yml(4 hunks)
⏰ Context from checks skipped due to timeout of 300000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (2)
- GitHub Check: e2e-ios
- GitHub Check: run_circuit_tests
transphorm
force-pushed
the
codex/find-opportunities-to-speed-up-ios-e2e-tests
branch
from
bfc871e to
6959ffb
Compare
transphorm
changed the title
.github/workflows/mobile-e2e.yml
Outdated
| fi | ||
| echo "📱 Verifying iOS Runtime availability..." | ||
| echo "Available iOS runtimes:" | ||
| xcrun simctl list runtimes | grep iOS || true |
There was a problem hiding this comment.
old leftover from previous logic that would go through a series of checks for the correct runtime - https://github.com/selfxyz/self/blob/dev/.github/workflows/mobile-e2e.yml#L260
the logic was mainly for testing the pipeline locally using act but we can just use github workflows for now.
it prevents a command run failure if no iOS runtimes are found. previously we should manually select a runtime.
will remove || true but shouldn't be a blocker
transphorm
deleted the
codex/find-opportunities-to-speed-up-ios-e2e-tests
branch
Issue seems be related to the github runner
Summary
Testing
yarn workspaces foreach -p -v --topological-dev --since=HEAD run nice --if-presentyarn lintyarn buildyarn workspace @selfxyz/contracts build(fails: Invalid account in Hardhat config)yarn typesyarn test(fails: exit code 37/1 in circuits/contracts)https://chatgpt.com/codex/tasks/task_b_68a2acabdf8c832db13a667f7eee6d9f
Summary by CodeRabbit
Chores
New Features