Argus Build Diagnostics

Argus Build Diagnostics

Use argus to analyze Xcode builds instead of parsing raw xcodebuild output.

Installation

mise install github:tuist/argus@0.4.0

Running a Build

Run xcodebuild with argus intercepting the build:

BUILD_TRACE_ID=$(uuidgen)
XCBBUILDSERVICE_PATH=$(which argus) BUILD_TRACE_ID=$BUILD_TRACE_ID xcodebuild build -scheme MyScheme

Querying Build Results

After the build completes, use these commands to analyze results:

# Get build summary
argus trace summary --build $BUILD_TRACE_ID

# Get compilation errors
argus trace errors --build $BUILD_TRACE_ID

# Get slowest targets
argus trace slowest-targets --build $BUILD_TRACE_ID --limit 5

# Get build bottlenecks
argus trace bottlenecks --build $BUILD_TRACE_ID

Dependency Analysis

Detect dependency issues in your project:

# Show what each target imports
argus trace imports --build $BUILD_TRACE_ID

# Find implicit dependencies (imports not declared as dependencies)
argus trace implicit-deps --build $BUILD_TRACE_ID

# Find redundant dependencies (declared but never imported)
argus trace redundant-deps --build $BUILD_TRACE_ID

Add --json flag for structured output.

Dependency Graph and Linking Analysis

Query the build dependency graph with product type and linking information to identify optimization opportunities:

# Get the full dependency graph with linking info
argus trace graph --build $BUILD_TRACE_ID

# Query what a specific target depends on (transitive)
argus trace graph --source MyApp

# Query only direct dependencies
argus trace graph --source MyApp --direct

# Query what depends on a specific target
argus trace graph --sink CoreKit

# Find the path between two targets
argus trace graph --source MyApp --sink CoreKit

# Filter by dependency type
argus trace graph --source MyApp --label package
argus trace graph --source MyApp --label framework

Available Labels

• target: Project targets
• package: Swift package dependencies
• framework: Framework dependencies
• xcframework: XCFramework dependencies
• sdk: System SDK frameworks
• bundle: Bundle dependencies

Output Fields

The graph output includes:

• name: Target name
• productType: Xcode product type identifier (e.g., com.apple.product-type.framework)
• artifactKind: The kind of artifact (e.g., framework, staticLibrary, dynamicLibrary)
• machOType: Mach-O type (staticlib, mh_dylib, mh_execute)
• dependencies: Array of linked dependencies with:
  • kind: Link kind (static, dynamic, framework)
  • mode: Link mode
  • isSystem: Whether it's a system dependency

Optimizing Launch Time and Binary Size

Use the graph information combined with linking types to make optimization decisions:

To improve launch time (reduce dylib loading):

1. Query the graph to find dynamically linked dependencies: argus trace graph --source MyApp --json
2. Look for dependencies with kind: "dynamic" or kind: "framework"
3. Consider converting frequently-used internal frameworks from dynamic to static linking
4. Targets with machOType: "mh_dylib" that are only used by one app are candidates for static linking
5. Fewer dynamic libraries means fewer dlopen calls at launch

To reduce binary size:

1. Query the graph to identify statically linked dependencies
2. Look for large targets with kind: "static" that are shared across multiple apps
3. Converting these to dynamic frameworks allows code sharing and reduces total install size
4. Targets with machOType: "staticlib" used by multiple apps benefit from being dynamic

Analysis workflow:

# 1. Get the full graph to understand the dependency structure
argus trace graph --build latest --json > graph.json

# 2. Find all targets that depend on a library you want to optimize
argus trace graph --sink MyLibrary --json

# 3. Check if a target is statically or dynamically linked
argus trace graph --source MyApp --direct --json | jq '.dependencies[] | select(.name == "MyLibrary")'

Decision criteria:

• If a library is used by only one target: prefer static linking (better launch time)
• If a library is used by multiple apps/extensions: prefer dynamic linking (smaller total size)
• For debug builds: prefer dynamic linking (faster incremental builds)
• For release builds with App Thinning: static linking may be preferred

Listing Previous Builds

# List all tracked projects
argus trace projects

# List builds for a project
argus trace builds --project my-app --limit 10