This test suite is designed to ensure consistency across all Spine runtime implementations by comparing their outputs against the reference implementation (spine-libgdx).

Purpose

Unlike traditional unit tests, this test suite:

Loads skeleton data and animations in each runtime
Outputs all internal state in a consistent, diffable text format
Compares outputs between runtimes to detect discrepancies
Helps maintain consistency when porting changes from the reference implementation

HeadlessTest Locations

Each runtime has a HeadlessTest program that outputs skeleton data in a standardized format:

Java (Reference): spine-libgdx/spine-libgdx-tests/src/com/esotericsoftware/spine/HeadlessTest.java
C++: spine-cpp/tests/HeadlessTest.cpp
C: spine-c/tests/headless-test.c
TypeScript: spine-ts/spine-core/tests/HeadlessTest.ts

Running Individual HeadlessTests

Java (spine-libgdx)

cd spine-libgdx
./gradlew :spine-libgdx-tests:runHeadlessTest -Pargs="<skeleton-path> <atlas-path> [animation-name]"

# Example with spineboy:
./gradlew :spine-libgdx-tests:runHeadlessTest -Pargs="../examples/spineboy/export/spineboy-pro.json ../examples/spineboy/export/spineboy.atlas walk"

C++ (spine-cpp)

cd spine-cpp
./build.sh  # Build if needed
./build/headless-test <skeleton-path> <atlas-path> [animation-name]

# Example with spineboy:
./build/headless-test ../examples/spineboy/export/spineboy-pro.json ../examples/spineboy/export/spineboy.atlas walk

C (spine-c)

cd spine-c
./build.sh  # Build if needed
./build/headless-test <skeleton-path> <atlas-path> [animation-name]

# Example with spineboy:
./build/headless-test ../examples/spineboy/export/spineboy-pro.json ../examples/spineboy/export/spineboy.atlas walk

TypeScript (spine-ts)

cd spine-ts/spine-core
npx tsx tests/HeadlessTest.ts <skeleton-path> <atlas-path> [animation-name]

# Example with spineboy:
npx tsx tests/HeadlessTest.ts ../../examples/spineboy/export/spineboy-pro.json ../../examples/spineboy/export/spineboy.atlas walk

Running the Comparison Test

The main test runner compares all runtime outputs automatically:

./tests/headless-test-runner.ts <skeleton-path> <atlas-path> [animation-name]

This script will:

Check if each runtime's HeadlessTest needs rebuilding
Build any out-of-date HeadlessTests
Run each HeadlessTest with the same inputs
Compare outputs and report any differences
Save individual outputs to tests/output/ for manual inspection

Example Usage

# Test with spineboy walk animation
./tests/headless-test-runner.ts \
    examples/spineboy/export/spineboy-pro.json \
    examples/spineboy/export/spineboy-pma.atlas \
    walk

# Test without animation (setup pose only)
./tests/headless-test-runner.ts \
    examples/spineboy/export/spineboy-pro.json \
    examples/spineboy/export/spineboy-pma.atlas

Output Format

Each HeadlessTest outputs:

SKELETON DATA: Static setup pose data (bones, slots, skins, animations metadata)
SKELETON STATE: Runtime state after applying animations
ANIMATION STATE: Current animation state with tracks and mixing information

The output uses consistent JSON formatting:

Hierarchical structure with 2-space indentation
Float values formatted to 6 decimal places
Strings quoted, nulls explicitly shown
Locale-independent number formatting (always uses . for decimals)
Circular references marked as "<circular>" to prevent infinite recursion
Each object includes a "type" field for easy identification

Development Tools

API Analyzer (Java)

Analyzes the spine-libgdx API to discover all types and their properties:

cd tests
npx tsx analyze-java-api.ts
# Output: output/analysis-result.json

Serializer Generator (Java)

Generates SkeletonSerializer.java from the analysis:

cd tests
npx tsx generate-java-serializer.ts
# Output: ../spine-libgdx/spine-libgdx/src/com/esotericsoftware/spine/utils/SkeletonSerializer.java

Claude Prompt Generator

Generates a prompt for Claude to help port the serializer to other runtimes:

cd tests
npx tsx generate-claude-prompt.ts
# Output: output/port-serializer-prompt.txt

Troubleshooting

If outputs differ between runtimes:

Check tests/output/ for the full outputs from each runtime
Use a diff tool to compare the files
Common issues:
- Number formatting differences (should be fixed by locale settings)
- Missing or extra fields in data structures
- Different default values
- Rounding differences