TinyGLTF v3 C (tiny_gltf_v3.h + tiny_gltf_v3.c + tinygltf_json_c.h) is the mainline. Legacy v1/v2/v3-C++ code lives under attic/ and is not built or tested.
make -C tests all && make -C tests runcmake -B build -DTINYGLTF3_BUILD_TESTS=ON && cmake --build build && ctest --test-dir build --output-on-failuremeson setup build && meson compile -C build && meson test -C build~/work/emsdk): make -C web && make -C web sample — then open http://localhost:8000 via make -C web servemake -C tests/v3/fuzzer (clang + libFuzzer)Run before every push. Fix or unstage anything found; the audit must pass clean.
Working tree sanity
git status --short — no staged build artifacts, no surprise files.git add -A / git add .: stage only intended paths. The repo root frequently contains untracked scratch files (Fox.glb, benchmark/synthetic_*, log*, .cache/, tests/v3/fuzzer/artifacts|corpus, stray *.gltf/*.bin files). Verify git status before staging.Secret scan
gitleaks detect --source . --no-banner -r /tmp/gitleaks.json~/local/bin/trufflehog git file://$PWD --no-update --only-verifiedB3G_LEFT_ARROW generic-api-key hits in attic/examples/common/OpenGLWindow/ are false positives (keyboard key constants).No build artifacts or unwanted large binaries
git ls-files -z | xargs -0 du -b | sort -rn | head -30 — review the top.attic/tools/windows/premake5.exe, models/Cube/*.png).*.wasm/*.exr/*.log/*.o/*.obj, generated output.*/cloth*/sim*.bin/out_*.exr files, and large .gltf/.glb files. Small test/regression fixture .gltf files (e.g. under tests/, models/regression/, tests/v3/security/) are OK.web/tinygltf_v3.js, web/tinygltf_v3.wasm, web/Cube.glb are generated by make -C web and must stay gitignored (never commit them).History is lean
git rebase -i origin/<branch>, git rm + --amend) so the blob never reaches the remote, then git gc --prune=now locally.git push origin <branch> — never --all or --force unless explicitly asked).Final smoke check
make -C tests run passes (all 4 testers).cmake/meson builds and make -C web still succeed.