v2.1.0

This is a significant update, primarily centered on replacing MeshRelation with MeshGL, which accomplishes the goal of maintaining material properties across operations much better and with a simpler, more performant API. I'm not considering this a breaking change, as MeshRelation turned out to be broken enough that I doubt it was getting much use anyway. For the WASM build on npm, this is definitely a breaking change as we've updated its API more significantly and switched to an ES6 module. This brings us up from v1.0 to v2.1 to keep a consistent version from now on.

New Features

• Added MeshGL. #269 (@stewartoallen, @elalish)
• Handle arbitrary properties in MeshGL. #290 (@elalish)
• Added transform relations to MeshGL. #315 (@elalish)
• Added normals update to GetMeshGL. #317 (@elalish)
• Simplified Polygons type. #343 (@elalish)

WASM bindings

• Added missing declarations. #267, #320 (@jirihon, @rafern)
• Update declarations for ES6. #342, #346 (@rafern)
• Refactored bindings. #268 (@rafern)
• Added JS Mesh class. #272 (@elalish)
• Updated to ES6 Module. #322 (@elalish)
• Accept Mat4 in transform. #329 (@pca006132)

Build/CI Updates

• Added Node.js tests. #262 (@elalish)
• Added C bindings. #292, #295, #303, #306, #310, #311, #313 (@geoffder)
• Updated CI to CUDA 12. #319 (@elalish)
• Added maze benchmark to Python run. #333 (@pca006132)

Bug Fixes

• Fixed memory leak. #280 (@pca006132)
• Fixed mirroring scale. #298 (@elalish)
• Made robust to overlaps by default. #299 (@elalish)
• Fixed Python binding bug. #307 (@pca006132)
• Changed enum case to avoid macros. #325 (@elalish)
• Hash table OMP fix. #331 (@pca006132)

Examples

• Created multi-material example. #309 (@elalish)

ManifoldCAD.org

• Fixed first run problem. #281 (@elalish)
• Fixed CSS. #296 (@pca006132)
• Switched from three.js GLTFExporter to gltf-transform. #322 (@elalish)
• Added show() and only() functions for visual debug. #330, #334 (@elalish)
• Added gl-matrix to editor for vector functions. #332 (@elalish)
• Made into a fully installable PWA. #336 (@elalish)
• Fixed module interface. #348 (@elalish)

v2.0.0

Thanks to all our collaborators, we've made enormous progress on this library, including some significant updates to the API that make this worthy of a v2.0 release. Following is a non-exhaustive list of changes.

New Feature

• Added Signed-Distance Function (SDF) support. #187 (@elalish)

API Changes

• Updated API to functional style. #131, #171 (@pca006132)
• Exceptions now optional, #170 (@elalish)
• Reduced dependencies #92, #189, #190, #194 (@elalish)
• Updated MeshRelation API, #184 (@elalish)

Build/CI Updates

• Windows compatibility. #88 (@yetigit)
• Godot compatibility. #93 (@fire)
• Cross-platform CI. #103, #111, #229 (@pca006132)
• Python binding. #113, #214 (@pca006132)
• Reorganized directories, #157 (@elalish)
• Added code coverage, formatting. #132, #140, #142, #195 (@pca006132, @elalish)
• WASM build. #155, #201, #221, #224 (@makc, @pca006132, @elalish)
• Now available on npm as manifold-3d.

Bug Fixes

• Make thread-safe. #125 (@pca006132)
• Fixed degenerate removal. #137, #179 (@elalish)
• Fixed triangulator. #154, #165, #180, #182 (@elalish)

Performance Improvements

• CGAL comparison. #84 (@ochafik)
• Optimizations. #106, #150, #235 (@pca006132)
• Managed memory. #121 (@pca006132)
• Dynamic parallel backend. #126 (@pca006132)

ManifoldCAD.org

• A new CAD app is born by using <model-viewer>, #198, and monaco, #208. (@elalish, @pca006132)
• UX design. #210, #212, #222, #230 (@elalish)
• Memory management. #213 (@pca006132)
• Operations off main thread and cancelable. #239 (@elalish)
• Examples. #231, #244, #249, #251 (@elalish)

v1.0.0

I'm marking this first release as v1.0 because I've been working on it long enough and it passes a rigorous enough batch of tests that I'm convinced it really is a next step in robust computational geometry. Obviously I can't promise it's bug-free, but I believe at this point the fundamental architecture is sound and can be built upon. Likewise, after many false starts I believe I have an API that is durable and can be expanded without major breaking changes.

In short, I feel comfortable recommending you try using it and building on it, and that I won't be wasting your time.