0.16.0 Release!

[timschmidt]

Our biggest release yet!

csgrs now has fully functional 2D and 3D subsystems!

The all new 2D subsystem is built around https://crates.io/crates/geo which supports several geometry types including MultiPolygons with holes, and has allowed for drastically improved extrude geometry. The shape library has also grown tremendously! Translate has a new shorthand using Reals instead of a Vector3. xor has been implemented for 3D shapes. TrueType glyphs are now available as paths and polygons. Many bugs have been fixed. I'm sure there are still a lot, but there's lots to love too!

As always the latest release is available from crates.io: https://crates.io/crates/csgrs

I'd like to thank https://github.com/ArchiyouApp for becoming a Github sponsor for csgrs. My next sponsorship goal is to get to 10 sponsors for the project. Please lend a hand if you can.

[ftvkyo]

Congratulations with this release! I am a frequent user of OpenSCAD and Rust, so programmatic CSG in Rust is a dream :D
I considered building something like this too. This is exciting.

I have recently been using geo to make an SVG to G-Code converter (still WIP) for milling a PCB on a CNC-Router, and it looks like csgrs would be a nice upgrade even if for just the integration of everything.

[timschmidt]

Thank you! And thank you for sponsoring the project! It means a lot to me.

Your SVG to gcode project is very interesting. I would like csgrs to provide all the geometrical tools necessary to produce high quality toolpaths. I would also like csgrs to be able to ingest and export SVGs. Would you be willing to help write SVG import/export for csgrs? It'd be just the same as for geo. You can find me in the GeoRust, Rust Gamedev, or Bevy Discord channels for quicker turnaround on replies.

[timschmidt]

I considered building something like this too.

Is that https://github.com/autoyolk/feather ? I was eyeballing that when you star'd csgrs.

[ftvkyo]

Yep, you got the right link!

I was looking for options to procedurally generate complex geometry for 3D printing, as OpenSCAD was getting in the way.
I'll throw in the thoughts I had in mind, maybe something would be of interest.

1. I wanted to control twist & scale in linear_extrude() using a function, like ImplicitCAD does this. Trying to approximate something like this with OpenSCAD took the render time to ~30 minutes 😅
2. I wanted better facilities for using the "escape hatches" line polyhedron(). It's annoying when e.g. you have to reimplement affine transformations: https://www.ftvkyo.me/project/openscad-snippets/#wavy-donut
3. While some ideas that the builtin language used were very interesting, I feel like with the right API, a more general purpose language would provide a much smoother experience. I am aware of PythonSCAD though, this all is just an excuse to have fun :D

Ultimately, feather is on the back-burner because I like to reinvent wheels, and I realised that this wheel is too big for me at the moment 😅

Yes, I would be interested in working on SVG IO for csgrs. Just joined the GeoRust discord :)

Some thoughts about this:

• SVG has CSS. CSS takes precedence over tag attributes. stroke-width is important for giving thickness to paths. I did stroke expansion by myself, and later switched to geo-offset (although I see a few things in it I want to improve). Regardless, the point is -- it might be necessary to also parse and resolve CSS. It's outside of the scope of the svg crate, but there might be something more advanced out there.
• Output would probably be a bunch of filled SVG paths made out of straight segments, that's probably OK. SVG also supports holes in filled paths, so that's not an issue. They have an example with reverse winding order here: https://developer.mozilla.org/en-US/docs/Web/SVG/Attribute/fill-rule

If you get curious, this is the WIP tool: https://github.com/ftvkyo/svg2gcode

Here are some contours I produced with it:

[timschmidt]

Responding to your ideas:

1: I'm generally in favor of more F-rep in csgrs. Doesn't seem complex to implement, we already have fast-surface-nets as a dep, anything you'd like to add on that front I'm open to.

2: I have actually tried to draw the very same shape in OpenSCAD. I failed, and I'm not surprised you had to resort to such extreme measures. I feel like having all the CAD kernel's types and methods public is a good first step, but I'd also like to implement more iterators within csgrs, accessor functions, that sort of thing, to make the internals a little easier to work with.

3: Agreed. I like that I can bolt csgrs into other applications relatively easily. And still use it to express ideas quickly for printing.

4: Understood.

5: Awesome. Many hands make light work.

6: I'm interested to hear more about what you see in geo-offset which could be improved. I am OK starting with what the SVG crate is capable of today. With basic support implemented, we can work on ways to improve it.

7: Sounds totally workable.

8: I looked through the code briefly. Looks useful.

9: That is some fancy linework! beautiful!