Implement 2D boolean operations

[hannobraun]

It should be possible to combine sketches (2D shapes) using union, difference, and intersection operations. This is related to 3D boolean operations (#42, #43, #44), which I'm currently working on. While the low-level details are going to be different (and most likely much simpler for 2D), the high-level algorithm should be pretty much the same, allowing for reuse there.

I'm not sure where this issue fits into the current plans. It could make sense to finish 3D boolean operations first, then work on making the algorithm more generic, until it supports 2D operations too. It could also make sense to go the other way: implement 2D boolean operations first (which should be much easier, given the simpler geometry), then work to expand the algorithm to also support 3D.

[rozgo]

A "3D boolean operations first" approach to generalize for 2D seems like a good direction. Starting with 2D could allow for too many easy shortcuts and assumptions that may not generalize correctly to 3D.

[hannobraun]

It's always a trade-off! On the one hand, what you say is totally correct. On the other hand, getting something working (and be useful) earlier would be worth a lot, as it could provide experience and feedback, which could be worth the extra work generalizing the 2D code to 3D.

I think if I were to start fresh, then I would do 2D first, then later generalize to 3D. I have no intent to do that though, and since I've already put a lot of work into the 3D part, I will continue along that path.

[joamatab]

We already have Boolean operations in gdsfactory, I hope you find it useful

[hannobraun]

I hadn't hear of gdsfactory. Looks interesting!

[hannobraun]

This remains a desired feature, but I've decided to close this issue for now. I've expanded the entry about boolean operations on the feature wishlist to include 2D.

Everything I wrote in #42 is true here too, albeit to a lesser degree. In addition, there are upcoming changes to the geometry representation that would make this issue both more practical to implement, and would essentially require rewriting any code that we'll write for this before then.

For these reasons, I've determined that the feature wishlist is the more appropriate place to track this, for the time being.