Integrate spade

[RobWalt]

This PR aims at bringing spade triangulations to geo. More specifically I'm looking into implementing constrained delauny triangulations for geo since it seems handy in a lot of situations and opens up possibilities for new algorithms.

Todos

• more tests?

Commit summary

---

chore: add spade as a dependency

although spade is a optional dependency, it looks like a good fit for the default features. Including it has the effect of bringing in 3 new dependencies:

• robust 0.2 (we already use this on version 1.1)
• optional 0.5
• spade 2.2

I'll try to speak to the spade maintainers. Maybe we can bump the robust version in their crate to reduce dependencies further. Also it might make sense to ask them if the optional crate provides any real value.

Update: I created PRs for the changes

• the robust version bump should make it since it comes with performance improvements
• the optional removal might make it depending on the maintainers taste

Update: Both PRs got merged, so this change doesn't add any additional dependencies

---

feat: initial implementation of TriangulateSpade trait

includes all the code for the integration of spade into geo and also features some tests

---

• I agree to follow the project's code of conduct.
• I added an entry to CHANGES.md if knowledge of this change could be valuable to users.

---

[urschrei]

This is great, @RobWalt, thank you! We'll wait to see how the proposed changes to spade go.

[RobWalt]

@urschrei Do you think it would be feasible to review & merge this as is and keep a tracking issue open for updating the dependencies which track the two spade PRs? I assume these two PRs are not going to be included anytime soon and this issue here is required to merge the spade boolops at some point

[urschrei]

I'm in favour of merging this, both because it's a pre-requisite for the new boolean ops functionality, and because geo should support Delaunay triangulation as a feature more generally. For some reason it's not currently showing up as a top-level item under the algorithms module – again, you can check for yourself by running cargo doc --open.

[urschrei]

Another thing that's needed here is a doctest example of how to use the algorithm to produce both constrained and unconstrained triangulations. This example should link to a good overview of what is meant by constrained / unconstrained, although your diagrams are also great.

[RobWalt]

• applied the review comments suggestions
• rebased on main
• sealed the helper methods on a separate private trait since they probably shouldn't be public in the first place

[michaelkirk]

Nice work! None of my comments should be considered blocking.

[RobWalt]

Thanks for the thorough review @michaelkirk! 🏅 I like your taste 👍🏼

[RobWalt]

Ok, the proposed changes to spade were accepted and are merged as of now. There's nothing major in the way now. The only things left are:

• explore review comment about epsilon
• add more tests (?)

[rmanoka]

LGTM; thanks @RobWalt .

[RobWalt]

😎