implement chaikin smoothing

[JosiahParry]

No description provided.

[atsyplenkov]

Hey, I think I can help with that. Are you currently accepting PRs?

Overall, It would be great to implement spline and other smoothing techniques that aren’t currently available in the geo crate. I believe they can be written in pure Rust quite efficiently. What do you think about that?

At the moment, I’m only aware of the smoothr R package, which has worked well for years but may be slow with large geometries. Having similar functionality, but with Rust’s performance boost, would be really cool.

[JosiahParry]

I don’t think there is much you can do to change CRAN’s build machines :/

If you want spline smoothing, my recommendation would be to first create a PR downstream in geo with that feature. The team is great and quite helpful. Then once that is done, we can address it in rsgeo. The current version of geo used in rsgeo will face a good number of breaking changes after upgrading.

[atsyplenkov]

Understood. What’s your view on the future of rsgeo? By the way, the Chaikin method is already in the geo crate.

Regarding CRAN, sorry for the confusion, my two issues are not connected. A couple of days ago, I was thinking about using rsgeo as a dependency in my centerline package (which is on CRAN), but I ended up implementing everything I needed with geos, so it’s no longer necessary. With all due respect to CRAN, it is not the only way to disseminate r packages

Now, I’m more interested in the potential for super-fast geo computations in the future.

[JosiahParry]

There was previously an interface to the smoothing method:

rsgeo/src/rust/src/lib.rs

Line 252 in 5c033a7

// /// Chaikin Smoothing

I forget why I removed the method.

The future of rsgeo is uncertain. If (or more-like when) I figure out how to connect geoarrow-rs to R using extendr, I will replace rsgeo with a geoarrowrs and focus contribution to geo then geoarrow-rs and then the geoarrowrs R package. But I have paused this for the time being.