James Lucas

Yeah the example there looks good. I was just referring to this statement: > the users of a model don't get to choose the types it acts on - these...

Have responded in #114 just to minimize noise here for anyone else reading.

I have some proof of concepts for the ideas discussed here: - [K-means that accepts `MatrixSlice`](https://github.com/AtheMathmo/rusty-machine/blob/151eed2b4ff791cc15e882a50a6092fcc7ce4a42/src/learning/k_means.rs) - [New model trait separation](https://github.com/AtheMathmo/rusty-machine/commit/7905e1872a9ecae7129d1e537bcf7195af4aec03) These have a few blocks before they make it...

Thanks for commenting! I agree with you totally that using an enum will lead to pretty clunky code. The trait version is definitely a lot cleaner. I also agree that...

I think it might be acceptable to restrict the actions to only work on `Matrix`. `Vector` only really exists to optimize some operations on a column of data (it could...

I see what you mean - we do indeed want to work with other types. Though I think this is a valuable enough update that we could first make the...

I wanted to try playing with this idea a little and ran into [an issue](https://play.rust-lang.org/?gist=0f3b46e7fd2e24ef0aceb7c6592e67f4&version=stable&backtrace=0). We can't create a trait object without specifying the associated types. This is [also indicative](https://doc.rust-lang.org/book/trait-objects.html#dynamic-dispatch)...

I think it is best to stick with `Matrix` and tackle convolutions later? Also once #117 lands we can make it operate on `T: BaseMatrix` - that will give us...

Please do! I'd be more than happy to give input when you're ready. It may be worth you being aware of the ongoing discussion in #117, #120 and #124 ....

I've added the macro and tested that the windows build goes through okay. In doing so, I've potentially shown my hand as not-really-a-c++-developer. Feel free to make changes directly or...