Method `rank` exists, but its trait bounds were not satisfied

[mmagician]

I'm trying to compute a rank of a square 256x256 matrix. It works for smaller sizes (up to 127), but for anything bigger I get:

the method `rank` exists for struct `Matrix<f32, nalgebra::Const<256_usize>, nalgebra::Const<256_usize>, ArrayStorage<f32, 256_usize, 256_usize>>`, but its trait bounds were not satisfied
the following trait bounds were not satisfied:
`nalgebra::Const<256_usize>: DimMin<nalgebra::Const<256_usize>>`
`nalgebra::Const<256_usize>: ToTypenum`rustc[E0599](https://doc.rust-lang.org/error-index.html#E0599)
dimension.rs(212, 1): doesn't satisfy `_: DimMin<nalgebra::Const<256_usize>>`
dimension.rs(212, 1): doesn't satisfy `nalgebra::Const<256_usize>: ToTypenum`

It sounds like a pretty basic operation to do, and the matrices aren't THAT big. Is there an easy solution to this? Many thanks!

[Ralith]

Can you use dynamically sized matrices for this? 256x256 is rather large for a value type.

[mmagician]

I'll try

[trueb2]

In embedded (no_std), a dynamically sized matrix isn't really an option, but there are still plenty of matrix operations that would be reasonable to perform on 128, 256, 512, 1024, etc. This rank invocation issue isn't really about rank, but rather the impl's of Const<T>.

Is there a limiting factor preventing add those to the library? I don't know the story behind Const only being implemented up to 127, i.e. why not at least stop on a power of 2. https://github.com/dimforge/nalgebra/pull/820 seems to imply that the Const generics improvement stopped at 127 because no one asked for more. I would be happy to have more on the power of two.

Here is a link to the macro impl for ToTypenum through 127: https://github.com/dimforge/nalgebra/blob/dev/src/base/dimension.rs#L309

[mmagician]

@Ralith Thanks, that indeed works with dynamically sized matrices. It's a shame that static allocation only works for such small matrices though.