Inconsistent API and unnecessary mut qualifier.

[wdanilo]

Hi! See the signatures of all uniform_matrixXXX_with_f32_array functions. Almost all of them consume data: &[f32]. However, the ones which are defined for non-square matrices, consume data: &mut [f32] instead. To keep the API consistent, all of them should consume data: &[f32].

[alexcrichton]

Thanks for the report!

By default we assume mutable access is needed (since we don't know otherwise) and we have a whitelist of methods that don't mutate. If you'd like a PR to extend that list would be most welcome!

[wdanilo]

@alexcrichton sure thing, here it is: #1957 :)

As this is a very minor change, I'd be thankful if it could be merged fast :)

[wdanilo]

Oh, one more question!
There is another issue - the generated functions are not specialized for all possible variants. There is function tex_image_2d_with_i32_and_i32_and_i32_and_format_and_type_and_opt_u8_array but there are not functions for other array types (sure we can do it via the JsObject casting). According to MDN docs we can provide much more types here.

Would you be so nice and tell me where can I fix it in the code? :)

[alexcrichton]

Ah unfortunately that's a lot more complicated since it's all based on the WebIDL that we're reading. Want to open a dedicated issue for that?

[wdanilo]

@alexandrestein of course!