cospectrum
cospectrum
1. "ownership variant" variant not always possible from existing implementations, for example, `match_histogram_mut(&mut img, &img);` doesn't compile and doesn't make any sense. 2. It's not about ownership variant, I don't...
Owned variant is possible only from inplace.
This is just an example. We already have a convention to implement `_mut` functions (in place or with an output buffer). So I don't see how `owned` brings anything to...
`let b = match_histogram(b, &a)` looks stupid. We don't want that
Lets just find functions that have only this kind of the signature `f (&I) -> I`, and will see if it makes sense to add `_mut` implementation. They should allocate...
Do you want to replace `func(&I) -> I`, `func_mut(&mut I)` pairs with single `func(I) -> I`?
> Do you want to replace `func(&I) -> I`, `func_mut(&I, &mut I)` pairs with single `func(I) -> I`? It's a question of style. Ask the owner.
Should we remove `match_histogram` then? There is `match_histogram_mut`
We should follow `image` codestyle in my opinion
> We should follow `image` codestyle in my opinion They have `imageops::blur(&img, sigma)`, but no `imageops::blur_mut(&mut out, &img, sigma)`. So the same questions araise for them. `imageproc` is some kind...