Shouldn't complement live in Colors.jl?

[yakir12]

Small function, small issue, but seems like complement which acts on a color should live in Colors.jl.

[SimonDanisch]

I think there are a few functions, that should actually live in Colors.jl instead of Images.jl.
E.g. clamp01, reexport of N0f8 - probably more! How about we make a list, confirm that they should move and then make a PR?

[bjarthur]

mapwindow also strikes me as misplaced, as i often use it for time series. seems equally suited to me to be in a signal processing package.

[yakir12]

I agree with all of that.

Since there are currently 448 packages that depend on Colors, there might be some resistance to adding any functions to it (thus making it ever so slightly larger).

I tried to find the methods that use Color in Images by looking at the output from methodswith(Color) after using Images, but didn't get any useful result. Can you compile a list of Images functions that act upon types of Colors?

As to mapwindow, yea I find myself using Images just for some of the filter functions there, even when I only need it for a Vector...

[kimikage]

BTW, if you move complement to another package, I think ColorVectorSpace.jl is more appropriate package than Colors.jl.

[wizofe]

I attempted a PR for both steps! Being a GSoC student; be kind ^_^

• Step 1: PR125
• Step 2: Bug/move std to cvs #879

Thanks!