Perhaps we could make it a suggestion instead? The advantages of the new function is that it is faster (I think), is more compatible with NA, and is stricter about casting. I think it would be an improvement and good preparation for the future. I suppose we could just just add the extra keywords to putmask as a workaround.

copyto(dst, src, casting='same_kind', where=None, preservena=False)

I am suggesting it in the docstring right? Do you want different wording, or something else? I don't understand why adding new keywords to putmask would help.

Might just add a mention of NA.. The preservena keyword of copyto keeps NA in the dst from being overwritten. But I agree that deprecating putmask is draconian and undeprecating it is the right thing to do.

Done.

I pushed this in 3d9bedb and opened enhancement ticket #2008 for implementing putmask with copyto.