".-" is not a unary operator

[eschnett]

I notice that .- is not a unary operator:

julia> .-[1]
ERROR: syntax: ".-" is not a unary operator

For consistency, I'd expect .+ and .- to be available as unary operators as well.

[stevengj]

It's hard to imagine a circumstance in which .-x != -x, so on that grounds I'm inclined not to add the former synonym.

[eschnett]

It's the same case as as for x - y and x .- y. If there's an element-wise subtraction operator, I expect an element-wise negation operator as well.

[stevengj]

@eschnett, that's not true. x - y is not defined if x and y do not have the same shape, where x .- y is (often) defined. As a binary operator, therefore, .- is a strict generalization.

[oxinabox]

Are there any element wise unary operators?
I can't recall if there is an element-wise "not" operator

[KristofferC]

Nothing really actionable here.

[stevengj]

Now that dot operators are fusing, we should probably reconsider this. e.g. imagine you want to compute log.(.-a) rather than log.(-a), in order to avoid the temporary array for a::Array.

[yuyichao]

I think we might still want to parse unary .- as broadcast.