absInt over vector?

[uneven yarrow]

Is there a way to run a scalar function on every element of a vector and put the result into a new vector? I'd expect like a @map but I don't see one.

(besides doing the for loop)

[loud relic]

zig and functional programming styles are kind of orthogonal

zig is pretty solidly imperative

[uneven yarrow]

well, there's a @reduce 🙂

[loud relic]

that applies to vectors, which map quite directly to hardware

and unlike arrays, vectors only hold stuff like integers, pointers, floats - not composite types

[uneven yarrow]

I'll be more specific - it'd be nice to have a vectorized std.math.absInt, but there's no such thing. So I'm wondering what the nicest way to deal with this is.

there's a vectorized std.math.sign, for example.

[loud relic]

well, there is @fabs, but that only applies to vectors of floats

[uneven yarrow]

indeed

[loud relic]

at any rate, there's maybe a way to do it with @select

[errant vale]

You would select

[solar bear]

sign() does some select() stuff, so you could go off that

[errant vale]

I like to model my vector stuff into data transformations and reducing the movements

[uneven yarrow]

ty

[loud relic]

you'd probably have something like

fn absVector(x: anytype) {
    const Child = @typeInfo(@TypeOf(x)).Vector.child;
    const pred = x > @splat(x.len, @as(Child, 0));
    const y = -x;
    return @select(Child, pred, x, y);
}

std.math.absInt could probably be doing this

[uneven yarrow]

yeah, that's approximately what I'm doing now. I guess maybe absInt isn't doing this because of fear of overflow.

I suppose you'd just check all the elements first.

[loud relic]

well, Idk about fear

[uneven yarrow]

sure, s/fear/possibility/ 🙂

[loud relic]

ye

[uneven yarrow]

absInt over vector?