Add 'do' syntax for <RACStream> about reactivecocoa HOT 5 CLOSED

I don't think a traditional do syntax would be very useful in objc.

If all you're doing is chaining bind:s, the syntax would be just a bit lighter, but the general readability would suffer because of it's inconsistency with the rest of the code.
If you're composing bind:s, objc already allows you to assign the result of bind:s to variables, so there's no need for a <- operator.

What really hurts readability as it is now, is splitting and merging streams through publish and combineLatest:, or working with a lot of tuples to avoid doing that. Especially since objc's type signatures don't allow you to specify what is in the tuples.

from reactivecocoa.

If you're composing bind:s, objc already allows you to assign the result of bind:s to variables, so there's no need for a <- operator.

This is a little bit of a red herring, because:

do
    x <- M
    …

maps to something like the following Objective-C:

[M bind:^(id x) {
    …
}];

That said, I agree with your overall point. I don't think do syntax would be that useful for how RAC is generally going to be used.

from reactivecocoa.

I think the value of do syntax is that deeply nested binds are gross.

So, compare:

do 
    x <- f1
    y <- f2 x
    z <- f3 x y
    return x + y + z

vs.

[[self f1] bind:^(id x) {
    return [[self f2:x] bind:^(id y) {
        return [[self f3:x and:y] bind:^(id z) {
            return [self.class return:[[x plus:y] plus:z]];
        }];
    }];
}];

Nesting aspolsion.

The question in my mind is whether a macro could be better enough to make it worthwhile.

from reactivecocoa.

A do macro would end up fairly similar to Clojure's:

do(
    x, [self f1],
    y, [self f2],
    z, [self f3:x and:y],
    [self.class return:[[x plus:y] plus:z]]
)

I just don't think people will be binding that often/intensely.

from reactivecocoa.

Yeah, I agree. Not much better and probably rare.

from reactivecocoa.