[swift-evolution] [Pitch] Retiring `where` from for-in loops

[Haravikk]

> On 13 Jun 2016, at 13:28, Xiaodi Wu via swift-evolution <swift-evolution at swift.org> wrote:
> 
> I think this discussion has made it pretty plain that what is claimed to be 'so useful' is barely ever used. Moreover, it provides no independent uses. The point of these pitches is to sound out arguments, not, as far as I was aware, to take a vote.

I think that the discussion has made it clear that far more people want this feature to remain, and feel it is a valid part of the language, whether or not it’s a mainly aesthetic one, and that it should stay. Whether or not the discussion is intended as a vote is irrelevant, if more people want a feature to remain than want it removed then that suggests the justification for removing it had better be a really good one.

So far we have redundancy, which isn’t really sufficient reason to remove something, as we have plenty of redundant syntax for convenience. Take conditional binding for example, which is essentially just a shorthand for pattern matching. Functionally they’re identical, and there’s not much difference in complexity (you don’t have to learn the pattern matching aspect in order to use it for unwrapping optionals), so really it’s just a neat shorthand similar to where, and could be removed easily, yet people would fight against it just as furiously.

Otherwise the argument is against confusion/mistakes, but this alone isn’t a reason to remove a feature vs addressing why confusion is caused. Besides, the main argument for it being confusing comes from its use in both while and for loops, but if it’s being removed from while loops then this ceases to be an issue. As a result it would be premature to remove the feature as we don’t know yet if it’s still confusing when only permitted on for loops.

After that the other issue is whether filtering deserves a special case, but this is muddied by a lack of use largely resulting from lack of awareness. For us to get useful evidence either way would require this to put more widely known and to see if it continues to be used infrequently; I know about the keyword and use it in maybe 20% of my for loops (plus I actually rarely use while loops, and mainly in testing).
Besides which there is a proposal for a variant of where (while/until) which gives access to another fairly common variant of if/guard that developers may wish to replace, which would address a lack of versatility.

The benefit of the where clause is eliminating a line by putting the same condition on the opening loop, you may not consider this worthwhile but clearly plenty of people do, therefore I’d consider this justification enough for the feature to remain as we find it useful and prefer how our code looks with this shorthand. You remain free as ever to not use it.