Tuple's are not allowed as lvalues

[adrientetar]

Currently, this is unimplemented:

fn main() {
    let (x,y) = (1,2);
    (x,y) = (3,4);
}

The line (x,y) = (3,4); will trigger an error.

This is allowed in Python for example and can come in handy:

def fib_increment(n):
  a,b = 0,1
  for _ in range(n):
    a,b = b,a+b
  return a

Here, b allocated value depends on a and vice-versa. Not sure how this translates into lower-level code but at least here it spares an alloc and a line of code.

Also, being able to allocate a tuple from a function would come in handy, cf. #7507#issuecomment-23618416.

See #7507, #7508, #10123.

[emberian]

cc @nikomatsakis

[emberian]

I was under the impression we weren't allowing this for grammar reasons.

[eddyb]

We can translate Expr to Pat, or even handle a tuple of values specially in trans, do we need anything a pattern can express that an expression can't?
ref and mut affect the resulting binding (so they don't make sense in destructuring assignment).
&x in Expr creates an rvalue, so there would be no ambiguity in the language (though it might be confusing for users) as to it being a dereference in a destructuring assignment.

[Aatch]

If anybody is still interested in this, an official RFC should be filed. https://github.com/rust-lang/rfcs/blob/master/active/0001-rfc-process.md

[adrientetar]

@Aatch Are you closing all issues tagged as B-RFC?

[Aatch]

@adrientetar not all, just a few that clearly should go through the new RFC process, I only did 7.

[adrientetar]

@Aatch Well this is a feature request and not a proposed implementation so it is not going to go any further.

[huonw]

RFCs are not (just) for proposed implementations but rather extensions to the language, which is exactly what this is.