RFC: Allow pure functions to mutate their `&mut` arguments

[catamorphism]

Suppose I want to write something like:

pure fn mk_str() -> ~str {
  let mut s = ~"";
  str::push_char(&mut s, 'c');
  s
}

This doesn't work because push_char isn't pure. Morally, it seems like it should work, though, because although push_char does modify memory it doesn't own, mk_str owns the memory and should be allowed to call a function that modifies it.

I'm sure this question has come up before, but I don't know if there's an issue on it. If we do keep purity in the language, it seems like it's worth thinking about.

[nikomatsakis]

I have a plan to address precisely this problem. I was hoping to write a blog post on it today.

[catamorphism]

Related: #3490

[nikomatsakis]

Here is a description of my proposal:

http://smallcultfollowing.com/babysteps/blog/2012/10/12/extending-the-definition-of-purity-in-rust/

[nikomatsakis]

I renamed the issue to describe my proposal and to include the term "RFC".

[graydon]

pure functions are scheduled for removal, so closing WONTFIX