blog post: recent & future pattern matching improvements

[Centril]

r? @nikomatsakis

[Centril]

NOTE: the file name needs to be adjusted depending on when this is merged.

[Mark-Simulacrum]

Overall looks good. Interested to hear what you think on some of the comments, but generally seems like we can likely move ahead.

[Mark-Simulacrum]

Why do we need the same binding modes? I would've expected e.g. it be fine to have Ok(ref x) | Err(x) with Result<u32, &u32> producing x: &u32.

Is this just a match/borrow check limitation?

[Centril]

It's possible that there's no good reason for this limitation (which is encoded in librustc_resolve) to exist. I'm not exactly sure why it does, so my answer is the non-answer "this is how it has always been" ;)

error[E0409]: variable `x` is bound in inconsistent ways within the same match arm
 --> src/main.rs:2:25
  |
2 |     let Ok(ref x) | Err(x) = Ok(0);
  |                -        ^ bound in different ways
  |                |
  |                first binding

Maybe @petrochenkov or @matthewjasper know why.

(Note to self: the diagnostic should not mention "match arm".)

[matthewjasper]

The compiler doesn't currently track binding modes separately for each binding. Since Ok(x) | Err(mut x) has to be an error, and Ok(ref x) | Err(&ref x) is allowed, there isn't much reason to change this.