ICE: Computing fictitious types

[alexcrichton]

This program

fn error(m: uint, positions: ~[()]) { 
  match m {                           
    i => {                            
      let (a, b) = positions[i];      
    }                                 
  }                                   
}                                     

fn main() {}                          

generates:

$ rustc ./foo.rs
error: internal compiler error: Asked to compute contents of fictitious type

Additionally, a similar program

fn error(m: uint, positions: ~[()]) { 
  let (a, b) = positions[m];      
}                                     

fn main() {}                          

yields

$ rustc ./foo.rs
./foo.rs:4:11: 4:12 warning: unused variable: `a`
./foo.rs:4       let (a, b) = positions[m];
                      ^
./foo.rs:4:14: 4:15 warning: unused variable: `b`
./foo.rs:4       let (a, b) = positions[m];
                         ^
error: internal compiler error: adt::represent_type called on non-ADT type

[jbclements]

+1, I'm running into this, too.

[jbclements]

in my case, it turned out to be a missing ampersand on the pattern match for the tuple. I spent a bit of time digging around in the compiler to figure out whether it would be possible to at least attach some span info to the error message, but I came up empty.

[nikomatsakis]

This seems like a missing error report to me. Clearly these programs should not be type checking, and those ICEs I believe come from passes that follow type checking. Perhaps an over-suppressed error?

[jbclements]

nominating for Production Ready

[ssbr]

I found this error in a different context, and another ICE in a related context (both for code that shouldn't pass the type checker):

<ssbr> rusti: match Some((1u, 'a')) { (x, _) => 1u, _ => 2u }
<rusti> error: internal compiler error: Asked to compute contents of fictitious type
<ssbr> rusti: match Some((1u, 'a')) { (_, _) => 1u, _ => 2u }
<rusti> error: internal compiler error: adt::represent_type called on non-ADT type

[catamorphism]

I've fixed this in a branch. Just testing now.

[catamorphism]

Fixed pending #6427

[catamorphism]

Fixed in 638b394