Is this supposed to work?

Void test()
{
  Obj i := 0

  ((Int) i)++
}

It results in a Expected expression statement compile error.

Casting in Fantom is done with the as keyword:

foo as Bar

But your example still won't compile even with as:

class Foo
{
  Void main()
  {
    Obj i := 0

    // won't compile
    //(i as Int)++

    z := (i as Int)
    z++

    echo("i: $i")
    echo("z: $z")
  }
}

But this does not increment i

Promoted to ticket #2012 and assigned to brian

That is just the compiler complaining that it thinks the increment is a no-op and an expression that doesn't map to a complete statement (because the increment is mixed in the with the cast). I think it just a tweak to the compiler that needs to a cast expression to pass thru in the check for Expr.isStmt.

Its actually a simple patch if someones to try their hand at making a compiler fix (with a test case!)

I actually didn't know that java-style casts worked in Fantom, I though you had to use as. I like as a lot better.

Anyway, I'll take a stab at this fix unless Jens wants to.

Hm. Trying to use the increment method instead, which ++ is supposed to be a short cut for, I notice that it doesn't actually increment the receiver, but return the new value:

Void test()
{
  Int i1 := 0
  Int i2 := i1.increment
  echo( "i1: " + i1 )   // Prints "i1: 0"
  echo( "i2: " + i2 )   // Prints "i2: 1"
}

This also seems to be a bug?

Also, the documentation for sys::Int.increment does not specify if it should return the value of the receiver before or after the increment.

Anyway, I'll take a stab at this fix unless Jens wants to.

I had a quick look. The code in isStmt seems fine. And the error is coming from the parser, in exprOrLocalDefStmt.

(In the case when isStmt fails in the CheckErrors phase, Not a statement is reported instead)

It seems like it thinks that the whole statement ends with the cast, failing to associate ++ with the same statement. So it reports the cast with Expected expression statement.

I find the parser a bit tricky to understand...

I notice that it doesn't actually increment the receiver, but return the new value:

Int, the receiver is a const type (same as all the Num types), so its value will never change - it has to return the new value.

Int, the receiver is a const type

Ah, of course, I see, thanks. It's a bit confusing with Ints double nature, value/const object.

Ticket cancelled

After looking at this a bit more, I'm thinking that it might be best to leave this a compiler error. The ++/-- operators a bit complicated in how they work on the runtime stack, and having a cast in there actually creates a lot of complication in the fcode b/w JVM and CLR runtimes.

It sure is a rather uncommon thing to do.

But it might be a good idea to have a better error message for it. Otherwise there might be a new bug-report on it some day.