Macros attempting to expand to multiple items silently only expand to the first one

[paulstansifer]

Users who want a macro that expands to a collection items are required to wrap the RHS in {} in order to get them to expand to a whole block. However, if they fail to do so, they unexpectedly just get the first item of the block, rather than an error message.

macro_rules! print_if_nonzero(
    ($msg:expr, $($field:ident),*) => ( /*{*/
        $(
            if t.$field != 0 {
                println($msg);
            }
        )*
    /*}*/ )
)

struct T { x: int, y: int }

fn main() {
    let t = T { x: 0, y: 0 };
    print_if_nonzero!("hello", x, y);
}

...expands to...

fn main() { let t = T{x: 0, y: 0,}; if t.x != 0 { println("hello"); }; }

[paulstansifer]

Thanks to kmc on IRC for reporting this.

[bstrie]

Perhaps this is really just another plaintive cry for someone to tackle #4375.

[huonw]

Smaller testcase:

rusti> macro_rules! t( () => { println("hi"); println("oops"); }) t!()
hi
()

[nejucomo]

I don't understand how this is different than #4375. If not justification for the distinction is given, can we mark this as a duplicate?

[paulstansifer]

This source of this problem probably isn't specific to items. I'm worried
that there's a lurking (if minor) usability problem in the lack of erroring
out in this case.

[kmcallister]

Yeah, jason_ on IRC provided this example using tuples:

macro_rules! omg(($n:expr) => ($n,$n+1))
omg!(100) // => 100, not (100, 101)