Why does izip!(iter) not return 1-tuples?

[ExpHP]

When given a single Iterator, izip just does into_iter.

I know it has been like this from the very beginning, but I never understood why. There isn't really much of a use case for calling izip with one iterable in real code ; the only time this ever happens is when another macro uses izip, in which case the lack of 1-tuples invariably just causes trouble.

Given some matched repetition like $($ident)+, I see no straightforward way to construct a pattern for these items.

#[macro_use] extern crate itertools;

macro_rules! fmt_columns {
    ($fmt:expr, $($c:ident),+ $(,)*) => {
        izip!($($c),+)
            // nope, matches A with (a,)
            // .map(|($($c,)+)| format!($fmt, $($c),+))
            
            // nope. (a) is not a valid pattern
            // .map(|($($c),+)| format!($fmt, $($c),+))
            
            // ???
    }
}

(I could've sworn this also used to be a problem with types as well, but it appears that (T) is a valid type in the latest stable)

[bluss]

There's no why. Best to make the case for a change (or not) if it's wanted

[ExpHP]

Right. The case for a change is because izip!(a) is not needed in handwritten code, but naturally arises in macros. 1-tuples compose better with macros because:

• it allows them to write ( $($a,)* ).
• it eases the use of a trait as an implementation detail of a macro, because an impl involving (A,) (for all A) does not overlap with an impl involving (A, B) (for all A, B) or etc.. In contrast, an impl involving A tends to overlap with anything else.

[bluss]

This should have the breaking-change label so that we find it when doing breaking version updates.

@ExpHP You have a good argument. My only counter is that it's pretty easy to have macro code adapt to the unary case not being a tuple, at least in pattern matching, I've done that several times. There it is just ($($x,)*) vs ($($x),*)