Integrate with either

[rust-itertools]

Integrate with either

[Stebalien]

Alternatively, one could name these functions interleave_{left,right} etc. to allow for more control.

[bluss]

Interesting idea. Either is certainly Iterator-related, it exists partly because I didn't want to reimplement the two-case iterator many times.

There's some overlap between the crates. itertools::Partition could just have been the Either enum. So for that reason it seems like a good idea. It's also a very small crate to add as a dependency. It has the risk of thrusting the either crate and type into a lot more rusties' hands, I wonder what they will think.

The left, right iterators don't seem necessary, since the .left(), .right() methods return options that are iterable. So that functionality is available already through composition.

The extend line would not actually be possible would it, since String::extend doesn't use any generics or AsRef<str> or so, it needs an iterator of &str.

[Stebalien]

The extend line would not actually be possible would it, since String::extend doesn't use any generics or AsRef or so, it needs an iterator of &str.

See rust-lang/rust#34143 (whether or not this gets accepted, or even should be accepted, is another issue).

The left, right iterators don't seem necessary, since the .left(), .right() methods return options that are iterable. So that functionality is available already through composition.

I'm not sure I follow here. I'm suggesting a helper function and type equivalent to iterator.map(Either::{Left, Right}). This is mostly to get namable types (i.e., avoid functions in types).

[bluss]

It sounds like you mean either Either::Left as a function (that already exists) or the method Either::left composed with .unwrap() (Either::left already exists)?

[Stebalien]

I mean Either::Left as a function. Basically, first.chain_either(second) would be shorthand for first.left().chain(second.right()).

[bluss]

aha. So .left() would be an iterator adaptor equivalent to .map(Either::Left)

[Stebalien]

...caps (yes).

[bluss]

The breaking change part of this is going to be done for 0.5. That's changing out Partition for the Either enum

[Centril]

I'm personally partial to / and biased in favor of the frunk crate which is more general in that it provides support for "anonymous coproducts" with arbitrary many variants.

[bluss]

@Centril sounds interesting, but frunk will not be a dependency of itertools, so we can't do so much with it.