Add F[TupleN] syntax

[danicheg]

This adds convenient _1F..._22F methods for F[Tuple3..22]. At the moment, Cats contains those methods only for F[Tuple2], so it would be great to have it for all Tuple families.

I'm open to any suggestions/points about this.

[satorg]

As I understand, the worst concern regarding changes like this one is a potential increase of the result jar size.
Did you happen to estimate by chance how significantly the sizes got increased?

Link to a similar PR: #4009

[johnynek]

this seems like a PR for mouse if you ask me:

https://github.com/typelevel/mouse

which, BTW, I would live to see merged into this repo since I believe cats is the only dependency. Then it would just be a submodule here.

[danicheg]

@johnynek yeah, I was thinking about this, but it just confusing a bit to have F[Tuple2] syntax in the Cats and syntax for F[TupleN] in the Mouse (or have intersected on the same methods in both libraries).

[danicheg]

Can we take yet another look at this? I have a concern about having this in the Mouse. It's just odd to have _1F, _2F for Tuple2 in Cats, but for TupleN in Mouse. But if it's common for all new syntaxes within cats.syntax, I doubtless will migrate this to Mouse.

[armanbilge]

It's just odd to have _1F, _2F for Tuple2 in Cats, but for TupleN in Mouse.

Yep, I agree with this. If this is the primary reason to target this PR to cats, maybe the real question is do _1F and _2F belong in Cats? IDK the history but maybe at the time they were added there was no mouse yet.

As I understand, the worst concern regarding changes like this one is a potential increase of the result jar size.
Did you happen to estimate by chance how significantly the sizes got increased?

Possibly off-topic, but I haven't been able to appreciate why this is such a big deal. See my comments in #3871 (comment). But I would like to know if there's something I'm not getting :)

[danicheg]

IDK the history but maybe at the time they were added there was no mouse yet.

Mouse is about six years this May :) And _1F and _2F were added a year ago.

[armanbilge]

I see, those methods were added in #3318.

[johnynek]

I think in the past reviewers (such as me) did not think about mouse.

Adding 1F, 2F also can be related to the idea that tuple2 is special: we do have tuple2 FlatMap for instance. We might not necessarily have TupleN FlatMap.

I think considering Tuple2 as special, since it is kind of the primitive product type, but argue that in scala2, higher arities should go in mouse.

But it is hard to draw a line on how much we want to trade off jar size for functionality. It would be interesting to consider how we could have a framework for these kinds of PRs.

[danicheg]

If we speak about the end-user experience it'd be totally bizarre to add a new library as a dependency to get syntaxes for TupleN (this and any future ones), as far I can see it. So if it's a clinched question for all these syntaxes, I won't make waves.