Consider freezing cats-kernel

[non]

Hi folks,

As per the comments on this algebra PR (typelevel/algebra#162) I want to start thinking about freezing cats-kernel. By freezing, I mean turning on MiMA compatibility checking, and committing to keeping the API stable for a relatively long timeframe (e.g. a year or more).

Before we do this I want to be be sure that the design of cats-kernel and algebra works for Spire (and Algebird too). We should also be sure that it works for Cats.

Has anyone identified problems in the design of cats-kernel that would break compatibility to fix? Is there anything people think is essential to add or change? I know the design is slightly inconsistent with Cats (necessarily for compatibility reasons). I'm asking more about the names, methods, and type classes than about these kinds of consistency concerns, but if you feel strongly about those feel free to bring them up here.

What do you all think? From a Cats-centric point of view, is cats-kernel ready to be declared done?

[non]

I should add that I think freezing cats-kernel is a required precursor to cats-1.0.0 so the process I mention about freezing could happen relatively quickly.

[adelbertc]

One thing I've noticed is the style of laws exposed in kernel-laws differs from that of cats-laws. I have a ticket to track this: #1272

[non]

@adelbertc I'm open to doing this, but I want to get the algebra team on board first. (I think the Cats way of defining laws is cleaner, but since we tore kernel out of algebra I wanted to do the minimally-disruptive thing.)

[kailuowang]

Should this one be closed now? according to this https://github.com/typelevel/cats/blob/master/build.sbt#L222 kernel is locked at 0.8.0

[ceedubs]

I think that this is resolved. @non feel free to reopen if I'm mistaken.