Add mima for kernel

[johnynek]

This adds mima for kernel, but it does not pass, so it is not added to travis yet.

As discussed when moving kernel from algebra, the thinking was we could be more conservative there. I would like to turn this on after the next release (I guess 0.7.0) and I hope we can be conservative with kernel updates so that algebra is not too tightly coupled to cats.

[johnynek]

relates to #814 (does not close, since we have failing test case now).

[codecov-io]

Current coverage is 88.80%

Merging #1159 into master will not change coverage

@@             master      #1159   diff @@
==========================================
  Files           232        232          
  Lines          3073       3073          
  Methods        3021       3021          
  Messages          0          0          
  Branches         49         49          
==========================================
  Hits           2729       2729          
  Misses          344        344          
  Partials          0          0          

Powered by Codecov. Last updated by c4f5d1c...3af6587

[kailuowang]

We could also make the previousArtifacts dynamically check previous bug fix version, like

  lazy val mimaSettings = Seq(
    mimaPreviousArtifacts := (if (scalaBinaryVersion.value != "2.10") {
      Version(version.value).flatMap {
        // we are currently only checking binary compat with the previous bugfix version
        case Version(major, Some(minor), Some(bugfix), _) if bugfix > 0 =>
          Some(Set(organization.value %% moduleName.value % Seq(major, minor, bugfix - 1).mkString(".")))
        case  _ => None
      }
    } else None).getOrElse(Set.empty)
  )

[benhutchison]

In my workplace, many Scalaz/Argonaut-based projects have become binary- and source- compatibility minefields. Pleased to see cats taking version compatibility seriously from the outset; I hope concern from compatibility becomes part of Cats' "cultural DNA".

[johnynek]

big +1 to this. binary compatibility should be taken very seriously for a library like cats that hopes to be at the bottom of a lot of dependencies.

[johnynek]

@kailuowang #1159 (comment) that is a great idea. In fact, I think we can do something like this to verify semver:

if major == 0 then we check previous bugfix, if major > 0 we check previous bugfix and previous patch at version 0 perhaps?

That said, my overall point is that I think cats-kernel should enforce binary compatibility sooner (maybe like now), so that algebra can move close to a 1.0 release.

[kailuowang]

Sorry not sure I follow completely. Are you suggesting that for kernel we maintain binary comp between minor versions between 0.7.0 and 1.0.0 ?

[johnynek]

Yes, I am. That was my understanding when we discussed moving that code
from algebra to cats (since it has changed very little in two years, this
does not seem onerous).

On Saturday, June 25, 2016, Kai(luo) Wang notifications@github.com wrote:

Sorry not sure I follow completely. Are you suggesting that for kernel we
maintain binary comp between minor versions between 0.7.0 and 1.0.0 ?

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#1159 (comment), or mute
the thread
https://github.com/notifications/unsubscribe/AAEJdmQq7UieVdS2tgCSTeUlCo89hNuvks5qPdpLgaJpZM4I-OQp
.

P. Oscar Boykin, Ph.D. | http://twitter.com/posco | http://pobox.com/~boykin

[ceedubs]

👍 thanks @johnynek. I'll create a follow-up issue to enable this after the next release.