Experiment: convert any CE effect type to a bifunctor effect

[neko-kai]

Seems to be working: cats-effect laws pass, but there may still be problems they don't reveal

[codecov-commenter]

Codecov Report

Merging #1766 (1c1553e) into develop (1623459) will decrease coverage by 1.80%.
Report is 2 commits behind head on develop.
The diff coverage is 61.03%.

❗ Current head 1c1553e differs from pull request most recent head 8a1c2ec. Consider uploading reports for the commit 8a1c2ec to get more accurate results

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

@@             Coverage Diff             @@
##           develop    #1766      +/-   ##
===========================================
- Coverage    65.39%   63.59%   -1.80%     
===========================================
  Files          557      479      -78     
  Lines        11061     9636    -1425     
  Branches         0      491     +491     
===========================================
- Hits          7233     6128    -1105     
+ Misses        3828     3508     -320     

[mucaho]

Would it be possible to generalize this even further?

Namely, given any monofunctor F[_] and a corresponding cats type class instance e.g. cats.Applicative[F], generate the respective BIO type class instance [+E, +A] => izumi.functional.bio.Applicative2[Bifunctorized[F, E, A]] for it.

Repeat that for all conversions in izumi.functional.bio.CatsConversions, just the other way around

[neko-kai]

@mucaho
Sure, just need to make sure it's sound first.

[mucaho]

Great to see this is still being worked on, thanks!

While this was being worked on, I have created my own two conversions I needed in the meantime, not sure if they are completely sound though, in case anyone needs: https://bitbucket.org/mucaho/ceffbanx/src/8986109fb7e9e97c7ae5a71ba311ff9d915634b8/src/main/scala/org/bitbucket/mucaho/ceffbanx/Util.scala#lines-130:180

[neko-kai]

cats-mtl added "submarine error handling" in similar vein - typelevel/cats-mtl#619 associated blogpost
The distinction between cats.Submarine and bio.TypedError is that Submarine uses an additional AnyRef to ensure valid nesting of handlers. Not sure whether that's a relevant addition for bifunctors.

[neko-kai]

NB: nesting of handlers may not be relevant for bifunctorization. But distinguishing between different bifunctorized effect types may be. (we could just use TagK instead of path-dependency?)