nikita-volkov / base-prelude Goto Github PK

1. New module: Control.Monad.IO.Class
2. New modules: Data.List.NonEmpty and Data.Semigroup
3. New modules: Data.Functor.Classes, Data.Functor.Compose, Data.Functor.Product, and Data.Functor.Sum
4. New module: Control.Monad.Fail
5. traceM, traceShowM (which we reimplement) have been generalised to Applicative – should we generalise them as well for older versions of base, or not? (I lean towards “not”, because together with #15 it would lead to weirdness – traceM would have an Applicative constraint when base-4.6 is used, Monad when base-4.7 is used, and Applicative again when base-4.9 is used.)
6. New function: forkOSWithUnmask (could be reimplemented)

I appears as if BasePrelude violates the PVP contract by advertising a version 1.0.1 and yet exposing a greatly fluctuating API (and not even placing a sensible upper bound on base anymore to contain the risk).

I'd like to remind that https://hackage.haskell.org/ states that "All packages should follow the Package Versioning Policy (PVP)." and base-prelude does not state in its description that it deviates from the PVP contract.

I'm worried because packages on Hackage are starting to make use of base-prelude and without proper guidance that will cause more busywork for Hackage Trustees to keep install-plans from bitrotting due to inaccurate version-bounds.

I'm using GHC 7.10.3 I don't know if it's a regression or not but they are already in the Prelude, so they need hidding.

My understanding is that base-prelude is meant to be mostly a drop in for Prelude. Prelude exports FilePath, but base-prelude explicitly hides this. Is there any reason for that?

As per: https://hackage.haskell.org/package/base-4.12.0.0/docs/Data-Semigroup.html#t:Option

No single version of base-prelude compiles with GHC 7.4.2, and yet the cabal file accepts base-4.5.

See https://ghc.haskell.org/~hvr/buildreports/base-prelude.html for details (look at the HTML source if the mouse-over text popups aren't readable). I've already edited all the base-prelude releases files on Hackage to have a tighter base >= 4.6 constraint.

Call it a feature request. I like getArgs and getEnv in particular.

base-4.8:

• Numeric.Natural
• Data.Void (moved from void)
• Data.Functor.Identity (moved from transformers)

base-4.7:

• Data.Proxy
• Data.Coerce
• void was added (could be reimplemented)
• isLeft, isRight were added (could be reimplemented)

Should I make a PR?

There don't appear to be any clashes, and this is a very useful thing to have available. Why not re-export it as well?

(I like printf, but if you have intentionally decided not to export it, I won't complain. Otherwise, having it would be nice.)