[ghc-steering-committee] #265: Unlifted Datatypes, recommendation: accept

[Simon Marlow]

I recommend that we *accept* proposal #265 (Unlifted Datatypes)

https://github.com/ghc-proposals/ghc-proposals/pull/265
https://github.com/sgraf812/ghc-proposals/blob/unlifted-data/proposals/0000-unlifted-datatypes.rst

It's a fairly conservative extension: the kind TYPE 'UnliftedRep already
exists with the required functionality, the only addition here is to allow
user-defined types to be declared with that kind. The semantics are clear,
and there already exists a prototype patch to implement it.

There are considerable performance benefits to be had for
performance-critical code, for instance the containers package.

A couple of minor issues remain:

   - Without special support, the type data unlifted Strict a = Force !a
   comes with an associated box, so this type isn't as useful as it could be.
   - It isn't possible to define values of kind TYPE 'UnlifedRep at the top
   level, which might be a surprising restriction to the programmer. (However,
   there's a reasonable workaround). Relatedly, GHC cannot lift expressions of
   kind TYPE 'UnlifedRep to the top level in the optimiser, which can lead
   to surprising performance behaviour. See
   https://gitlab.haskell.org/ghc/ghc/issues/17521

Nevertheless, we shouldn't let the perfect be the enemy of the good, and
Unlifted Datatypes is a clearly useful addition in my view.

Cheers
Simon


On Thu, 28 Nov 2019 at 10:06, Joachim Breitner <mail at joachim-breitner.de>
wrote:

> Dear Committee,
>
> this is your secretary speaking:
>
> Unlifed Datatypes
> has been proposed by Sebastian Graf
> https://github.com/ghc-proposals/ghc-proposals/pull/265
>
> https://github.com/sgraf812/ghc-proposals/blob/unlifted-data/proposals/0000-unlifted-datatypes.rst
>
> I propose Simon Marlow as the shepherd, as the expert on low-level stuff.
>
> Please reach consensus as described in
> https://github.com/ghc-proposals/ghc-proposals#committee-process
> I suggest you make a recommendation, in a new e-mail thread with the
> proposal number in the subject, about the decision, maybe point out
> debatable points, and assume that anyone who stays quiet agrees with
> you.
>
> Thanks,
> Joachim
> --
> Joachim Breitner
>   mail at joachim-breitner.de
>   http://www.joachim-breitner.de/
>
> _______________________________________________
> ghc-steering-committee mailing list
> ghc-steering-committee at haskell.org
> https://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-steering-committee
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-steering-committee/attachments/20191202/c05207cb/attachment.html>