Re: [ghc-steering-committee] #390: Fine-grained pragmas, recommendation: accept

Dear Committee, I'm happy to repeat my recommendation to accept Alejandro's proposal which is now back from revision: Fine-grained pragmas for classes, families, and instances https://github.com/ghc-proposals/ghc-proposals/pull/390 https://github.com/serras/ghc-proposals/blob/instance-pragmas/proposals/0000... All the issues that arose in the previous discussion are now resolved. Does anyone have something to add? Vitaly сб, 13 мар. 2021 г. в 12:38, Vitaly Bragilevsky :

чт, 11 мар. 2021 г. в 16:31, Vladislav Zavialov (int-index) < vlad.z.4096@gmail.com>:

...

I like the proposal and I would be happy to vote for its acceptance. However, it requires a little bit of polishing up with regards to its use of modifiers. In particular, it seems to treat modifiers as built-in units, whereas the core idea behind modifiers is that they are based on actual promoted types (with the exception of %1).

Hence it seems appropriate to send it back for revision until this point is addressed.

Yes, I agree. There is also another issue. I think we need either a deprecation story for UndecidableInstances, or a statement that it should survive with some motivation behind that.

So, I'm sending this proposal back to Alejandro for review and set the "needs revision" label on GitHub.

Vitaly

...

- Vlad

...

On 20 Feb 2021, at 14:34, Vitaly Bragilevsky wrote:

Dear Committee,

Our own Alejandro has been proposed Fine-grained pragmas for classes, families, and instances https://github.com/ghc-proposals/ghc-proposals/pull/390

https://github.com/serras/ghc-proposals/blob/instance-pragmas/proposals/0000...

...

His idea is to bring the flexibility of

overlaps/overlapping/overlappable pragmas to termination checking, type inference, and constraint solving. Alejandro proposes to introduce the following %-modifiers (as in already accepted #370, https://github.com/ghc-proposals/ghc-proposals/blob/master/proposals/0370-mo... ):

...

%NoTerminationCheck %LiberalCoverage %LiberalInjectivity %Overlapping %Overlappable %Overlaps to liberate conditions for classes and instances, type families, forall-types, etc. The first three modifiers can be used instead of the scary-sounding UndecidableInstances extension. The last three modifiers are supposed to be used instead of the overlap*-pragmas for instances we already have. Note that this proposal doesn't suggest deprecating those extensions and pragmas.

I think that this proposal goes in the right direction and recommend accepting it.

As far as I understand, all the committee members, including those with terms about to expire, have the right to either support this proposal or to raise a voice against it either here or at https://github.com/ghc-proposals/ghc-proposals/pull/390.

Regards, Vitaly

_______________________________________________ ghc-steering-committee mailing list ghc-steering-committee@haskell.org

https://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-steering-committee