Even aside from committee feedback easily being weeks away for such a complex proposal, I’m also concerned about how the current proposed design changes to core seem a bit fragile with linear types. Perhaps I’m misunderstanding some of the current planned details , but I’m fairly confident that I’ve got a median or better comprehension.
Type checking as if the code were inlined (per join points and related case expressions as the linear core doc says ) tends to be a symptom of the types not quite modelling the right information. Likewise would not that sort of checking create a possible quadratic blowup when linting/ type checking core? (And quadratic blowups are bad when debugging/checking possibly large core programs in the core of any ghc debugging or the like !)
That said, putting a check point on phab for feedback of a technical sort is def something that would help. The ghc proposal spec is vaguer than I’d like for something like this. And a lot of important details I care about will be visible in the code that are lacking in the associated proposal and paper.
Matthew Pickering <matthewtpickering@gmail.com> writes:
> Perhaps Nested CPR will be ready :) ? https://phabricator.haskell.org/D4244
>
> I am also working on the linear types branch. Arnaud is quite keen for
> it to be ready for 8.6 but we still have a bit to go.
>
I'll admit that I'm a bit worried that the linear types branch may be a
bit late given that the proposal only went to the committee last week.
That being said, I'm happy to keep all options on the table.
Regardless, it might be a good idea to put up a patch sooner rather than
later so we can begin the review process.
Cheers,
- Ben
_______________________________________________
ghc-devs mailing list
ghc-devs@haskell.org
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs