#14770: Allow static pointer expressions to have static pointer free variables -------------------------------------+------------------------------------- Reporter: TheKing01 | Owner: (none) Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.2.2 Resolution: | Keywords: | StaticPointers Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by gershomb): No concrete use-case at the moment (in that I'm not working on a system that does cloud distribution). I wanted to track this ticket as I've been thinking a lot about questions related to this ticket, including possible other uses for `static` and how it relates to Contextual Modal Type Theory (and also the question of staging). I think the ticket has the right of it in terms of the problem. Semantically, the "correct" thing to do is, treating `StaticPtr` as a modality, allow computation _within_ the modality such that we have `StaticPtr (a -> b) -> StaticPtr a -> StaticPtr b` directly. Instead, we have the workaround the ticket points to, which makes use of an opaque GADT. So that means, as the ticket says, this gives (on its own) a potential efficiency gain, but not expressivity gain. I think we'll need more people using static pointers "at scale" to see if the efficiency cost becomes a genuine pain point. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/14770#comment:5 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler