Support for pure functional types.
I've been wondering for some time why there is no support for pure functional types. Like function int -> int. In most type inference techniques, there is. I know, side effects. But usually we can tell that function has no side effects at all, so it would be nice to reduce it to lambda type, not just remember to enter it's block. It would make some things much easier. Anyway: when calling some lambda or local function not all bindings are clear (depends on context), but we can leave it and got bidings later and use it normal way (not degenerate return value to some-object, which happens usually)
Hi Maciek, On Sun, Jan 22, 2006 at 07:35:01PM +0100, Maciek Fijalkowski wrote:
I suppose you are referring to supporting dynamic closures in RPython. We don't support them mostly because we didn't need them so far. The first step in supporting them would be to figure out how they should be implemented at the C level. Everything else can be worked out upwards from there. A bientot, Armin
Hi Maciek, On Sun, Jan 22, 2006 at 07:35:01PM +0100, Maciek Fijalkowski wrote:
I suppose you are referring to supporting dynamic closures in RPython. We don't support them mostly because we didn't need them so far. The first step in supporting them would be to figure out how they should be implemented at the C level. Everything else can be worked out upwards from there. A bientot, Armin
participants (2)
-
Armin Rigo -
Maciek Fijalkowski