On Tue, Jul 2, 2013 at 4:07 PM, Patrick Walton <[email protected]> wrote:
> On 7/1/13 5:32 AM, Corey Richardson wrote: > >> Variable width floating point code is also dangerous - frequently code >> has implicit assumptions about accuracy. Using floating point >> correctly is already hard enough, introducing variable width types >> makes it even harder. >> > > I proposed this in the past, and Graydon objected on the principle that > it's similar to `int` in that `float` is the size of a register, and it > leaves the door open for 128-bit floats if CPU manufacturers ever go to > that. I'd still personally be OK with it though: C doesn't have a concept > of "floating point type as big as a register" (rather float is de-facto > 32-bit and double is de-facto 64-bit). As far as I know it will probably > never acquire such a thing, so it seems best to me to just follow C here > and only support fixed length floating point types. > Seems to me you could quantify the performance impact of variable-width types. E.g. try making 'int' 32-bit on a 64-bit architecture and see how much it hurts. I'm suspicious they're a premature optimization. Rob -- Jtehsauts tshaei dS,o n" Wohfy Mdaon yhoaus eanuttehrotraiitny eovni le atrhtohu gthot sf oirng iyvoeu rs ihnesa.r"t sS?o Whhei csha iids teoa stiheer :p atroa lsyazye,d 'mYaonu,r "sGients uapr,e tfaokreg iyvoeunr, 'm aotr atnod sgaoy ,h o'mGee.t" uTph eann dt hwea lmka'n? gBoutt uIp waanndt wyeonut thoo mken.o w * *
_______________________________________________ Rust-dev mailing list [email protected] https://mail.mozilla.org/listinfo/rust-dev
