[MLton] MLton rules! (was: filedes = int)

skaller skaller@users.sourceforge.net
Tue, 19 Jul 2005 05:59:23 +1000

Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable

On Mon, 2005-07-18 at 11:20 -0500, Henry Cejtin wrote:
> No, I put the GMP support in to MLton exactly because I wanted fast large
> multi-precision integers.  I need the speed.

We're talking about the *compiler*, not generated code.

> I agree that you could re-implement the library using the same representa=
> in the compiler, because the compiler doesn't do any thing that much with
> big integers, but why bother.=20

Two reasons come to mind:=20

(a) portability: tools dependent on other tools and libraries
are much harder to port -- especially stuff that uses
as many C macros and autoconf lunacy as GNU code.

(b) licence

> Also, why not take advantage of the speed.

It isn't clear it is even remotely relevant to the
compile times of most programs, and those for which
it is relevant should probably be taken out and shot :)

> Even more importantly, why depend on the format of libgmp numbers in a
> separate implementation.

I never suggested that.

John Skaller <skaller at users dot sourceforge dot net>

Content-Type: application/pgp-signature; name=signature.asc
Content-Description: This is a digitally signed message part

Version: GnuPG v1.2.5 (GNU/Linux)