two possible coding porjects

Karl Hasselström kha@treskal.com
Tue, 29 Jul 2003 02:10:10 +0200


--HcAYCG3uE/tztfnV
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On 2003-07-29 08:58:27 +1000, Kevin Ryde wrote:

> stephen-nuchia@utulsa.edu writes:
>=20
> > this would at the very least admit error reporting when
> > non-reallocable storage overflows.
>=20
> You mean at the mpn level? Routines using those must simply take
> care not to go past their defined operands. Some of the hairier ones
> have some ASSERTs to see that theory accords with practice, but
> that's about the limit of the checking we think is needed.

As for temporary space needed _during_ the calculation, each mpn
routine allocates memory for itself as it is now. As far as I know,
that'll gradually change into a system where each mpn routine takes a
pointer to all the scratch space it needs as an argument, thereby
reducing allocation overhead and/or use of stack space (depending on
whether heap or stack memory was allocated), and hopefully reducing
memory use through reuse of temp space.

--=20
Karl Hasselstr=F6m, kha@treskal.com
      www.treskal.com/kalle

--HcAYCG3uE/tztfnV
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQE/JbtiN7iPzXSoOQoRAhBMAJ9ONO0whwWaYV6OWPo11XNk4ycGDQCfWN9N
0mw/6yGEQNN/Ko3jUKafieg=
=mmny
-----END PGP SIGNATURE-----

--HcAYCG3uE/tztfnV--