[MLton] Unicode / WideChar

Wesley W. Terpstra wesley@terpstra.ca
Wed, 23 Nov 2005 02:15:14 +0100


On Nov 23, 2005, at 1:51 AM, Wesley W. Terpstra wrote:
> Now, for the map, I propose that we use a minimal perfect
> hash.

Hmm. Or perhaps an 'optimal bloom filter'.
(Hadn't heard of these till reading recent perfect hash papers)

I'd still need the minimal perfect hash for the to{Upper,Lower}.
However, this could perhaps reduce the space for the is*
methods to one bit per member. The executable overhead
would remain the same, but maybe only  20k + 5k would
be necessary for the to* and is* methods respectively.

Actually, an optimal bloom filter might also remove the need
for the second WideChar in the to* maps.

Must... read... more...