Re: sxhash redux

Adam Warner wrote:
>
> If CLISP limits its 64-bit version of SXHASH to the range of its 32-bit
> non-negative fixnums then it is worthless using the hash code as a index
> to any vector greater than 16777215 in size.
>
> This is a serious limitation upon the performance of any hash table with
> many millions of elements and a practically unacceptable limitation upon
> 64-bit platforms.

The clisp developers agree. This has been fixed in the clisp CVS a week ago
for normal hash tables, and six months ago for package symbol-tables.

Bruno
.