[r6rs-discuss] Hash function return value constraints

[Lynn Winebarger]

On Fri, Apr 3, 2009 at 5:16 PM, leppie <xacc.ide at gmail.com> wrote:

> ----- Original Message -----
> From: "Brian Harvey" <bh at eecs.berkeley.edu>
>
> > Okay, I'm confused,  A hash value is an index into an array.  If the
> > hash value doesn't fit in a fixnum, then the array doesn't fit in memory,
> > or even in the virtual address space (by a factor of 4 on a 32-bit
> > machine).
> > What am I missing here?
>
>
> Something other than the most extreme naive implementation of a hashtable,
> ever.  :)


I'm with Brian - I haven't been following the "hash table" literature, but
to the best of my knowledge its normal usage refers specifically to
associations recorded in an array.  Maybe the term has acquired a wider
meaning - certainly R6RS does not retain this standard meaning - but I think
it's hard to claim the standard definition of X is "the most extreme naive
implementation of " X.



It's up to the implementation to choose how it converts the hash value into
> a meaningful index (or perhaps something else).
>

If both R6RS conformance and the extra bit is so important, why not add a
special unsigned fixnum to your internal data types?

Lynn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.r6rs.org/pipermail/r6rs-discuss/attachments/20090403/cb271709/attachment.htm