John Cowan <xxxxxx@mercury.ccil.org> writes:

> Taylan Ulrich Bayırlı/Kammer scripsit:
>
>> I just don't see much of an improvement in changing hash function
>> signatures to something entirely novel, when '(obj, [bound]) -> int'
>> works perfectly fine and is familiar for programmers and more compatible
>> with existing implementations.
>
> Even this is problematic, though.  For example, the Larceny implementation
> of SRFI 125 (and R6RS and SRFI 69) does not pass a "bound" argument to a
> hash function.  If one is written to demand such an argument, it will break.
> This is what I mean about a fixed protocol.  If (I say again) the framework
> were able to ask the hash function what arguments it accepts, no problems
> with "bound" or "salt" arguments would arise.

The bound argument is optional in SRFI-126, as explained in the "hash
functions" section.  I think I should clarify that in the specification
of 'make-hashtable' as well though.

Taylan