Dan <greenash@xxxxxxxxx> writes:
> I'm replying privately as this may become boring for those not
> interested.
It's an appropriate discussion for sisc-devel -> cc'ing that.
> Like you said initially, the "right way" is a lot of work.
Yes, but in SISC we generally prefer to do The Right Thing :)
>> we want all hashtable implementations to be sub-types of a common
>> hashtable base
>
> Sorry if this is obvious, but why the type hierarchy requirement?
The SISC type system is used for things like generic procedures. If I
write a generic proc that takes a hash table as an argument, I'd want to
be able to specify a single type.
> We could let hashtable? return #t only for Java hashtables, and SRFI
> 69 hash-table? return #t both for Scheme hashtables and Java
> hashtables. Luckily, the names of the functions are distinct.
> This would not break any existing SISC-based code, and would allow
> code based on SRFI-69 to run -- if I'm not missing something.
You might as well use the srfi-69 reference implementation in its
entirety, keeping it completely separate from the existing hashtable
implementation in SISC. That is the quickest way to srfi-69
compatibility, but is far from The Right Thing To Do.
Matthias
-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems? Stop! Download the new AJAX search engine that makes
searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click
|
