RE: [FR-devel] Database slot-type addition

[Rich Kilmer]

> > > Well, I see your point where slot.data={}
> > > is concerned. But for the others, we could
> > > make #hash create an object that was a
> > > subclass of Hash, with the [] and []=
> > > doing notification. Couldn't we? This
> > > would be more convenient than a #get and
> > > a #put, IMO.
> > >

The problem is that the slot[] and []= are used to retrieve subslots.  I
don't want to confuse the issue of slot access ([], []=) and slot
storage.

> > > Actually I guess a 'get' operation doesn't
> > > need to notify a subscriber anyway.
> >
> > I like this... a very useful addition.
> 
> On thinking a little more: Couldn't we
> simplify it even further?
> 
> Change the data= accessor so that when it
> was passed a hash (empty or not) it converted
> it to the subclassed hash (and did any needed
> notification).

I want to add those methods (get/put/hash/is_hash_slot?) to maintain the
symmetry of the API in the way it handles storage/access.

> 
> Shouldn't that work? Then we could get rid of
> #hash, too (though we might need to keep
> #is_hash_slot?).
> 
> Hal
> 
> 
> 
> _______________________________________________
> Freeride-devel mailing list
> address@hidden
> http://mail.freesoftware.fsf.org/mailman/listinfo/freeride-devel