Re: MV Keys
Date: Wed, 08 Mar 2006 06:23:01 GMT
Message-ID: <9LuPf.4406$eJ1.2362_at_trndny05>
"Marshall Spight" <marshall.spight_at_gmail.com> wrote in message
news:1141787070.196520.53850_at_e56g2000cwe.googlegroups.com...
> David Cressey wrote:
> > "Marshall Spight" <marshall.spight_at_gmail.com> wrote in message:
> > >
> > > The RA does not have any operations for looking at elements. :-O
> >
> > But the RA does rely on the existence of an operator that can test two
> > references to elements to see whether or not they refer to the same
element.
> > You can't do all the RA operators, if you can't tell the elements apart.
>
> Definitely. For every domain the RA operates on, it needs a
> value-equality
> test. This might make for some requirements on the type implementor or
> in might be possible to do automatically, but it must be done.
>
>
> Marshall
>
A little while ago we had some back and forth about the "relational engine"
vis a vis the "type engine".
I think your phrase "type implementor" refers to the same division of
labor. "Type implementor" might be better phrased.
How about this phrasing?
"type services implementation"
BTW, I remember some discussion in here about some programming language where relational services are built in, and users can add their own types, but they have to provide and equality test service, and also an ordering test service, in order for the relational services to work on that type. I can't remember the name of the language... (This old age memory stuff is just awful). Received on Wed Mar 08 2006 - 07:23:01 CET