Re: If you were to implement the original relation algebra language...

Marshall Spight wrote:
> "Bob Badour" <[EMAIL PROTECTED]> wrote in message news:[EMAIL PROTECTED]
>
> > "Marshall Spight" <[EMAIL PROTECTED]> wrote in message
> > news:[EMAIL PROTECTED]
> >
> > > "Bob Badour" <[EMAIL PROTECTED]> wrote in message
> >
> > news:[EMAIL PROTECTED]
> >
> > > > "Marshall Spight" <[EMAIL PROTECTED]> wrote in message
> > > > news:[EMAIL PROTECTED]
> > > >
> > > > > I think we would be better off providing the language with enough
> > > > > flexibility so that the programmer can choose the most appropriate
> > > > > form of expression at the time. cf. Haskell.
> > > >
> > > > Are you suggesting you prefer redundant languages?
> > >
> > > Sometimes I have trouble understanding what you mean
> > > because you are so very terse.
> >
> > You seem to have understood well enough.
>
>
> Sure. But I have to guess what you mean sometimes, and
> sometimes I guess wrong. I think what I was trying to say
> was, forgive me if I guess wrong this time.
>
>
>
> > > I'm happy giving the programmer the flexibility to pick the right
> > > syntax for the current context. YMMV.
> >
> > Fair enough. What contexts favour prefix and what contexts favour infix?
>
>
> I generally reserve infix notation for binary operators that are traditionally
> infix in math textbooks. I suspect, but have not investigated, that
> the operators of the relational algebra would work well in infix form.
>
> Simplistic example with union:
> (use your imagination with the uppercase U :-)
>
> A U B U C
> is decidedly better than
> union(union(a, b), c)
> and possibly better than
> union(a, b, c)
Why is "union(union(a, b), c)" worse than the others? From a user's 
perspective I can see that one might be skeptical about its 
ease-of-use...however, do yo have any research saying that that one is 
decidedly better than the other?
Oh, and if the rel alg is written like this union(union(a, b), c), the 
method signature would be easy: public Relation union(R1, R2){return 
resultrelation;}
> The last one is interesting, but for full generality, it would need
> to accept a variable number of arguments, and I'm suspicious of
> vararg functions.
>
>
> Marshall