Re: separation of church and state?
Date: Mon, 08 Oct 2007 12:48:28 GMT
"paul c" <toledobythesea_at_ooyah.ac> wrote in message
> Roy Hann wrote:
> > "Marshall" <marshall.spight_at_gmail.com> wrote in message
> > news:1191742694.188301.91520_at_k79g2000hse.googlegroups.com...
> >> On Oct 6, 9:48 am, "Roy Hann" <specia..._at_processed.almost.meat> wrote:
> >>> "paul c" <toledobythe..._at_ooyah.ac> wrote in message
> >>> news:hoONi.6504$_K.2827_at_pd7urf3no...
> >>>> It seems a little doctrinaire to me. I can agree that the "result
> >>>> isn't a
> >>>> relation" but on the other hand a user could see such a result
> >>>> knowing that "ORDER BY" was involved and not be faulted for taking it
> >>>> to
> >>>> be a relation.
> >>> What user ever would? Users never see relations. They see various
> >>> of
> >>> reports (using the word "report" to mean anything users get to see).
> > [goofy stuff snipped]
> >> All that goofy stuff I wrote above can be argued for, but why?
> > That was pretty much my point: Paul was thinking about possibly true
> > but in an irrelevant framework.
> > Roy
> Could be. Maybe one of Date's meanings is that no system that supports
> both ordering and some relational algebra is purely relational, even if
> the "pure" relational part of it could be isolated in some way from the
> rest! If so, calling the paragraph "doctrinaire" might be a bit of a
> Also, I should say that my main interest in this stuff is nearly always
> implementation, this includes making sure that an implementation doesn't
> veer off in directions that aren't sound.
In that case, the implementation of Oracle/Rdb should be of interest to you. The market penetration of Oracle/Rdb was limited by the fact that it was bound to the DEC platform. It's still a worthwhile example of an implementation to study for better or worse.
If it will help at all, In Oracle/Rdb a view with an ORDER BY in it is read-only. This obviates at least some of the problems that theoreticians may have with the concept.
> When Codd himself talked of projection and "desired permutation"
> (admittedly he was talking of columns, not rows) I think an ordinary
> person could be forgiven for not separating the above two "parts" of a
> system. When it comes to implementations, I'd fault one that performed
> a presentation sort that wasn't "any desired permutation" (as Codd put
> it) but I wouldn't blame one that happened to echo some ordering that
> was already present in its "stored representation".
Because of the way data is stored in a computer, what one typically gets for the representation of an unordered set is one of the possible permutations, where the permutation is intended to represent the entire class of permutations of the same unordered set. Ordering the set in some particular way does not change the unordered set that is thus represented.
Unless you work around this problem, you run into a real problem when you try to implement the test for equality. If you are given two different permutations that represent the same unordered set, it can be a real bear to determine that the "equal sets" function should return true. Received on Mon Oct 08 2007 - 14:48:28 CEST