Re: Date and McGoveran comments on view updating 'problem'
Date: Mon, 15 Dec 2008 12:25:01 -0800
Message-ID: <N1z1l.5522$E46.2642_at_newsfe21.iad>
Brian Selzer wrote:
> "paul c" <toledobythesea_at_oohay.ac> wrote in message
> news:Wve%k.3357$yK5.661_at_edtnps82...
...
>> Heath's Theorem ( see the first question at >> http://www.bridgeport.edu/sed/fcourses/cs450/Assignments/Selected_Problems_From_Ch11.pdf) >> that D must be equal to any of its projections that include {S#, P#} so if >> A and B include the S# and P# attributes, they must have only one row >> each, otherwise D would have more than one tuple/row.
>
> Stop right there. Your analysis is faulty. Even if A and B include the S#
> and P# attributes, it is still possible for either A or B or both to have
> more than one row. Consider the counterexample:
>
> A: {{S#:'S1', P#:'P1'}, {S#:'S1', P#:'P2'}}
> B: {{S#:'S1', P#:'P2'}, {S#:'S2', P#:'P2'}}
> ...
Well, I wasn't very clear there, I had in mind the 'scope' that
McGoveran refers to. Since that post, I've been trying to write one for
a new thread that tries to pin it down better.
,,,
>
> Your application here of Heath's theorem is strange, but possibly sheds some
> light on the flaw in your reasoning. ...
It may seem to have come out of the blue but that may change whenever I post my latest attempt to make Date's and McGoveran's comments more concrete. In it, Heath is necessary, I don't say always necessary for anybody else, just convenient for me to explain my 'purely algebraic' conclusions so far, which are basically that McGoveran's comments are close to at least a partial solution to the general problem. Received on Mon Dec 15 2008 - 21:25:01 CET