David Cressey wrote:
> "Marshall" <marshall.spight_at_gmail.com> wrote in message
> news:1176756330.363208.181810_at_o5g2000hsb.googlegroups.com...
>
>>On Apr 16, 1:20 pm, "David Cressey" <cresse..._at_verizon.net> wrote:
>>
>>>>Something like that. Requires RVAs and an aggregate union.
>>>>Clearly no standard way to do in SQL, but there might be
>>>>some hack that lets you do the union as a string operation
>>>>which might be sufficient for some applications.
>>>
>>>OK, but what I offered was not a hack. For once, I was trying to give a
>>>serious response to this kind of question.
>>>
>>>(You may not recognize it as such, but it's the same question that Dawn
>
> has
>
>>>raised a dozen times in the last few years.)
>>>
>>>Pushing the result of a SQL query through a hierarchical report writer
>
> is
>
>>>really not hacking. It's serious work, albeit without much theoretical
>>>merit at all.
>>
>>Oh, totally. I didn't intend to detract from your answer at all.
>>Your answer was practical; mine was almost entirely impractical.
>
> Then again, I'm thinking that maybe there is a lesson for theoreticians to
> learn in this exchange.
>
> Turning relational data into hierarchical data is "trivial", in every sense
> of the word. turning hierarchical data into relational data is, in some
> cases, non trivial. This suggests that storing database data in relational
> form is better than storing it in hierarchical form, provided there is some
> utility in doing so.
>
> Most theoreticians already know this, and know it better than I have
> verbalized it. But it's worth making it explicit.
Actually, expressing a hierarchy in relations is trivial too. In fact,
expressing general graphs in relations is trivial. What's not trivial is
expressing a general graph in a hierarchy.
Received on Tue Apr 17 2007 - 07:02:05 CDT
