Re: Aggregates: Largest Groups
Date: Fri, 9 Apr 2010 20:19:27 -0700 (PDT)
Message-ID: <d214218c-15c5-4a23-88e2-b1dcbe302259_at_w42g2000yqm.googlegroups.com>
On Apr 10, 2:12 am, paul c <toledobythe..._at_oohay.ac> wrote:
> The context he had in mind was
> the programming usage of databases, especially the dominant big iron
> ones of the day such as IMS, which were hierarchical or network based
> and full of the 'details' you mentioned. I don't think he ever claimed
> his model was aimed at general programming, in fact, I think I've read
> that right to the end he was in favour of data sub-languages.
Exactly correct. He just tried to improve what he saw, and hit upon gold with RM. Then he and perhaps a whole generation of folks stopped there. I on the other hand think the evolution Codd (and many other, different kinds of advocates for high level primitives+structure) started should continue. In *theory* it could overtake procedural programming as a paradigm as well, but practically I don't think it ever will (there are a number of complexity proofs in the way). I think Codd saw most of the same stuff in his day because the reasons to think so aren't exactly spankin-new.
Yet as I just said above, we do have examples of languages which fare better in certain departments, because they're higher level and/or more suited to the task. My own thinking is that at least all of the current languages' best sides could probably be tied together, and that relational algebra might be a big part of that.
-- Sampo Of courseReceived on Sat Apr 10 2010 - 05:19:27 CEST
> I don't speak for him, just going by what he's written of which I think
> the best is so good because it is so concise. Personally I can't
> imagine taking an obvious array application and using sql tables or
> Tutorial D relvars instead (I say 'obvious' just to exclude array
> oriented apps that should have used Codd's 'R-Tables' in the first place).