Re: Extending my question. Was: The relational model and relational algebra - why did SQL become the industry standard?

From: Paul Vernon <paul.vernon_at_ukk.ibmm.comm>
Date: Sat, 15 Feb 2003 19:26:36 -0000
Message-ID: <b2m4he$eli$>

"Lauri Pietarinen" <> wrote in message [snip]
> I am just trying to get a hold on this matter, what is it exactly that
> Date get's wrong or missunderstands.
> Are they trying to achieve different goals? Maybe my head is a bit
> thick but I am not convinced.

For what it is worth, from this discussion I am still convinced that bags are a bad model for users, because to quote from Date's second article,

"[with bags] the range of possible results that can be obtained from any given database is
vastly expanded, though not usefully so" "[with bags] the range of logically distinct query expressions that can be formulated is
vastly expanded, too, though again not usefully so"

In otherwords, Occams razor comes into play.

The extra complexity of a bag algebra (even if it only 'a bit' more complex) is of no use for users, at least in mine and Chris Date's opinion. Reminds me of transactions actually, again possibilities get expanded but not usefully so (except maybe for those that teach ;-) ;-) ).

As an internal bag algebra might well help implementations provide better performance, I can see arguments for exposing such an algebra to users to possibly make the implementators life easier. However, that is the tail wagging the dog. I can only see costs in a bag algebra for users.

Paul Vernon
Business Intelligence, IBM Global Services Received on Sat Feb 15 2003 - 20:26:36 CET

Original text of this message