Re: Relationals vs. Objects Databases I

In article <01bd30c2$2cb91e40$7261b693_at_HP5079Q>, dogmat <macdonaldrj_at_bv.com> writes
>> Jeremy Rickard wrote in message ...
>> >>No. All OO programming languages are computationally complete
>> >>(only requires arrays, if and goto (loops)). OO centainly 'fits' the
 way
>> >>people think about problems.
>> >
>> >In practice, SQL seems complete enough in my experience.
>
>Not in mine. I have an application with a simple constraint: x1+x2=x3.
>But this is repeated many times over recursively, i.e., x3+x4=x5, x5+x6=x7,
>et al.
>SQL cannot handle this. I have to know a priori how many levels of
>recursion I need
>(and I cannot). Even if I did know how many levels I need, I would have to
>do a recursive
>SQL that many times even for the cases which only go one level deep. So I
>am forced
>to handle the constraints in the programming language. Now that I am forced
>to program
>in something besides SQL, why should I have to tolerate the impedance
>mismatch that adds
>30-50% to my work effort?

I'll just "unretire" myself from this thread momentarily...

I did point out that recursion was a good "line of attack" if you want to criticize the limitations of SQL; but also that some database vendors support recursive queries, albeit with syntax variations.

I can't say for certain whether recursive SQL could solve your problem, since I don't understand the problem fully. If you email me with more details (bearing in mind that my physics is rusty), then I'll see what is and is not possible - purely as an academic exercise since you've already started on your own solution.

-- 
Jeremy Rickard

(To email, change "SPAM" to "jbdr" in address.)