Re: feature & performance comparison

That was exactly my original point.

Oracle does not support the principle of Orthogonality in the purest sense. They break the rule every now and then, and then bail out saying, well, we support it through some other feature. It makes me wonder whether the Oracle SQL compiler designers ever knew the concept of orthogonolaity. Microsoft and Syabase are worse. Their implementation of subqueries is a joke.

You learn SQL only once. And use it only one way. You do not wait till the next release to use a feature that should have been delivered using Orthogonality.

Well, Orthogonality is the best kept secret of the SQL implementation. It would save companies such as Microsoft and Oracle tons of money (by not having to support multiple custom features).

Finally, Mark, Blair's BOM query is way too cool compared to the connect by feature. Yes it does descend to all children of all levels. The other two examples in the BOM trio are also very cool.

Cheers.

Mark Townsend wrote:

> in article B7498743.15384%markbtownsend_at_home.com, Mark Townsend at
> markbtownsend_at_home.com wrote on 6/10/01 8:21 PM:
>
> The PDF link works OK, however. This IS cool, it's a nicely orthogonal
> syntax. I checked the Oracle9i doc and it seems that a named subquery cannot
> reference it's own name, so I guess Oracle doesn't do this this way
> (although CONNECT BY does provide equivalent capability - restrictions may
> be different however).
>
> See Serge - I CAN give credit where credit is due :-)
>
> > in article 3B243445.E53F5D19_at_ca.ibm.com, Blair Kenneth Adamache at
> > adamache_at_ca.ibm.com wrote on 6/10/01 8:00 PM:
> >
> >> See appendix M of the SQL Reference, at this page:
> >>
> http://www-4.ibm.com/cgi-bin/db2www/data/db2/udb/winos2unix/support/v7pubs.d
> 2>> w
> >> /en_main
> >
> > Cheers - but, um, we have a problem Houston. Trying to access the HTMl
> > version I get the following error message
> >
> > DTWF019E: Net.Data failed to connect to a backend The Net.Data ODBC language
> > environment could not allocate the environment handle.
> >
> > Planned downtime, right :-)
> >
> >> There are three examples, and it talks about controlling depth. Recursive SQL
> >> has been in DB2 since 1995.
> >>
> >> Mark Townsend wrote:
> >>
> >>> in article B749727F.1536B%markbtownsend_at_home.com, Mark Townsend at
> >>> markbtownsend_at_home.com wrote on 6/10/01 6:52 PM:
> >>>
> >>>>> WITH RPL (PART, SUBPART, QUANTITY) AS
> >>>>> ( SELECT ROOT.PART, ROOT.SUBPART, ROOT.QUANTITY
> >>>>> FROM PARTLIST ROOT
> >>>>> WHERE ROOT.PART = '01'
> >>>>> UNION ALL
> >>>>> SELECT CHILD.PART, CHILD.SUBPART, CHILD.QUANTITY
> >>>>> FROM RPL PARENT, PARTLIST CHILD
> >>>>> WHERE PARENT.SUBPART = CHILD.PART
> >>>>> )
> >>>>> SELECT DISTINCT PART, SUBPART, QUANTITY
> >>>>> FROM RPL
> >>>>> ORDER BY PART, SUBPART, QUANTITY
> >>>
> >>> Hmm - actually, there is more to this than I first saw - presumably this
> >>> iterates down through all levels of child ? How is this transformed ? Can
> >>> you push predicates into the traversal ? This may be way cool.
> >>
> >
Received on Mon Jun 11 2001 - 21:03:55 CDT