Re: [HACKERS] Wishlist for next version: group by clause

I didn't realize PG could not do

    group by [function on column]

Ouch!

I *think* all the "real" RDBMS can do this.  If Oracle and
Sybase both support it, that makes it more or less a de facto
standard :-) I'm sure we use this syntax in several places in
our apps -- certainly in our time-series analysis package.
Implication is that 6.3 is still not functional enough to
replace an existing commercial SQL server such as Oracle or
Sybase for production apps, without expensive manual proofing
and rewriting of embedded SQL statements.

Does anyone know whether this group by syntax is ANSI SQL92?

            ---------------

There must be many sites in the same boat with mine:  running
an outmoded version of one of the Big Guys' engines, unwilling
to pay the outrageous support and upgrade fees required to get
current, wanting full Linux support, yet unable to switch to
PG because of small gotchas like this one.  It's a small
gotcha if you are writing a brand new app, but it's a large
gotcha if you have to comb through thousands of embedded SQL
statements in hundreds of production apps and manually fix it
in each instance.

Is there a list of the "PG is different" gotchas like this,
and is their elimination being given a high priority? I think
"plug-n-play" replacement of existing servers with PG is a
good practical goal -- so long as the app writers have wisely
avoided vendor-specific syntax in their SQL, of course :-)
I think conversions of this sort would be good publicity
for PG, and I would be willing to write up a public report
on mine if and when PG evolves to the point where I can
do it!

What do you all think about the PR value of new PG-driven
apps vs conversion of existing production apps?

de

.............................................................................
:De Clarke, Software Engineer                     UCO/Lick Observatory, UCSC:
:Mail: de@ucolick.org | "There is no problem in computer science that cannot:
:Web: www.ucolick.org |  be solved by another level of indirection"  --J.O. :