Re: rq23 v1.409

On Fri, Jul 01, 2005 at 09:31:57AM +0100, Seaborne, Andy wrote:
> 
> I'd still like to hear the alternative design that covers the result forms 
> in the match-modify-result processing model (or a different processing 
> model).  For ASK, the solutions modifiers only affect the empty/non-empty 
> test of the solution sequence in trivial ways so don't add anything in 
> practice.  But the symmetry of the same process everywhere seems valuable.

I would drop all the SMs for DESCRIBE. 

While I don't think the SMs applied to CONSTRUCT are really worth it, I do
understand the design better now and don't have an alternative one that's
practical. I'd certainly vote yes to drop SMs applied to CONSTRUCT (in fact,
I'd probably vote yes to drop CONSTRUCT, today), but, as I said, all of
that's moot now. (What I wanted from CONSTRUCT was, say, the ability to
build RSS 1.0 channels from queries on the fly, and the template mechanism
of CONSTRUCT doesn't seem flexible enough to do that. As a completely
different issue, I'd like to see that added to our issues list and
immediately postponed.)

Kendall

Received on Friday, 1 July 2005 11:45:25 UTC