Re: Draft response to GK-1

Related to the issue of ORDER in GROUP_CONCAT we had a mail thread starting at

 http://lists.w3.org/Archives/Public/public-rdf-dawg/2010OctDec/0041.html

and ending at

  http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2010Oct/0001.html

... this was then discussed in the TC on 2010-10-26:

http://www.w3.org/2009/sparql/meeting/2010-10-26#line0100

there we had a strawpoll on whether or not we want an order feature in GROUP_CONCAT...

 Options for group_concat: 1) no order_by 2) simple order_by 3) full ordering by expressions (e.g. order by second letter of a word, etc.)

where a clear majority voted for 1), mainly because of concerns that anything else would not be doable with the remaining time/resources.

As far  as I understand, what is now being discussed is 3), correct? Back in october, no one voted for 3). 
Unless someone thinks we have substantial new information, I think we should stick with this position.

BTW: I just see that, on IRC, in that TC, Steve suggested to make a postponed issue out of it: http://www.w3.org/2009/sparql/meeting/2010-10-26#line0133  but so far we haven't added it as an issue.

best,
Axel

On 14 Feb 2011, at 10:33, Steve Harris wrote:

> On 2011-02-14, at 09:34, Andy Seaborne wrote:
> >
> > On 14/02/11 09:17, Axel Polleres wrote:
> >> Hi steve,
> >>
> >> On 8 Feb 2011, at 16:31, Steve Harris wrote:
> >>
> > ...
> >>
> >>> He raises a reasonable point about the interaction of ORDER BY,
> >>> http://www.w3.org/2009/sparql/docs/query-1.1/rq25.xml#convertSolMod
> >>> could be construed as meaning that ORDER is preserved in aggregate
> >>> operations, though the algebra does say it's operations on
> >>> multisets.
> >>
> >> Is this something that needs to be discussed? Could an alternative
> >> behaviour to the current one be achieved with reasonable effort and
> >> the time/resources we still have?
> >
> > This relates to Jeen's point about GROUP_CONCAT and ORDER BY
> >
> > http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2011Feb/0004.html
> >
> > Maybe we should add "; ORDER BY" to GROUP_CONCAT.
> 
> ; ORDER BY (and ; LIMIT) would be very useful, but on the other hand, I'm thinking that aggregates are quite complex as it is.
> 
> - Steve
> 
> --
> Steve Harris, CTO, Garlik Limited
> 1-3 Halford Road, Richmond, TW10 6AW, UK
> +44 20 8439 8203  http://www.garlik.com/
> Registered in England and Wales 535 7233 VAT # 849 0517 11
> Registered office: Thames House, Portsmouth Road, Esher, Surrey, KT10 9AD
> 
> 

Received on Monday, 14 February 2011 11:01:08 UTC