Re: Draft Response to ED-1

On 2/12/2010 12:25 PM, Steve Harris wrote:
> On 12 Feb 2010, at 16:56, Andy Seaborne wrote:
>>
>> On 12/02/2010 4:08 PM, Steve Harris wrote:
>>> I believe that it's equivalent to Eric's proposal, which was on the
>>> table for a while, but I admit I didn't dig into it too far.
>>>
>>> Your right that it allows you to do the same thing with less bytes, but
>>> I don't think it offers additional features, does it?
>>
>> I haven't understood all the details but it seems to produce result
>> with multiple groupings per row which could either be hard or not
>> possible in that exact form.
>>
>> Could we point to Eric's design?
>
> I spent some time trying to track it down, searching IRC logs and the
> list archive, but couldn't find anything concrete, just oblique mentions.
>
> I'm fairly sure I remember the syntax being discussed on the list, but
> I'd had no luck finding it.
>
> IIRC it used the AGGREGATE keyword. It's possible that Eric showed it on
> a laptop at a F2F, and it was never sent to the list.

I know that it was discussed in person at times, and don't remember if 
it was discussed on list. Eric advocated a design that involved 
aggregates inline within a query. I don't know whether or not it was the 
same / similar to the commenter's suggested design.

When the working group did discuss aggregates, there was strong 
consensus on the SQL-like syntax and semantics.

If there are WG members who would like to re-consider this decision in 
light of the commenter's message, I'd like to ask that you volunteer to 
present the summary of the alternative design at a teleconference in the 
near future. If no one wishes to do this, I'd suggest that we conclude 
that there is consensus on the design as is, based on experience and 
familiarity with SQL aggregates.

Lee

Received on Friday, 12 February 2010 17:45:42 UTC