Re: Comments on 10.1 "Solution sequences and result forms" (CR 6 Apr 2006)

Fred Zemke wrote:
> 
> 
> 
> 10.1.1 Projection
> The last sentence of the formal definition uses set notation
> for the result of projecting a solution sequence into a new
> solution sequence.  This is not desired, because:
> a) sets are not ordered, but solution sequences are
> b) sets do not permit duplicates, but the intent is that the
> result of a projection might have duplicates.
> 
> This can be corrected by using some notation denoting a sequence.
> Earlier we used (S1, ..., Sn) to denote a sequence, and that
> could be done here, for example,
> ( (project (S1, VS), ... project (Sn, VS) ). 
> Or we can use the mathematical definition of a sequence as a
> function whose domain is the positive integers, in which case
> the sequence is represented { (i, project (Si, SV) ) | i = 1, ..., n }

rq24 (and earlier suggested changes) : "10.1.2 Projection"

"""
For a solution sequence S = ( S0, S1, . . . , Sn)
and a finite set of variables VS,
project(S, VS) = ( project(S0, VS), project(S1, VS), . . . , project(Sn, VS) )
"""

> 
> 
> 10.1.3 ORDER BY

rq24 (and earlier suggested changes) : "10.1.1 ORDER BY"

> The formal definition does not support the following features:
> a) ordering in descending order

The ordering condition C is arbitrary in the definition so gives the
direction. To order in the opposite order, C orders Si before Sj if Sj
"greater than" Si.

> b) ordering by multiple sort keys.

The ordering condition can be based on multiple sort keys.

I changed the first sentence to make that clearer:
"""
The ORDER BY clause takes a solution sequence and applies an ordering
condition based on all the expressions and directions specified in the ORDER
BY clause.
"""

> 
> Fred
> 

	Andy

Received on Monday, 24 July 2006 09:04:41 UTC