Re: Algorithm for converting graph patterns

Lee Feigenbaum wrote:
> Hi DAWG,
> 
> I wanted to point your attention to 
> http://www.w3.org/2001/sw/DataAccess/rq23/rq25.html#convertGraphPattern 
> where, in response to points raised from commenters, Andy has drafted 
> some possible text (search for "Reworked") to clarify the algorithm in 
> light of the comments we've received.
> 
> The main snag is that it's a non-trivial change to make at the 11th hour 
> of our process. As a member of the WG, I have reviewed the changes and 
> aside from a couple of changes[1], I believe that it both clarifies the 
> text greatly and is true to the existing semantics (as given by the 
> existing spec text and by the test suite).
> 
> As Chair, though, I'm uncomfortable accepting this last minute change 
> without more eyes on it. If you are interested, please do take a look. 
> We'll talk about it on the teleconference and decide what the best path 
> forward is.
> 
> thanks,
> Lee
> 
> 
> [1] Andy, I believe that the following has two fixes needed:
> 
>    If A is of the form OPTIONAL(Filter(F2, A2))
>      G := LeftJoin(G, A2, F)
> 
> should be:
> 
>    If A is of the form Filter(F2, A2)
>      G := LeftJoin(G, A2, F2)
> 

Done (actually, call the set of filters FS and a single filter expression F).

The text for clarification of re-use of IRIs in dataset clauses is also the 
doc now.

 Andy

> 
> 

-- 
Hewlett-Packard Limited
Registered Office: Cain Road, Bracknell, Berks RG12 1HN
Registered No: 690597 England

Received on Tuesday, 16 October 2007 08:31:45 UTC