Re: implementability + HP last call position

>I heard your ruling as chair that discussion of CR exit criteria is 
>not in order yet.

I don't think I made such a ruling - I made a suggestion that barring 
critical issues w/respect to implementability we should hold off

>I think that means that HP should probably abstain on the last call 
>votes, but indicate a concern that the recommendation we are 
>drafting is not currently implementable as is. This concern is best 
>addressed with appropriate CR exit criteria and actual 
>implementations of one sort or another.
>I suspect there was a communication failure over the specific 
>proposals I was making: you seemed to see them as unreasonable, and 
>they were not intended as such, but we can have another go at that 
>when such discussion is in order.

I thought them overly harsh, but notice I carefulyl refrained from 
any detail - I just wanted to move towards LC without derailing 
unless something serious arose

>You also indicated that you believed we would have DL reasoners as 
>part of our implementation report. I think HP's objection would 
>substantially be met by one or two DL reasoners which could in some 
>sense be called complete (e.g. the implementors believe them to be, 
>and they pass all the DL test cases) and practical (in the Horrocks' 
>sense of the word).

I agree, I also think it possible for someone to build a Lite 
implementation from scratch in a short order of time, but it would be 
a tableaux reasoner, and I am very aware of the issue that if this is 
the only way to do it, is Lite enough different from DL to be worth 
separating - and that I don't know how to address in CR/PR criteria

>So I hope that an abstention at this stage, combined with a last 
>call comment by the HP team as to what we would be expecting to see 
>as part of the implementation report, would be sufficient to permit 
>us to raise a formal objection at say, the transition to PR, without 
>surprising anyone.
>In this light I withdraw my message:
>"Entailments and normativity"
>still believing that it expresses a way of correctly setting user 
>expectations, but prepared to be compelled by the evidence in our 
>implementation report.

OK, that's understandable, and I do hope we will discuss the issues 
you raise in [1], because I do think they are important to getting 
things right before LC.



Professor James Hendler
Director, Semantic Web and Agent Technologies	  301-405-2696
Maryland Information and Network Dynamics Lab.	  301-405-6707 (Fax)
Univ of Maryland, College Park, MD 20742	  240-731-3822 (Cell)

Received on Wednesday, 22 January 2003 10:29:34 UTC