Re: SpecGL LC16 and 39: CP5.4 (o.ld 8.4) Discretionary items

Good points regarding the evolution of the checkpoint

I don't disagree with having the checkpoint as long as it is clear and 
people can understand the conformance requirements.  At this point, I 
doln't understand it.  Perhaps I am in the minority here and the rest of 
the WG understands it.

Since we are getting ready to publish SpecGL soon, I propose the following:
Get input from the rest of the WG during the telecon as to the clarity of 
the checkpoint and these options:
1. Leave it as proposed by David.  If there are problems with it, we can 
handle them during CR
2. Continue to draft the CP  (volunteers?)
3. Delete it and handle it in the next SpecGL version.

I have already recommended #3, but the other options would be acceptable.

I apologize for not addressing these LC issues and your proposal earlier - 
I had not anticipated the difficulty I would have with it.

--regards
Lynne

At 11:11 PM 9/5/2003, david_marston@us.ibm.com wrote:

>Lynne writes:
> >1. [making the CP apply to consolidation of discretionary items]
> >is not the original intent of the checkpoint (as I was the original
> >author). The original intent was that I would get the same results,
> >under the same conditions - e.g., if I choose an option, like 'large
> >fonts', and view a page, then if I view another page or that same
> >page (after refresh), I would get the same 'large fonts'.
>
>This CP definitely has a long, tortured history. My recent rewrite
>was based on the last published version of 8.4, which was basically
>Dom's wording trying to express the consolidation thought. He said
>"factorize" but meant the same thing. The rewrite that you worked
>from was the result of further interchange among me, Dom, and
>Lofton.
>
>I think we also need to say something about your interpretation,
>that whatever choice is made by the implementation should be
>deterministic. That will need an escape clause for the upcoming
>full-text-search enhancement to XQuery, but is a good idea in
>general. It's especially useful when an implementation is
>capable of more than one behavior and offers the user a choice.
>
>Nevertheless, the pressure to consolidate discretionary items
>will have a very positive effect on interoperability. I don't know
>why you disagree with applying such pressure. I think we should
>have two checkpoints here.
>..................David Marston

Received on Saturday, 6 September 2003 20:27:51 UTC