XML Processing Model WG

Meeting 43, 9 Nov 2006


See also: IRC log


Norm, Alex, Paul, Rui, Alessandro, Michael, Richard, Henry, Mohamed, Murray


Accept this agenda?

-> http://www.w3.org/XML/XProc/2006/11/09-agenda.html


Accept minutes from the previous meeting?

-> http://www.w3.org/XML/XProc/2006/11/02-minutes.html


Next meeting: telcon 16 Nov 2006

Michael possible regrets for 16 Nov

Technical agenda

Norm: The question of declare-* or not is the first big issue
... Any strong feelings?

Henry: I liked the idea in the XML representation of XML Schema that we used the same element types for both declaring and referencing elements, attributes and a few other things.
... In general, I have not found that to be a source of confusion, but it is occasionally critizised.
... But I thought it was a good idea to keep them separate in our language.

Richard: I say get rid of them.
... The divison between them doesn't seem to line up neatly enough. Sometimes declare means declare-and-use so there's no advantage in being explicit about it.

Henry: That's not true of parameters.
... It's always the case that a given locus with respect to parameters is either a declaration or a binding, never both.

Norm: In the case where you're assigning defaults, it looks exactly like declare-and-use

Henry: I'm not sure I see it that way, but I understand how you might.
... Is there value in keeping the distinction for parameter but not input/output?

Michael: Can someone type an example?

-> http://www.w3.org/XML/XProc/docs/alternate/#ex.p.for-each

Norm: The input element both declares the existence of a port named "chap" and binds it to the given URI.

We need more examples.

<MSM> when you speak of a port named "chapter", do you mean a port named "chap" ?

Henry: That's not quite right, viewport does more than that. But we tried hard to fix that in Ontario and didn't find a better proposal.

<MSM> I don't see anything that looks like it's declaring a port named "chapter"

Yes, I meant to say "chap"

Micheal: It's not really a reference, it's an initializer.

Richard: You're right. It's assignment.

Consider: http://www.w3.org/XML/XProc/docs/alternate/#dt-input-here

<ht> HST maintains that these things are a blend of input (source/href attrs) and declare-output (name attr)

Micheal: It seems to me that this mixture could lead an outside observer to believe that we don't have our fundamental notions well understood.

Henry: That's where we were in Ontario where we did finally get the semantics clear and agreed there were two things going on plus some special meanings, but that syntactically, it was obnoxious to have different bits of syntax to bear the different notions.

Micheal: I'm not understanding the distinctions.

Richard: The distinctions are in the attribute names and not in the element names.
... It seems really strange to distinguish three cases with two different elements.

<ht> HST agrees with richard -- name= establishes something input ports can bind to, source/href/etc. bind to a document source

Richard: What we have now seems neither rigerous nor convenient.

Norm: I've heard some support for removing the declare-* forms, some "concurs", and no opposition.

<MSM> [My instinct is that when things are semantically distinct they should have different GIs.

<MSM> If that means we end up with a cumbersome syntax, then it sometimes means we aren't defining the right abtractions.

Norm: We shall go with a draft that does not have the declare-* forms for publication on 17 November?

<MSM> And sometimes it doesn't. I am not sure which class of case this is.]


Norm: Any other issues that people feel must bee resolved before publication?

None heard.

Norm: Alex published a list of components this morning but it's unlikely that we'll have time to review them. Does anyone object to leaving them out of the 17 Nov draft?

Mohamed: I think we should put some of them in.

Murray: Would it be reasonable to produce a companion note that we can work on in parallel?

Norm: No, I don't think that's practical.

Richard: Looking through Alex's list, it appears that some are obviously good and some that need more discussion.
... I think it would be good to put the most uncontroversial ones into the draft.

Henry: I nominate 1.1 through 1.6 with one change.
... The change being that the XSLT component should be clear about what version of XSLT.

Norm: What about 1.7?

Henry: No, that raises security issues.

Alex: That's no different from the endpoints of a pipeline.

Henry: Yes it is, they're outside the spec.

Murray: I don't think we need firm agreement before we put something in the draft.
... I think we should include it all if we're going to put it in.

Richard: I'm not sure I agree.
... When we put the WD out, we want to really direct attention to the core of it.
... What about serialize and parse?

Alex: Yes, we've talked about them, but those are probably somewhat controversial.

Murray: There's a middle ground, include an editorial note to say which are firmer and which are softer.

Norm: Proposal: we incorporate 1.1-1.6 in the 17 November draft.

Murray: Point of order: Alex provided the whole list.
... why are we rejecting the whole list?

Henry: I don't have any objection at all.

The scribe considers how to deal with this

Proposal: Include section 1 of Alex's component list in the 17 November draft.


Proposal: Include section 2 of Alex's component list in the 17 November draft.

Henry, Mohamed, Richard, object. Rejected.

Proposal: Include section 3 of Alex's component list in the 17 November draft.

Henry objects.

Richard: It has some value in that it shows quite different kinds of components.

Henry withdraws his objection.


Proposal: Include all of Alex's component list in the 17 November draft.

Henry objects.

Alex: What's wrong with section 2?

Richard: Section 1 has obvious things, section 3 has fairly straightforward things that are somewhat different, but section 2 has some controversial things.


Mohamed: I suggest that we add 2.3.

Norm: The chair would prefer not to address individual cases for 17 Nov

Henry: What about parameters?
... I'm not sure about Jeni's most recent post, but I think it should be possible to refer to parameters from other parameters.
... Is it obvious that we can do that, and how?

Alex: It's not clear, but you do the $-thing.

Norm: I think that it's unclear, but that we did address it.

Henry: I'd like to encourage the editor to say something about that in the 17 Nov draft.

Alex: What about references to parameters declared on the same component?

Henry: I'd like to be able to make $x the default value for $y.

Norm proposes to deliver a new working draft by close-of-business (EST) on Tuesday that will be published on 17 November. Unless someone objects on the 16 Nov call, that draft will be published.


Any other business?



Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2006/11/30 17:14:36 $