W3C

- DRAFT -

XProc telcon

Meeting 90, 1 Nov 2007

Agenda

See also: IRC log

Attendees

Present
Henry, Alessandro, Andrew, Richard, Alex, Michael
Regrets
Norm, Paul, Rui
Chair
Henry S. Thompson (pro tem)
Scribe
Henry S. Thompson

Contents


Agenda

http://www.w3.org/XML/XProc/2007/11/01-agenda

Accepted as distributed

Minutes of last meeting

http://www.w3.org/XML/XProc/2007/10/25-minutes.html

Accepted as distributed

Next meetings

F2F in Cambridge MA next Thur and Friday: Norm, Henry, Paul, Alex, Michael (in part)

HST: We will try to announce some summary of discussion and decision making times, for those who are dialing in

Next telcon: 15 November, usual timing

Regrets: Michael

Review of Action Items

No changes to published list

Comment 29: Determining whether a pipeline has a (defaulted) output

http://www.w3.org/XML/XProc/2007/09/lastcall/comments#C029

RT: At the end of the last meeting I was leaning towards requiring declarations for p:pipeline
... They are short, don't require any bindings
... I think we all rejected the extreme interpretation of the status quo which would require arbitrary recursive analysis

HST: RT and I discussed this after the call last week, and developed a possible alternative: use syntax to distinguish simple pipelines which would get defaulted input/output declarations from complex ones which would not. The distinction might be made with two element types, or an attribute, or . . .
... In the absence of email preparation, let's move on

AV: Please do send an example, but yes, let's move on

<scribe> ACTION: HST to send an example of a 'new syntax' resolution to issue 29 to the list [recorded in http://www.w3.org/2007/11/01-xproc-minutes.html#action01]

Comment 18: Scope of step types

http://www.w3.org/XML/XProc/2007/09/lastcall/comments#C018

RT: Don't we have approximate consensus on this, action A-87-03 refers

Comment 13: Saxonica comments on sections 5-7

http://www.w3.org/XML/XProc/2007/09/lastcall/comments#C013

HST: First, let's look at comment 3, section 5.7.3 option 2

http://www.w3.org/XML/XProc/docs/langspec.html#opt-param-bindings

RT: I think MK has misunderstood it
... The prefixes whose bindings are in question are those in the result of a 'select' XPath, not in the 'select' XPath itself

HST: The problem arises because there is no example to hand

AM: I think we have to clarify with an example and with better text, what the purpose of 'default namespace bindings' actually is at this point

RT: There is an example further down

HST: We need a simpler example earlier

RT: Aha, we should be looking at the first numbered list -- OK, yes, I see the problem

HST: Enough here to guide the editor, let's leave it with him

<scribe> ACTION: NW to rewrite 5.7.3 by adding a simple 'select=' example alongside the 'match=' one at the beginning, and trying to clarify what the default namespace binding is for early on [recorded in http://www.w3.org/2007/11/01-xproc-minutes.html#action02]

HST: Moving on to comment 4, a clarification

http://www.w3.org/XML/XProc/docs/langspec.html#p.document

AM: p:load is a step, p:document is not
... p:document leaves it implementation-defined whether we validate or not

HST: No, it says you must not validate

AM: I'm surprised that is there. . .

RT: What does it mean to say 'must not validate' ? 'Must not fail for validity errors' I could understand

HST: I think we're looking at the result of Norm trying to respond to MK's comment here. . .

RT: I think this needs to change to clarify that p:document doesn't fail because of a validity error
... I think we should follow XSLT here and require that the external subset be read
... so that all entities are expanded

HST: Do we need to be more explicit about any other processor-dependent options?

AM: I would be sorry to disallow the possibility of a secure environment in which all input of any kind to a pipeline had to be valid

RT: I guess we need to discuss this at the f2f
... I think AM's point should be an 'at user option' feature. . .

HST: OK, done until the f2f
... I think wrt MK's second point, NW's change in the 3 Oct. draft is sufficient

HST: Moving on to comment 6
... This is subsumed by issue http://www.w3.org/XML/XProc/2007/09/lastcall/comments#C070
... Issue 13 Comment 6 and Issue 70 should be responded to jointly

RT: The idea is to do things much more cheaply than would be the case if XSLT or XQuery were used

HST: Yes, determines whether you get an efficient implementation without waiting for an XSLT implementation which detects streamability

RT: I've suggested in the past that we use XSLT stylesheets to provide exemplary implementations of the steps like these
... removing any ambiguity as to how they work

HST: Interesting idea -- volunteers?

<alexmilowski> Richard suggested it...

<alexmilowski> :)

RT: There may be problems in the details

HST: We'll leave that for now, as a start on subsequent discussion of issue 70

RT: Let's not get bogged down in details of individual steps

HST: Moving on to comment 9, on p:load

http://www.w3.org/XML/XProc/docs/langspec.html#c.load

RT: What is meant by 'namespace aware DTD validation'?
... assume it means namespace-aware parsing

HST: I think, by contrast with the reference to Namespace Well-formed for p:document, this means that if validate='true', then we require Namespace Validity

RT: Right, e.g. IDs must be NCNames

<scribe> ACTION: NW to clarify by adding reference to Namespace Validity to the description of p:load with validate='true' [recorded in http://www.w3.org/2007/11/01-xproc-minutes.html#action03]

Summary of Action Items

[NEW] ACTION: HST to send an example of a 'new syntax' resolution to issue 29 to the list [recorded in http://www.w3.org/2007/11/01-xproc-minutes.html#action01]
[NEW] ACTION: NW to clarify by adding reference to Namespace Validity to the description of p:load with validate='true' [recorded in http://www.w3.org/2007/11/01-xproc-minutes.html#action03]
[NEW] ACTION: NW to rewrite 5.7.3 by adding a simple 'select=' example alongside the 'match=' one at the beginning, and trying to clarify what the default namespace binding is for early on [recorded in http://www.w3.org/2007/11/01-xproc-minutes.html#action02]

Minutes formatted by David Booth's scribe.perl version 1.128 (CVS log)
$Date: 2007/11/02 12:36:33 $