See also: IRC log
No regrets heard.
A-207-01: completed. there's only one comment
A-209-01: continued, ETA 29 Mar
Norm: We only have one, from cmsmcq.
Norm: Two substantive issues,
standalone and validation.
... What about standalone?
... What are the cases?
... No external decls, it's irrelevant
... External decls, standalone=no, (the default) that's a validity constraint
... External decls, standalone=yes, then what we do is ok.
Henry: Perhaps we should ask
Michael if he made the same mistake that I did, that
standalone=no does not require a processor to read the external
... We should also see if he made it more clear in his previous comments what he wanted.
Norm: Looking at Michael's comments, I think he has misinterpreted the spec.
<scribe> ACTION: Norm to respond to Micheal and see if we can come to the same place. [recorded in http://www.w3.org/2012/03/15-xproc-minutes.html#action01]
Alex: Reading Micheal's original comment, I think there are two interpretations. One is that these are things he thinks the XML Rec says and one is that these are the things he would like *us* to say.
Norm: You think perhaps Micheal
is saying that *we* should enforce this behavior wrt to
... The second issue he raises is validation, which he at least agrees we improved.
... I think Micheal makes a good point that it would be easy to read this spec and missunderstand that validation is forbidden.
... I wonder if we could improve things by mentioning validation in 2.3 and 2.4
... We could say in point 1 "non-validating or non-valdating"
Henry: We could add a note that says "this requirement is satisfied by any conformant validating parser"
Murray returns the discussion to standalone with the observation that the XML spec says there's an algorithm for turning standalone=no into standalone=yes
Some discussion of how "algorithm" is to be interpreted; does a validating parser suffice?
Norm: I'm not sure a validating parser covers the case of an invalid but well-formed document with standalone=no
Henry: I think my problem with the standalone declaration is that it's a significant increase in complexity for an area that's very little used.
Alex: What would standalone help us with?
Norm: I'm not sure what Micheal
meant, so I'm not sure how to answer that.
... In the basic and id profiles, standalone is irrelevant becase we don't read the external decls and it's a validity constraint and a validating parser can't be used to do the first two profiles.
Murray: I think what Michael is
saying is what I've been saying, validatity and standalone=no
are things that would change the result of processing.
... If you have a document that requires validation and/or requires fetching external subsets is going to result in a different document.
... And the truth value of that document changes depending on whether you validate or not.
Henry: That's why we put in the
stuff about invariants, so we could be very clear that what you
get may change.
... Maybe we should make it clear that a validating processor cannot implement 2.1 and 2.2 if the documents have an external subset.
... Validating parsers *must* read the external subset.
<scribe> ACTION: Henry to draft notes for 2.1 and 2.2 saying that a validating parser cannot be used if there's an external subset and to say in 2.3 and 2.4 explicitly that validation could be performed [recorded in http://www.w3.org/2012/03/15-xproc-minutes.html#action02]
Norm: Murray, what do you want to say about a standalone=no document with external decls if it's parsed by a basic or id processor.
Murray: I think I just want to say that you may have lost information.
Norm: I have no objection to adding a note to that effect, I just don't think changing behavior is within our remit.
Murray: I still think there
should be a profile that takes in an XML document which was
composed with a notion that it would be validated.
... so it's truth value would be determined by that profile.
... But the WG doesn't agree with that position, that's fine. I think that Michael feels the same way. I think the amelioriting text will help.
<scribe> ACTION: Norm to draft notes for 2.1 and 2.2 to describe the consequences of information loss for a standalone=no document when it has external declarations. [recorded in http://www.w3.org/2012/03/15-xproc-minutes.html#action03]
Murray: What would be most satisfying to me is that if we had a profile that covered validation and then if we deprecated it in V.next.
Henry: I think we're doing better than that. The external declarations profile gives you the infoset you want, whether or not its validated, and then you can decide independently to validate it.
Norm: Perhaps a note to that effect in 2.3 would be a good idea.
Henry: In section 3, for class Extended perhaps we should say explicitly that may be absent under 2.1 and 2.2.
<scribe> ACTION: Norm or Henry to implement Micheal's editorial changes, raising any issues we see, if any. [recorded in http://www.w3.org/2012/03/15-xproc-minutes.html#action04]
Norm: Murray are you interested in working on the use cases and requirements.
Murray: Yes. I was talking to Alex, and I think we might work on it together.
... Proposed ETA?
... How about 12 April?