XML Processing Model WG

Meeting 136, 05 Feb 2009


See also: IRC log


Norm, Mohamed, Paul, Vojtech, Alex, Henry


Accept this agenda?

-> http://www.w3.org/XML/XProc/2009/02/05-agenda


Accept minutes from the previous meeting?

-> http://www.w3.org/XML/XProc/2009/01/29-minutes


Next meeting: telcon 12 Feb 2009?

No regrets heard.

Review of open action items

ACTION-2009-01-29-01 completed

Mohamed: I didn't see very much in common between our specs. They're mostly using binary offsets.
... There are a few others that go inside ZIP to check for files.

Norm: So you didn't see anything that seemed out of the ordinary?

Mohamed: They're using some new space characters and they're doing a case-insensitive comparison in some places.
... I'll be watching those things.

Norm: It doesn't sound like there's anything we as a WG need to comment on.

Mohamed: I don't think so.

ACTION-2009-01-29-02 continued

031. Redefining standard steps

-> http://www.w3.org/XML/XProc/2008/11/cr-comments/#C031

Norm: This is about attempts to redefine steps in the p: namespace.

Vojtech: I thought err:XS0036 would cover it.

Norm: Yes, but I think we also want the error to cover the case of declaring p:foo

Vojtech: Ok, then we don't have an error for that.

Norm: I think we should just create a new error for this, any objections?

None heard.

059. “Document sequence”

-> http://www.w3.org/XML/XProc/2008/11/cr-comments/#C059

This is about the term "document sequence". Should we define it?

Norm: I've never thought we meant more or less than what the English language words mean.

Vojtech: If we have a formal definition of sequence, then we'd need to define other things.
... The word sequence is almost the definition.

Alex: Since the term sequence in XQuery/XPath 2 has a particular concept, perhaps we need a definition is looser.
... XPath 2 has a bunch of loaded semantics that we don't want to inherit.

Norm: True, you never get an XPath 2.0 "sequence" from our "sequence of documents".

Some discussion about the fact that you can't actually access our sequences as a XPath 2.0 sequence.

Norm: Does anyone think we need to try to tie this down?

Henry: I think it's likely to be harder to get right than to say nothing about it. It's very hard to get right.

Mohamed: Especially if you want to have room to do parallel optimization.

Vojtech: In XQuery (and XSLT) we do say that the sequence becomes the default collection.

Norm: But that's a collection not a sequence.
... I propose that we close this with no action.


077. Required serialization methods

-> http://www.w3.org/XML/XProc/2008/11/cr-comments/#C077

Norm: The request here is that we support the 'text' serialization method.

Alex: I think this is a quality-of-implementation issue. There's nothing that prevents implementations from doing more.

Mohamed: I agree. I think XML serialization is the bare minimum. Getting text right is actually quite hard.

Norm: Anyone want to argue for including more than XML as mandatory?

None heard.

Norm: I propose that we decline and leave other methods as implementation-defined.


080. Content model of c:multipart

-> http://www.w3.org/XML/XProc/2008/11/cr-comments/#C080

Norm attempts to summarize.

Norm: I just don't know if useful headers can be associated with a body.

Vojtech: I think the body can have arbitrary headers. That's what the text of the step says.

Alex: I think you're right.

Norm: That makes me want to put a wrapper around each collection of (header*,body), but maybe it's too late for that.

Alex: There's more work that you have to do to encode the pieces.

Norm: I just wish we had c:part wrappers around them, but I don't think we can do that now.

Vojtech: We don't handle nested multipart bodies either.

Norm: So I guess the proposal is to fix the grammar so that it allows a mixture of headers and bodies.


-> http://www.w3.org/XML/XProc/2008/11/cr-comments/#C081

083. http-request: if detail is true, XML is not parsed?

Norm attempts to summarize.

Norm: I think the intent was to flip the 2nd and 3rd paras of and make the "translation of the text into a Unicode character sequence" only apply to non-XML media types.

Mohamed: I think that was the intent.

Alex: The intersection between these two paragraphs is not zero.
... If you have an XML media type or a text type, then you can make a sequence of characters. If it's an XML media type, then you should parse it.

Norm: So this is intended to be two-part process.

Alex: Maybe the right thing to do here is leave most of that first sentence and just at the end say that you're supposed to construct a sequence of characters.
... Make the part about making a c:body a separate part.

Norm: What I hear is that the intent is to get Unicode characters first, then parse them if it's an XML media type.
... Anyone disagree?

None heard.

Norm: I propose we get our editor to fix this.


084. Handling of ‘non-XProc' pipeline sources

-> http://www.w3.org/XML/XProc/2008/11/cr-comments/#C084

Norm: This is about what should happen if you hand a random XML document to a processor.

Vojtech: I think it should be separate static error.

Norm: My concern is: should we mandate the behavior or say that it's implementation defined.

Mohamed: Use XPointer if you want to embed pipelines.

Vojtech: Importing would be a problem too.

Norm: I think the proposal is make it a new static error if the pipeline document doesn't have a root of p:pipeline, p:declare-step, or p:library.

Mohamed: yep.


087. Ordering of steps in a subpipeline

-> http://www.w3.org/XML/XProc/2008/11/cr-comments/#C087

Norm: I think some examples would be useful, perhaps in a non-normative appendix.
... Does anyone else think that would be valuable?

Vojtech: I have problems with understanding all the details.

Mohamed: If you have explicit connections, why do you have to reorder them?

Norm: Only so that you can get the execution order right.

Mohamed: I think reorder and execution order are different things.
... I'm trying to find out why we're trying to make the process harder than necessary.

Vojtech: So is there a use case for writing the steps out of order?

Mohamed: Only to make authoring easier.

Some discussion of how to achieve the order.

Norm: Make the implicit connections explicit, then look for cycles. If you find a cycle, the author loses. If you don't, then pick one of the partial orders and you're good to go.

Vojtech: Ok, I'm satisfied for now.

Norm: In that case, I think we should just close this without action.


Any other business?

None heard.


Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.133 (CVS log)
$Date: 2009/02/05 17:10:43 $