See also: IRC log
Skipping the rest of December for seasonal festivities.
No regrets heard.
Henry: I want to behave as much like XQuery as possible.
Alex: I think it's XSLT not XQuery that we're copying.
Vojtech: In the first version of the document there were different rules. Now we have rules more like XQuery wrt curly braces and quotes.
Some discussion about the current rules.
Vojtech: I think we're currently consistent with what XQuery does.
Norm: I think the only question is, if you see "}" in xpath-mode, do you look for another "}" or do you end the expression?
Vojtech: No, I proposed that if you're not in XPath mode and you see "}" then you just treat it literally.
Norm: So "}}" remains "}}"?
Vojtech: No. If you see "}}" you output "}", if you see an unescaped "}", you just recover from the escaping error and output the "}"
Alex: I think the current rules could allow nested expressions in the future.
Vojtech: I'm just observing that you could recover, I'm not pushing for it.
Norm: I think it will be confusing to do error recovery, so I propose not.
Alex: We could do that, and be done with it.
Proposal: Add a new rule to "regular-mode" which states that an unquoted "}" is an error.
Norm: I'll also add a note to the
xpath-mode section to note that "}" doesn't look for "}}", it
... I'll update the spec and republish it in our space, with a plan to make it an official note in January if no one sees any other problems.
Vojtech: Does the minimum profile
require parsing of namespaces?
... Now I think it's clear in the text.
... The other issue is the one that David Lee raised about having a much simpler profile.
Henry: That amounts to subsetting XML.
Norm: I'm very conflicted. I think what David Lee wants makes logical sense, but it's not clear that we have remit to go there.
More discussion about infosets and subsets of XML and the fact that our section 2 says we start with a namespace well-formed document.
Henry observes that if we named a smaller subset, then parsers that could do it only wouldn't be able to handle any well-formed XML which is not the case today.
Proposal: Politely decline on the basis that it wouldn't be XML. It's not illogical, but we can't go there.
Vojtech: What about my question about connecting output ports of compound steps?
Norm: I think 5.11 overstates what we intended.
Vojtech: Our implementation allows it.
Norm: Yeah, I guess it makes sense. I don't think my implementation allows it but that's neither here nor there.
Vojtech: I think the rule in 5.11 is quite convenient.
Norm: And allowing it is an editorial change were forbidding it would be a technical changes, because 5.11 says it's currently allowed.
Proposal: Add an erratum to say that the output port of a compound step can be directly connected to any of the compound step's readable ports.
<scribe> ACTION: Norm to write an erratum to allow he output port of a compound step can be directly connected to any of the compound step's readable ports. [recorded in http://www.w3.org/2010/12/16-xproc-minutes.html#action01]
Alex: Do we produce a new version with the errata merged in?
Norm: We can, but we don't have to.
Henry: It's polite to do it.
Paul: That would be a second edition.
Henry: I think it's fine to wait at least a year.
Norm: So do I.
... Happy holidays and happy new year to one and all!