See also: IRC log
Henry gives regrets for 1 May.
Henry's items are on the agenda; no other progress reported.
Henry: It's a long message, but
basically, I propose to adopt the defacto model for non-REC but
... Follow the approach of the RECs themselves, but a little less formally. There's a URI that will be point to a mutable resource and another, dated, URI, that will never change.
... We should update the spec to point to both the dated and undated URIs.
Norm: Attempting to recall why we
don't point to the pipeline library. At one time, you could
import it if you wanted to; we had some weird rules about what
to do with steps that werent' recognized, etc.
... We changed the rules at some point so that you can't import it and when we did that we removed the link; but I think we should put it back.
... It's still used in the construction of the spec itself.
Jim: I think we should put the link back.
Norm: I think we should do what
... I think Henry can make the dated and undated URIs and I can propose the spec errata.
Henry: Yes, I think that makes sense.
Norm: Any objections to this course of action?
<scribe> ACTION: A-230-01: Henry to create dated and undated versions of the schemas and pipeline library in appropriate locations [recorded in http://www.w3.org/XML/XProc/2013/04/24-minutes#action01]
<scribe> ACTION: A-230-02: Norm to propose errata along the lines described in msg 2013Mar/0028 [recorded in http://www.w3.org/XML/XProc/2013/04/24-minutes#action02]
Jim: No progress. I'm trying to get zip/unzip done.
Jim: No progress there either.
Norm: If the p:xpath-context is omitted, then the default readable port is used, I believe. If it doesn't say that, we should make an erratum to do so.
Henry: Yes, but what about Vojtech's comment?
Norm: I think what I said still applies; the default readable port is used if there's no p:xpath-context and it's explicitly not an error if there's no default readable port. The context is simply undefined.
Norm: (with the standard XPath 1.0 hand wave at what undefined means)
<scribe> ACTION: A-230-03: Norm to propose the erratum to resolve bug 21003 [recorded in http://www.w3.org/XML/XProc/2013/04/24-minutes#action03]
Jim: This appears to just be an editorial error.
Norm: I bet it's a cut-and-paste
error by the editor.
... I think the solution is simply to remove the apparently conditionality of the select expression.
s/If a select expression is
given, it is/The select expression is/
<scribe> ACTION: A-230-04: Norm to propose the erratum to resolve bug 21004 [recorded in http://www.w3.org/XML/XProc/2013/04/24-minutes#action04]
Henry: It happens in
... The note in 'variable bindings' should be amended to be more precise. There's no straightforward referent for the phrase 'that variable' in the note.
<scribe> ACTION: A-230-05: Henry to propose an erratum that fixes the phrase 'that variable' in the note in 'variable bindings' in 18.104.22.168 [recorded in http://www.w3.org/XML/XProc/2013/04/24-minutes#action05]
<ht> "An option that has neither a specified value nor a default value will not appear as an in-scope variable. Consequently, an attempt to refer to that variable will raise an error." should change to "An option that has neither a specified value nor a default value will not appear as an in-scope XPath variable. Consequently, an attempt to refer to an XPath variable whose name is the name of such
<ht> an option will raise an error.
Norm: Looks good to me.
<scribe> ACTION: A-230-05: Norm to put Henry's erratum text in the errata document [recorded in http://www.w3.org/XML/XProc/2013/04/24-minutes#action06]
Henry: There's still a a problem
because I don't think 'in-scope specified options' is well
... No, I take that back. I think we can just use the phrase 'specified options' in clause 5 or perhaps in a new clause 6. Yes, a new clause 6 probably.
Norm: What happens if you say p:namespaces binding=fred and fred is an optional option with no value; is that just a gaping whole in the spec?
Henry: Seems likely.
... But I think the proposed cure is way more expense than is necessary.
... I think it's true if it's a static error if the namespace binding isn't there, but it's also a dynamic error if something you thought was going to be there turns out not to be.
Norm: Yes, I agree that special casing that one issue seems better.
Henry: The reason it's not
statically known is that you might have a declared step with a
declared option with no default which is invoked in two
different places in the pipeline and in one of those places the
value is supplied and in the other it isn't. So without
complete NP-complete flow analysis, you can't tell whether it's
going to be called with or without the option.
... You could imagine we make a rule that says that static analysis has to make the worst case assumption....but that seems unreasonable.
Norm: I'd like to do an experiment or two.
<scribe> ACTION: A-230-07: Norm to setup a test case for the optional option/p:namespaces binding and see what implementations do. [recorded in http://www.w3.org/XML/XProc/2013/04/24-minutes#action07]
Norm: We'll see what happens in the wild and then come back to this one.
Jim: Why aren't p:when and p:otherwise steps?
Norm: Uhhh...because they don't inherit from their preceding siblings etc.
<ht> because they can't appear outside of p:choose
<ht> Right, "substitutions for the body" -- think of it that way
Henry: I'd be tempted to actually think about re-casting the whole section or at least adding a note that says something along those lines. i.e., processors should behave as if the p:choose was the p:choose was replaced by a step consisting of the chosen branch.
Norm: It's ugly; there's static and dynamic behavior.
Henry: Yes, but that's about picking which one is chosen. The fact that you don't know until runtime which one is going to be chosen, but you have to know the static features of the step, you get a whole bunch of constraints that follow naturally.
Norm: Henry, would you be willing to review the section and see how much violence you think would be inflicted if we attempted to recast it along the "substitutions for the body" idea?
<ht> yes, happy to give that a go
<scribe> ACTION: A-230-05: Henry to examine our discussion of p:when etc. and see if it can be easily recast along the lines of 'substitutions for the body' without the confusion about subpipelines. [recorded in http://www.w3.org/XML/XProc/2013/04/24-minutes#action08]
Jim: When might we be able to have another f2f?
I'm going to Balisage and TPAC.
scribe: We could attempt to have a f2f of our own.
<jf_2013> 15-26 July I am on holiday
Norm: There's been some discussion of a meeting in July
<jf_2013> I can do any other time in July
Norm: In Edinburgh. Not specifically XProc, but...
Henry: If Jim and Vojtech could come to Edinburgh or London...
<jf_2013> London better, but I hvae friends in Edinborough
<ht> Week of the 8th would be good
<scribe> ACTION: A-230-06 Norm to send email about possibly meeting in London/Edinburgh the week of 8 July [recorded in http://www.w3.org/XML/XProc/2013/04/24-minutes#action09]
<ht> Sounds like 10--12 July would suit
Jim: I'm giving a talk on XProc at XML London.
Norm: I'll consider getting to London in mid-June