See also: IRC log
http://www.w3.org/XML/XProc/2008/05/15-agenda.html
HT: Add to the agenda PSVI
handling, in first place under Technical
... Agenda accepted as circulated and amended
... Review minutes from last week: http://www.w3.org/XML/XProc/2008/05/08-minutes.html
... Approved without comment
ACTION: Norm Walsh to fix minutes of 1 May per http://www.w3.org/XML/XProc/2008/05/08-minutes.html [recorded in http://www.w3.org/2008/05/15-xproc-minutes.html#action01]
HT: Meeting next week: Regrets from Moz, Norm, HST
NW: Call next week is cancelled, next meeting is 29 May
http://lists.w3.org/Archives/Public/public-xml-processing-model-wg/2008May/0095.html
http://www.w3.org/XML/XProc/docs/langspec.html
Discussion of the PSVI preservation proposal in the above message
Detailed discussion of the viewport [validity] case
RT: Points out that since the document element was not the validation root, the fact that IDREF validity isn't recorded there is not a bug
<scribe> ACTION: HST to go think what we do with the [validation root] property inside viewport [recorded in http://www.w3.org/2008/05/15-xproc-minutes.html#action02]
MZ: Wrt outputs of compound
steps, what about a choose
with PSVI on one branch and not
another?
HT: Not ruled out, author should
know better
... XPath2 has to respect PSVI if possible -- OK
... Point e) no unspecced PSVI preservation
NW: What if RelaxNG validation does type assignment? Must I lose them?
AM: Why not make an exception for identity?
HST: Because it's an exception, and will in practice rarely be used. . .
NW: Exceptions are messy
RT: What if there is no schema for it, i.e. if XSLT2 does some type assignment?
VT: Also when you don't have a primary output. . .
NW: I'm coming around. . .
MZ: What about split-sequence?
VT: Identity can be very useful to break up the order of steps
HT: Consensus is that we have a list of steps which MUST preserve the PSVI, starting with split-sequence and identities, i.e. steps which cannot change their input
NW: What about a select
expression on the input?
... Is it the same as viewport?
HT: Yes, so same rule applies
Voytech: Note that in the sequence case, even as input to identity, some may have PSVI and some not
NW: Yes, I'll add a note that points that out
HT: Add validate-with-xml-schema and validate-with-relaxng to the list in my point 4, impl. defn. what PSVI properties they output
NW: What about a more fine-grained psvi-available() function?
HT: I don't need it
MZ: I had a use case -- see email http://lists.w3.org/Archives/Public/public-xml-processing-model-wg/2008May/0098.html
HT: [fails to explain why he thought it wasn't needed]
MZ/VT/HT: What about dynamic access to the global system properties such as xpath-version, psvi-required (and ignore-inline-prefixes)?
NW: Please have a look at the simple suggestions for updating individual steps in today's agenda, and object or not via email
HT: Adjourned