TEI Stand-Off Markup WG 2003-10-10 Conference Call Notes
Initials Used for People
- SB Syd Bauman
- JC Jean Carletta
- DD David G. Durand
- JH Jessica Hekman
- NI Nancy Ide
- FV Fabio Vitali
Commenced ~14:37 with DD, SB, FV, JH.
Nomenclature
DD suggests ‘hub document’ for the file which includes others; it's the one you start parsing with. This avoids the words ‘master’ and ‘slave’. Could also just use ‘start of the chain of inclusion’. FV points out the topology need not be a star with this document as the center, which he feels is implied by the word ‘hub’. We agree topology could be any shape that doesn't include a loop. Eventually we agreed on ‘root document’, realizing we will need to be vigilant to avoid abbreviating it ‘root’, which is the root element.
Discussed question of whether or not root document should necessarily have <TEI.2> , <teiHeader> , and <text> . Eventually agreed that such should be strongly recommended, but not a requirement for conformance. Although it isn't entirely clear that TEI conformance applies to files before XInclude processing.
XPointer improvements
XPointer document defines two schemes: xpointer() and point(). The former is an extended XPath syntax — you can place any XPath in there and a few other things: functions (must be 1st), step productions (inside XPath as a single step), and node tests (identify & select elements of a given type in the current node set). Key issue is that xpointer() provides no simple way to get to the internals of a text node (i.e., into PCDATA). Only mechanism is to use string-range() function, which permits access only via string matching (thus you have to know what is in the string to start with) and, per section 4.5.2 2nd sentence by leaving the string operand empty, via numeric ranges.
Point() scheme is an extension of element() scheme, not xpath(), so you have to navigate via node numbers (3rd child node's 4th child kinda stuff).
Large discussion on FV's suggested new scheme which would be XPointer with the addition of 2 new functions: regexp() and word(), where the latter is syntactic sugar for regexp for finding word(s)?. DD suggests it is politically infeasible to expect that any changes to XPath to permit pointing into strings would ever be accepted by W3C, and thus it is unlikely software would support it. Some members had not seen FV's mail on the subject sent immediately pre-call, so no decision reached.
Appendix A:
See also DD's notes.
Wrapped up at ~15:40; next call Fri, 17 Oct 03 at 14:30. JH will be a few minutes late.