See also: IRC log
sebastian: one additional topic: what does an implementation mean?
yves: that is handled in the conformance section, basically it is "use selection" for the data category
...
we handle that during the "implementation" topic
yves: three proposals:
... 1)version attribute at the root
... 2) same as 1), but if the attribute is
absent, it is "1.0"
... 3) version attribute at the root, or if there is a <rules> element, version attribute at <rules>
Sebastian: two issues: default for versioning, and position of versioning
... how about XInclude? Where do
you process the version?
... do you process this before or after inclusion?
felix: we process what is in XML 1.0, which includes entities, but not DITA or XInclude etc.
sebastian: we should articulate that in the draft
<scribe> ACTION: editors to write a paragraph about relation of inclusion of external information (XInclude, DITA inclusion)versus ITS processing [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action01]
sebastian: versioning again: implimentation of version attributes if they don't have a fixed root
yves: for example DITA has no fixed root
... summary of discussion: version attribute should be at the root
element
example: <x its:version="1.0"><its:rules...</x>
<x> is a root element
<its:rules version="1.0"
<its:rules> is root element
felix: define the version as a global version attribute
sebastian: you would need to modify the whole schema to do that
yves: no consensus yet, we handle the topic at the end of the meeting or later again
yves: felix proposed to point to the W3C ruby TR, and for directionality XHTML 2.0
christian: what will happen if an editor encountors ruby markup?
yves: it depends if the editor wants to complie with the markup part or the processing part
... if the
latter is the case, it has to be compliant with external specs
felix: I don't propose s.t. new, but to see if you are fine with the current spec
yves: we don't have a processor for the global rules of ruby and ITS
... we might have to drop at some point
the selection mechanism and replace it with real mapping
<scribe> ACTION: editors to make conformance of ruby and directionality clearer (esp. with regard to processing of global rules) [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action02]
<scribe> ACTION: editors to repair http://www.w3.org/1999/xlinkf to give the right namespace [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action03]
<scribe> ACTION: editors to add markup to "alt" [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action04]
christian: description on how to associate the host markup with ITS
yves: question if there is more information in ITS or in the host vocabulary with respect to a data category
... if it is the same amount of information, it is fine to associate both, if it is different, it is maby not possible
... this is a general
problem of ITS. The specific one for term: what is a term?
... there is no agreement on what a term is
<scribe> ACTION: editors to repair numbering of conformance clauses in sec. http://www.w3.org/TR/its/#conformance-product-processing-expectations [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action05]
yves: we need examples for terms in sec http://www.w3.org/TR/its/#terms
<scribe> ACTION: all to write examples of what a term is, and what it is used for [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action06]
yves: dt element in XHTML is an index marker.
sebastian: TEI has a term element.
... we need a better example for example 22
<scribe> ACTION: all to make a better example for example 22 [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action07]
<rahtz> TEI: <desc>contains a single-word, multi-word, or symbolic designation
<rahtz> which is regarded as a technical term.</desc>
<rahtz> TEI example: A computational device that infers structure from grammatical
<rahtz> strings of words is known as a <term>parser</term>, and much
<rahtz> of the history of NLP over the last 20 years has been occupied
<rahtz> with the design of parsers
christian: what if the host vocabulary has a different data type?
<rahtz> its:termRef="http://en.wikipedia.org/wiki/Parser"
christian: e.g. anyURI versus NMTOKEN
felix: that would not work
... is that only a problem for anyURI, that is all pointer attributes, ref
attributes?
christian: how about the pointer attributes? Do they select all content, e.g. including child elements or not?
sebastian: its:termRef="http://example.com/termdatabase/#x142539"/> : what more information would you need here?
christian: people want to know that this reference is e.g. a password secured java data base
sebastian: the storage mechanism is hard to grap here
yves: it might be different with termRefPointer, e.g. you want to be sure that it is an URI
<scribe> ACTION: all to provide an example for termRefPointer for the draft [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action08]
<rahtz> "To point a link to a term reference in the current document, a termRefPointer can be used"
<rahtz> "To point to a link to term reference in the current documentation, a termRefPointer can be used"
<scribe> ACTION: editors to implement Sebastians change for termRefPointer definition [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action09]
<rahtz> "To point to a link to a term reference in the current document, a termRefPointer can be used"
sebastian: example: dt with href attribute
<rahtz> example: We may define <term xml:id="TDPV" rend="sc">discoursal point of view</term>
<rahtz> as <gloss target="#TDPV">the relationship, expressed through discourse
<rahtz> structure, between the implied author or some other addresser,
<rahtz> and the fiction.</gloss>
<rahtz> We may define <term target="#TDPV">discoursal point of view</term>
<rahtz> as <gloss xml:id="TDPV">the relationship, expressed through discourse
<rahtz> structure, between the implied author or some other addresser,
<rahtz> and the fiction.</gloss>
summary of associating discussion: If you have semantics / processing expectations in the host vocabulary
scribe: which are an addition to what ITS specifies, the assocation can be done
christian: I am fine with what is informally described at http://www.w3.org/TR/its/#d3e180
yves: "Associating existing markup with ITS data categories can be only done if the processing expecations are
the same
... or if the processing expecations of the host markup cover at least the same as ITS"
<scribe> ACTION: editor change "Mapping of ITS Data Categories to Existing Markup" to "Associating ITS Data Categories with Existing Markup" [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action10]
<scribe> ACTION: editors to add Yves sentence about associating ITS with existing markup to sec http://www.w3.org/TR/its/#purpose-mapping [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action11]
<rahtz> <its:termRule selector="//tei:term" termRefPointer="@target"/>
felix: as for anyURI in the "ref" attributes, we can just say the types should be compatible
... as for
XPath expressions in pointer attributes, we can say "the interpretation is the same as the non pointer functionality"
sebastian: we need to make clear that all pointer attributes are relative XPath expressions
<scribe> ACTION: editors to make clear that all pointer attributes are relative XPath expressions [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action12]
sebastian: is rules an "island" in the host vocabulary, or is it interleaved. Goes back to the question on
<rules>
... e.g. the question "who writes the rules?", or "how are they interleaving?"
... you are inventing a new markup
language
christian: I would like to see versioning information interleaved with the meta data
... I see a need for
having meta data
felix: what would you choose? SAP manual, TEI header, dublin core etc.?
christian: we could use whatever you like, or we define at least a standardized set of meta data
sebastian: should rules be part of meta data, or the other way around?
felix: it would take a lot time to do this
sebastian: we could describe as a technique "use dublin core for meta data"
christian: putting a remark in the techniques document would be fine
yves: what vocabulary would that be? ITS or s.t. else?
christian: we could stay away from ITS, just say "use what you like"
sebastian: if the rules are embedded within a bigger document, are they embedded within the ITS rules?
christian: in the ITS rules
sebastian: we have identification with xml:id, no need to invent that again
... is it o.k. to have
<rules> as an island in another document? if that is possible, you could have all kinds of metadata in that
... we could say "it is good
practice to have a time stamp on your file"
... and refer people to the techniques document
yves: does it need to be in the ITS tagset draft or in the techniques document?
sebastian: processing of the xlink attribute is like descendant-or-self in XPath
<scribe> ACTION: editors to change example 13 to add meta data to it [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action13]
yves: I would say "version 1.1"
sebastian: we could embedd these examples in the schema
yves: is it possible to process this separatly?
sebastian: yes
christian: will we provide a web service for checking these constraints?
felix: W3C has no resources for that :(
<scribe> ACTION: Felix to write the constraints for pointer versus not pointer, to be implemented as schematron schema annotations, useable also separately [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action14]
yves: current consensus: wait for comments on the spec
yves: this has been postponed to see if the get feedback from the public.
"These conformance types are defined complementary. " -> "These conformance types complement each other. "
"subsections in Section 6: Description of Data Categories": this must also talk about sec. 5
"All local ITS attributes should be declared at all elements which are part of the schema."
yves: why this?
sebastian: if you have valid XML documents, you don't need this
felix: this conformance level is for schema authors, without the instance document
<chriLi> Could we look at http://www.imsproject.org/metadata/mdbestv1p1.html#Conformance
yves: proposal: you have to implement one of the declarations: rules, at least one local attribute, ruby,
span
... still have clause 1-1,3,4
... but not 1-2
... one additional clause / paragraph: an implementation which implements
everything has "complete" markup declarations
<chriLi> How about: A schema can fullfil one of two conformance levels: full or partial.
<chriLi> Full conformance is defined as follows ... Partial conformance ...
<chriLi> And have a note which explains the rationale and advantage.
<chriLi> Ex. Advantage of partial conformance: Do not pollute ...
<scribe> ACTION: do rewording of conformance sec. 4.1 together [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action15]
sebastian: section 4.2: "generated or validated relying on"->"valid against"
yves: "element within text" should not include child elements
<scribe> ACTION: editors to change default selection for "element within text" [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action16]
<scribe> ACTION: editors to add a link to data category for conformance clause 2.3 [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action17]
christian: how about not slicing global versus local selection?
yves: people might want to use only one of them
"it must process all of the XLink href attributes found in the rules elements." -> "it must process an XLink href attribute found on a rules element"
"Processing expecations are the information which are necessary to attach ITS information to one or more nodes in an XML document"
(proposal by Christian / Sebastian)
<chriLi> ITS markup allows information to be attached to XML nodes. Processors need to compute the information which is attached to an XML nodes (or set of nodes) correctlly.
<chriLi> Addition: Correct computation especially involves support for selection mechanism, defaults, and precedence.
summary: MUST for processing at least one data category, and then 2-1 - 2-3 as a nested list
christian: add a sentence to "description":
... "An implementation of a data category MUST (SHOULD?)make
explicit what selection (global or local) it supports"
<chriLi> selections _and_ data categories?
yves: for each datacategory: say you support it or not, and say which selection you support
sebastian: how would a conformance claim look like?
... each conformance claim MUST have a matrix about
datacat (yes/no), selection global (yes , no), selection local (yes, no)
<scribe> ACTION: editors to add xml:id attributes to all div elements (to avoid automatically generated id attributes) [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action18]
<scribe> ACTION: felix to clean up old change markup from the draft [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action19]
<scribe> ACTION: discuss change of "may" to "should" at The markup may be valid against a schema which is conform to the conformance clauses in Section 4.1: Conformance Type 1: ITS Markup Declarations. [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action20]
<scribe> ACTION: discuss moving the examples of ITS applications (editors etc.) in the conformance "normative" sub section [recorded in http://www.w3.org/2006/04/18-i18nits-minutes.html#action21]
See if changes have been integrated properly.