See also: IRC log
Resolution: Pratul will send email request
Resolution: approved by attending members
Note: no quorum present; decision is to discuss issues today and bring issues up to rest of group when a quorum present
Kumar: wants to have confidence that a valid model is really valid.
Ginny: use case where schema author does not want to make a target required but, if present, wants it be a specific type.
Sandy: we have only 2 states -
valid and invalid
... if we separate concerns we can cover all use cases
... targetRequired controls whether a ref resolves not
targetType
Kumar: make this dependent on the whether we decide on an "unknown" state
Sandy: model validator could
state that it cannot perform e.g. network down.
... this affects more than just this bug
Kumar: if validator cannot reach the target and the target is actually not the correct type; user will not know
Sandy: memory bad or no network is not an issue on this bug only; this is an issue for the validator as a whole
Kumar: if this kind of an error
we cannot say if the model if valid or not.
... then this "unknown" does not apply to a reference but to
the model as a whole
... rephrasing: When you are trying to resolve a scheme and it
cannot be resolved due to an error such as network, etc., the
entire model is declared to be unknown.
... then ok with targetType/Element in this bug be specified as
"satisfied"
Resolution: When you are trying to resolve a scheme and it cannot be resolved due to an error such as network, etc., the entire model is declared to be in an unknown state. Then for an unresolved reference, targetType/Element is declared "satisfied".
Sandy: no inheritance for
particle restrictions so argument is not the same as Bug
4643
... need to specify how inheritance happens
Pratul: original spec wanted
constraints to be inherited
... why do we need to look at schema 1.1?
Sandy: describes the inheritance problem...
<scribe> ACTION: Sandy to add an example of the problem to the bug [recorded in http://www.w3.org/2007/11/08-sml-minutes.html#action01]
<trackbot-ng> Created ACTION-149 - Add an example of the problem to the bug [on Sandy Gao - due 2007-11-15].
Sandy: preference is that we don't inherit
Resolution: mark as editorial; editors to ensure that all issues in this bug have been addressed.
Sandy: working on a proposal covering inter-document references that will cover this bug as well as others
Resolution: Review this bug after reviewing Sandy's proposal.
Marking this bug as needsAgreement.
Resolution: mark as needsAgreement and wait for Sandy's proposal
Kumar: proposes to replace schemaLocation uri's with alias to document
Sandy: but base uri also applies to sml:uri elements and possible other new schemes.
Kumar: if xml:base brings other semantics with it, one possibility is to back to baseURI
Sandy: having a single baseURI is
not sufficient; several documents may need different
baseURIs
... candidates that need a base include schemaLocations and
sml:uri but should also cover additional extensions like new
schemes
... normally expect that relative URI is resolved based on
containing document
... so alias must retain information on directory structure
that is lost when packaged into an SML-IF document
... each document has its own baseURI
Kumar: sml-if doc baseURI applies
to every embedded document?
... each document has its own baseURI, a schemaLocation, and
sml reference. SchemaLocation is written with baseURI in mind
but sml:uri may not be.
Sandy: can always add other aliases
Kumar: are aliases absolute?
Sandy: yes, they need to be
Pratul: every relative URI is based on baseURI
Kumar: uncomfortable with each
document having its own baseURI
... if 10 incoming references and each with its own baseURI,
will need 10 aliases for the document.
Sandy: don't need different URIs
<Sandy> A: file://dir/a.xml B: file://dir/b.xml in A, there is a reference "../b.xml#xyz".
Kumar: 2 docs with different baseURIs. A has ref to E in B. A's ref is relative to A. To reach B, take baseURI of A plus relative ref to reach document B. So B will need an alias to match.
<Sandy> B: file://dir2/b.xml
<Sandy> reference: "../dir2/b.xml#xyz"
Sandy: in most cases, baseURI and alias will be the same
Pratul: could go back to one baseURI or change all uris to be absolute
Ginny: suggest that we limit a "conforming validator" to the #ALL phase; this does not mean a validator does not support other phases.
Resolution: Ginny will add another bullet point to address Sandy's concern about the #ALL phase
Updated scribe list for next meeting Last Scribe Date Member Name Regrets pending 2007-08-30 Wilson, Kirk 2007-08-30 Lipton, Paul 2007-09-20 Lynn, James 2007-10-16 Valentina Popescu 2007-10-15 Waschke, Marvin 2007-10-17 Eckert, Zulah 2007-10-17 Kumar, Pandit 2007-10-24 Boucher, Jordan 2007-11-01 Gao, Sandy 2007-11-08 Smith, Virginia 2007-06-12 Tabbara, Bassam Exempt Arwe, John Exempt Dublish, Pratul Exempt MSM Exempt PH