[cam] Revision 0.12 draft ZIP

[David RR Webber - XML ebusiness <Gnosis_@compuserve.com>]

Team,

Attached is the ZIP file of the latest edits.

I've received comments from Jim Wessely, Martin Bryan,
Bruce Peat and Duane Nickull.  I also received lots of
FAQ type questions from Ed Buchinski - those 
details are at our SWIKI.

So the attached reflects the technical edits
 (note - track changes is "On"
and that lets' you see all the changes directly).

Martin Bryans comments were the most technical in nature, 
and so I've included them here below, with my actions.

Duane made suggestions for improving the context passing
mechanism in the addendum, that I have adopted and
adapted.

Please review all this for our Thursday conference call.  We
will have more time for follow-on items and issues regardless
after Thursday.

I may also add that generally reviewers so far have been
very positive about the draft as it stands.

Thanks, DW.

p.s. We now have 20 participants in the TC, I'll look to get
a full roster out shortly - Karl has still to add another two people.
Our family is growing!

==========================================================

Martins comments:

I don't seem to be able to control the minimum number of an element, e.g.
if
member type is new or renew then number of items must be 4 or more.

>>> OK - added a new function - setRequired() to handle this.

I also don't seem to be able to do ranges, e.g. from 13 to 16 characters,
or
4-8 elements or 2-5 tokens in an attribute.

>>> OK - changed functions so you can overload them with range parameters.

The difference between useElement() and useChoice() is not clear.

>>> OK - basically useElement() does not require setChoice() first - so its
more easy to use.  Updated the documentation to stress this.
<<<

Does the treepath associated with useTree() identify the point the tree is
to be attached to or the tree to be attached to some point? In either case
something is missing as you need to specificy a) where the tree is to start
and b) which tree to use.

>>> Is relative node driven - changed documentation to clarify behaviour.

The list of types shown for StructuredType() in the table conflicts with
the
values assigned to this attribute in the DTD.

>>> Ooopps!  fixed the table entry.

The use of greater() as the opposite to lessthan() is confusing. Stick to
greaterthan()

>>> Changed to greatthan() throughout document

I would have expected a contains() extension for XPath to allow you to look
for strings within contents or attribute values. Starts and ends are easy,
its the centres that are runny.

>>> Ooops! contains added (was supposed to be in there!).

In figure 4.5.1 what does expression="'//RefCode + //UnitPrice'" actually
mean? It says to me that if any RefCode or any UnitPrice fails the test
report, but take no further action. What I think you are trying to say is
//RefCode(sibling::(UnitPrice)), i.e if the UnitPrice for the sibling of
any
RefCode meets the condition. The plus sign is ambiguous here and does not
imply that there is a relationship between individual members of each set.

>>> Its a string concatenation.  Added documentation and function.

The XPath predicates given for the EDI example in figure 4.6.2.2 are
invalid
and should not be referred to in terms of XPath. You could easily change
them into string searches that XPath would honour.

>>> This is a tougher one - not resolved this yet - needs more research.

For things like @multiply(p1,p2) you need to restrict the predicates to
being numeric, and state what the conversion rules are if they are of
different types. Simarly for @upper(p1) and @lower(p1) the predicate must
be
a string of Latin, Cyrillic or Greek characters: for most other languages
case is irrelevant. Incidentally you need generally to state the role of
ISO
10646 in all this.

>>> OK done.  Documentation enhanced.

Is there an on-line source of the spec I can point people to?

>>> I've put the ZIP for the 0.12 revision at our SWIKI -

 http://cam.swiki.net

Incidentally, if you want CAM to be used generally, as an extension to RNG
along the lines of Schematron, you need to sanitize the preamble to 
stress this is an open approach. If you do this I can submit CAM as a
possible part for the DSDL spec.

>>> Done.  Hopefully we now pass muster on this aspect!  Also longer
term - I want to make the Addendum where we cover off non-normative
bindings and implementation examples to the various sectors out there,
just as RosettaNet, EDI, CEFACT, UBL, and so on.

Attachment: CAM-v1-draft-022503.ZIP
Description: Binary data