Re: Notes on "Document Assembly and Context Rules"

[nagwa <nagwa.abdelghfour@sun.com>]

I would like to second Jon about the assembly spec,
this is the one I talked about during our con call for being
ready for public review

Nagwa

Jon Bosak wrote:

> I said during our Friday call that I thought the documents that I
> had read up to that point needed another pass by their editors
> before going to public review.  This weekend I have read the
> document titled "ebXML specification for the application of XML
> based assembly and context rules," and today I had a chance to
> speak to one of the document's authors, Eduardo Gutentag.
>
> I believe that this document, while not yet complete, is a fine
> piece of work.  It's well written and it's concrete enough that
> one could actually implement it.  I have some editorial comments,
> but I will send those in on hardcopy to James Whittle.
>
> Beyond editorial changes I see a couple of substantive issues,
> which I will raise here in the hope that they will be taken into
> account by the team working on the "assembly and context"
> document.
>
> (My mail system is having trouble talking to the list server, so I
> am addressing this not just to the QR list but also to individuals
> on the QR team, and I'm including the two editors of the "assembly
> and context" document as well as the person who generated the DTDs
> used as examples.  I will leave to Tim the question of how to
> handle followup, if any.)
>
> 1. The DTD in Section 7.2 of "assembly and context rules" contains
>    (line 299) a Condition element that can be used to apply
>    context information at the assembly stage.  An example of this
>    appears in lines 435-440 (Section 7.4).
>
>       435  <Condition test="Region='UK'">
>       436    <Rename from="address" to="addressUK"/>
>       437    <Rename from="City" to="Place"/>
>       438    <Rename from="address/State" to="County"/>
>       439    <Rename from="address/ZIP" to="PostalCode"/>
>       440  </Condition>
>
>    The process described in the document takes place in two phases
>    that are (as I think they should be) specified in two
>    documents, one containing assembly rules and the other context
>    rules.  But the declaration of a Condition element in the
>    assembly rules blurs this distinction and loses a lot of
>    conceptual coherence for no apparent reason.  Unless we can
>    point to some terrific practical advantage to be gained by
>    allowing context-sensitive schema rewriting to take place in
>    both phases, it seems to me clearly better to reserve the
>    context-driven rewriting for the context rules phase in which
>    one would expect to find it.  Surely the convenience of knowing
>    how rewriting is specified just by looking at the context rules
>    is a benefit that should not be discarded lightly.
>
> 2. The change I'm suggesting would have an impact on the diagram
>    at line 148 of the document currently titled "ebXML The role of
>    context in the re-usability of Core Components and Business
>    Processes."  The box labeled "Context Rules" in "The role of
>    context" is shown as applying to both "Assemble Types" and
>    "Type Use Rules" (as well as to "Application Component," which
>    is possible but should not be required).  If my point above is
>    accepted, then this diagram should change to reflect it.
>
> 3. In general, the process described in "assembly and context
>    rules" is not aligned with what purports to be the same process
>    in "The role of context."  The latter process is summed up in
>    diagrams appearing in Section 5.1 of that document at lines 148
>    and 156.  In my opinion, these diagrams in "The role of context"
>    should be aligned with the process described in "assembly and
>    context rules," and then the prose in "The role of context"
>    should be rewritten to reflect the same concrete technique.
>
> Jon