Interview with Chuck Jazdzewski

MURATA Makoto eb2m-mrt at asahi-net.or.jp
Wed Jun 26 03:07:12 CEST 2013


Dear colleagues,

Here is my proposed text (as part of the minutes) about the interview
with Chuck during the WG4 meeting in Bellevue.  I thank Microsoft
for providing this great opportunity.

Regards,
Makoto

Interview with Chuck Jazdzewski

WG4 had a chance to remotely interview Chuck Jazdzewski.  He
contributed to the design of original MCE.

1) Declaring application-defined extension elements

WG4 asked why there are no mechanisms for XML documents to declare
application-defined extension elements.  Chuck provided a very clear
answer: Application-defined extension elements have to be chosen when
original markup consumers are created, and thus we must not allow new
producers to declare new application-defined extension elements, which
cannot be handled by existing producers.

Note: WG4 should reconsider its decision to introduce the
ExtesionElement attribute.

2) Dropping PreserveAttributes and PreserveElements

Chuck agreed on the WG4 decision to drop them and further said that
they were incorporated into the first edition of ISO/IEC 29500-3 by
mistake.

3) Droppping namespace subsumption

At first, Chuck was surprised to hear that WG4 has decided to drop
namespace subsumption.  It has been used for XAML for example.
However, he agreed that namespace subsumption can be done by some
technologies for namespace renaming.  Such renaming works as a thin
layer between the MCE processor and application programs.  He
continues to believe that his implementation, which does namespace
subsumption as well as MCE processing, is conformant.  WG4 agreed.

4) Non-MCE attributes and child elements of AlternateContent elements

The current design allow them only when they are in the ignorable
namespace.  Chuck said that this restriction is intended to simplify
the design/implementation, if not mandating use of infosets.  He did
not oppose to allow any foreign attributes and child elements.

5) Inherited attributes

WG4 asked why MCE mentions inherited attributes in the XML namespace
and no other inherited attributes. Chuck said that this is
intentional.  When some elements are unwrapped, consumers do not know
how to handle these elements and thus their attributes should be
ignored together.  Chuck also said that all inheritted attributes in
the XML namespace should be mentioned in 29500-3.

6) Copying attributes

The current draft mentions copying of attributes from ancestors to
subordinates and mentions the use of infoset for such copying.  Chuck
thinks that the use of infosets would make MCE simpler.  He further
pointed out that text chunk directly under an unwrapped element cannot
hold inherited attributes and that some nicely-defined infoset can
allow text chunk to have attributes.

7) Multiple invocation of MCE

The current working draft does not allow MCE resumption within
application-defined extension elements.  Rather, it is assumed that
subordinate elements of such application-defined extension elements
may be handled by another invocation of the MCE processor.  Chuck
agreed on such multiple-invocation and said that the original MCE
concentrated on a single invocation.


More information about the sc34wg4 mailing list