Re: binaryXML, marshalling, and and trust boundaries

Dan Connolly writes:


> B: well, what you do in the privacy of
> your own home/trusted-net is your business.
> If this isn't about interoperability
> between arbitrary parties in the net/web,
> then you really don't need our endorsement,
> do you?


Hmm, it seems that standards are more than just about interoperability
and "endorsement."  A big part of XML's success is simply from the network
effect of lots of people "agreeing" to stop reinventing the
grammar/syntax/parser wheel with every application.  So, there's something
to be said for standardizing on one or more alternative syntaxes for the XML
infoset that would meet the needs of specialized communities so they don't
all have to reinvent the "efficient XML serialization format" wheel.  XML is
big enough now that this would not fragment it into non-interoperating
communities (after all, many XML tools operateat the Infoset level anyway).

I see a number of advantages for having a menu of standardized serialization
formats rather than insisting that one size fits all.  One might be
optimized for simple authoring, a la WikiML.  Another might be optimized for
fast parsing, something on the order of serialized SAX events.  Another
might be optimized for bandwidth, a la WAP/WML "done right", in light of
what people learned from that experience (if indeed something other than
gzipped XML 1.0 is what they need, I don't know).

The only one I have personal knowledge of is the "serialized SAX events"
idea; I know of experiments in this area by middleware vendors who have the
profiling data to  show that XML 1.0 parsing is a significant bottleneck.
(Note -- we're talking about software that does very little BUT parse, peek
at and tweak, and then serialize XML, so it should not come as much of a
surprise that the XML parsing takes a significant percentage of its
resources!) I'm not sure what, if anything, the TAG should be doing about
this.  Obviously if the vendors who are working in this area want to come
together and  form a WG or TC to define standards so that the network effect
can work its magic, any such initiative should come from them rather than
the TAG or the XML Core WG. Nevertheless, I for one would be quite
disappointed to see you make some a priori declaration that this is a Bad
Thing and off-limits for the W3C.

[not speaking for employer or WSA WG]

 

Received on Monday, 2 December 2002 19:27:41 UTC