Lists Home |
Date Index |
email@example.com (Kian-Tat Lim) writes:
>I think there may be another useful distinction between
>camps in the XML world besides the well-known "document"
>vs. "data" one. I think Walter, Simon, and others
>belong to a "publishing" group that stands in contrast to
>the "messaging" group.
Thank you for writing this, as it is a clearer expression of the
situation than I've been able to express myself. This message makes
nicely clear how publish/message are distinct from document/data.
>In a publishing model, as we observe on the Web,
>there is little opportunity for the creator of a document
>to negotiate with its recipient. As a result, fewer,
>more fundamental, more syntactic agreements are required
>to achieve interoperability.
Excellent. This meshes very well with my own early expectations for
"SGML on the Web".
>In a messaging model, the creator and receiver of a
>document have the opportunity to negotiate prior
>agreements such as schemata, use of binary serializations,
>particular APIs, etc. in order to produce more efficient
>Systems that rely on accessing "publish"-style data,
>like Walter's and Google, are hopelessly lost when faced
>with "message"-style data, since they cannot be parties
>to all the necessary agreements. That is how currently
>working integrations can be harmed by the proposed
>suggestions, unless they are used in a true messaging
>context internal to a system and never leaked to the
>outside world, or even the remainder of the enterprise.
This conclusion seems mostly correct. I have little interest in systems
that deal with extensively-negotiated frameworks and prior agreements,
in large part because I find the costs of those things to outweigh the
benefits on a lot of levels. Working with "evolving" or "found"
documents makes far more sense to me than expecting all parties to come
I'm not sure that publish-style systems are necessarily "hopelessly
lost" when they encounter message-style information, but they certainly
don't use the same supporting framework. Things like W3C XML Schema and
WSDL documents strike me as bad jokes rather than usable toolkits, and
certainly make integration more expensive for my style of work than it
would be otherwise.
Tim Bray gets it right when he says:
>To the extent that you minimize the negotiation, you win.
XML 1.0 gives me far fewer details to negotiate and far more room to do
things my own idiosyncratic way than does "the XML family of standards".
XML 1.0 represents enough of an agreement to be useful but not so heavy
an agreement that it impairs my work.
(I've argued for a long time that XML 1.0 has too many negotiable
options itself, of course, but it's a far better balance than the
Ring around the content, a pocket full of brackets
Errors, errors, all fall down!
http://simonstl.com -- http://monasticxml.org