[
Lists Home |
Date Index |
Thread Index
]
10/27/2002 7:46:52 AM, "Michael Kay" <michael.h.kay@ntlworld.com> wrote:
>I think that many of the design decisions in XML (and SGML) can only be
>justified by considering it primarily as an interchange format.
>
>No-one designing a storage format, for example, would have designed it
>so that you can't locate an element by ID except by reading the file
>serially from the beginning. That's why XML databases, even native XML
>databases, don't actually store XML: they store a representation of the
>XML InfoSet optimized for storage and retrieval.
OK, I see your point and more or less agree. SGML or document-oriented XML
may be the "native" storage format for lots of industrial-strength
applications (aircraft maintenance manuals being the canonical
example) but the requirement for interchange, interoperability, and usability
by future generations of software totally dominates the design.
I guess I wouldn't want to distinguish "XML" from "a representation of
the XML Infoset" however, but that's another perma-thread.
|
