Lists Home |
Date Index |
5/20/2002 1:05:27 PM, Bob Hutchison <firstname.lastname@example.org> wrote:
>What XML parser isn't DTD aware? Perhaps I misread the XML document and
>misinterpret the OASIS compliance tests, but it seems pretty clear that
>there are two kinds of XML parser: validating, and non-validating. Both have
>to read the DTD, parse it, and do certain things with it. Among the
>non-validating parsers there are a few types: those that read the external
>subset and those that don't. Even the ones that don't have to deal with the
>This statement you make about XML 2.0 I find a bit worrisome. Where are the
>entity references going to be defined if not the DTD? Surely not the
In DTDs, but I was envisioning a future where DTDs are (like W3C XSD, RELAX NG,
etc.) optional layers on top of an XML core rather than something that have
to be supported by every parser claiming to handle "XML". This allows the
"docheads", "dataheads", and "SOAP-heads" to share a common underpinning, but
not tangle up each other.
Even without this in XML 2.0, I envision parsers optimized for XML 1.0
+ namespaces - DTDs if SOAP or some other "Common XML"-based vocabulary
becomes widely deployed. Imagine a world where routers, firewalls,
virus scanners, proxys, etc. all have to look into thousands of XML messages
per hour to figure out what to do with them, and cellphones, PDAs, etc.
all have tools to produce and manipulate these messages. It seems likely that only a
tiny fraction of those messages will use DTD-based constructs such as entities.
What's the value proposition for Microsoft, Intel, Cisco, Palm, etc. to put
fully XML 1.0 conformant parsers in these devices? Even if it can be shown that
there is no performance overhead to having DTD-aware code in a parser, that
stuff SURELY adds to the time it takes to build and fully test the parser.
Whether or not this is a Good Thing has spawned dozens of thread here and on
sml-devl, and I won't reiterate them. I do think that it has become VERY
clear in the last couple of years that the major players won't let the W3C
specs dictate their development plans if they don't see a good business case
for full conformance.
Anyway, that's all hypothetical, but is the reason why I'm leery of the "all
the tools I use support [namespaces, DTDs, W3C Schemas, etc.]" argument for
not worrying about the problems that one or more may cause.