[
Lists Home |
Date Index |
Thread Index
]
8/20/2002 6:49:50 PM, bob mcwhirter <bob@werken.com> wrote:
>> Now putting it all together:
>>
>> "DOM: a specification of implementation of
>> representation of representation of our data".
>>
>> Any problem with this definition?
>
>Many might argue that the DOM is an in-memory representation of our data,
>and XML is just a serialization of that repsentation to octets.
Right. I think it's better to say that DOM is an
application programming interface to the XML InfoSet. The InfoSet could
represent HTML after being fixed up by the parser, XML after the parser
has done its job, or some other syntax (WikiML, for example) after being
parsed into Infoset-compatible form.
[OK, "an" infoset, not "the" InfoSet as far as namespaces are concerned!
Dare, if you want a concrete illustration of why namespaces are problematic,
consider how differently they are represented in the various specs.]
So, what's an "interface"? "a specification of implementation of
something whose implementation is susceptible of specification" may
be a joke and I'm humor-impaired, but it's not a particularly great
definition. How about "a specification of a software operation
as seen by a program that uses it, without reference to its internal
implementation."
So, the DOM is a "specification of a standard set of operations on
an XML InfoSet, without reference to its physical data structures or
internal implementation."
|