[
Lists Home |
Date Index |
Thread Index
]
From: "Elliotte Rusty Harold" <elharo@metalab.unc.edu>
>
> At 10:26 AM -0500 10/30/02, Seairth Jacobs wrote:
>
> >2) I see no problem with having a parser that ignores portions of XML
spec
> >if it will never encounter those portions, such as DTDs or PIs (CDATA in
his
> >example). He claims this would cause interoperability issues. Where and
> >How?
>
> It depends on what you mean by "ignore". If they simply don't report
> them to the client app that's one thing. However, they are absolutely
> required to read processing instructions and the document type
> declaration and report any well-formedness errors in these
> constructs. A parser that does not do this is not an XML parser. I
> have seen more than one parser vendor try to leave this required
> functionality out of their toolkit.
>
> With regard to the internal DTD subset, it goes a little further.
> Even at the minimum conformance level, parsers are required to apply
> default attribute values and expand internal entity references
> declared in the internal DTD subset, at least up to the first
> external entity they don't resolve. Parsers that don't do this can
> cause real interoperability problems.
But all that presuposes that the parser will be encountering these parts of
XML. As I mentioned, if the parser will never encounter a particular part
of XML, then it should be okay to leave that code out. For instance, if a
parser handles only "standalone" documents, why should it need code to handl
DTDs? I absolutely agree that a parser should not ignore parts of the XML
spec that it will encounter, but that is not what I said above.
---
Seairth Jacobs
seairth@seairth.com
|