W3C home > Mailing lists > Public > xml-dist-app@w3.org > October 2001

Re: Issue 4 Proposed Resolution (was: why no doc type declaration and PIs in SOAP)

From: Bob Hutchison <hutch@xampl.com>
Date: Mon, 01 Oct 2001 20:26:54 -0400
To: <francis@redrice.com>
CC: Andrew Layman <andrewl@microsoft.com>, xml dist <xml-dist-app@w3.org>
Message-ID: <B7DE820D.7A9%hutch@xampl.com>

The proposal allows for a choice of either failing or ignoring the PI/DTD in
the message. The reason the 'ignoring' option is there is because some feel
that a simple XML parser cannot detect the PI/DTD. I'm simply questioning
that argument. If that argument doesn't hold, then we don't need the
'ignore' option and we can simply fail.

I would add that if we really wanted to support simple parsers, then we
should support the parser that doesn't recognise the DTD or PI at all and so
cannot parse the message. We are requiring the parser to ignore the message
or fail with an error code identifying the source of failure as either DTD
or PI. What if the simple parser cannot tell which it is and cannot ignore
it? Is there another way to fail? What happens?

If I understand the proposal, you are prohibited from making use of the
information in the DTD, you must ignore it. This means that the XSLT based
parser that you mention would not be acceptable.

I'll let others respond to your last paragraph, because I don't understand
it. The only comment I'll make now is that, in my opinion, this is a much
more significant issue than 'issue 4'.

Cheers,
Bob

On 01/10/01 6:46 PM, "Francis Norton" <francis@redrice.com> wrote:

> Bob Hutchison wrote:
>> 
>> Actually, what I asked was how to skip over a PI or DTD without knowing what
>> it was you were skipping over. In your example you know it is a PI. If you
>> know it is a PI then why can't you fail immediately with the proper code? If
>> you know it's a PI or DTD, why allow anything other than failure? What's
>> this 'simple' processor that can skip a PI and DTD and not know what it was
>> that it skipped?
>> 
> Before MS SOAP came out, we implemented a simple SOAP library using our
> server-side JScript-based integration tool and XSLT. XSLT will tell me
> if there is a PI, but the only way to detect a DTD is to parse the XML
> document manually, which seems a perverse kind of simplification to me.
> 
> Why the assumption that SOAP tool developers will all be implementing
> their own XML parsers? Why not use platform standards or open source
> micro-parsers like Aelfred (the default parser for Saxon)? Or better,
> assume that the library will be passed an in-memory representation such
> as a DOM, SAX event stream or .NET XmlWriter?
> 
> Francis.
Received on Monday, 1 October 2001 20:27:38 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:59:04 GMT