RE: Impact of XML 1.1 and namespaces 1.1 on WebDAV

> From: w3c-dist-auth-request@w3.org
> [mailto:w3c-dist-auth-request@w3.org]On Behalf Of Brian Korver
> Sent: Saturday, November 16, 2002 11:37 PM
> To: Julian Reschke
> Cc: WebDAV
> Subject: Re: Impact of XML 1.1 and namespaces 1.1 on WebDAV
>
> ...
>
> Julian,
>
> The principal you object to (Postel's robustness principal, spelled out
> in RFC-793, the TCP/IP draft), informs the design of many IETF
> protocols, especially when dealing with issues of backward
> compatibility.  Within the IETF, this principal is believed to
> increase interoperability, in contrast to what you state.

However, XML is conservative in both directions (malformed requests are
never accepted), and has reached very high interoperability. Are you
suggesting that an XML parser should accept documents with minor
wellformedness errors if it thinks it can fix them?

For example: Microsoft's XML parsers used to accept control characters that
aren't allowed in XML. People solely on the Microsoft platform started using
this as a feature -- they never tested against other implementations. Guess
what they said when Microsoft finally fixed their parser?

So yes, I think that this well known principle is *causing* interoperability
problems. It's much better to have a as-precise-as-possible protocol
definition and to mandate that both clients and servers stick to it.

> As for forcing servers to reject non-compliant requests, I think
> we'll all agree that non-compliant implementations are a problem,
> but how can the protocol document force these implementations
> into compliance?  And if vendor Y decides to interoperate with
> non-compliant vendor X, there is little that this working group
> can do to prohibit vendor Y from doing so.

It can't.

However, Eric was suggesting that the protocol explicitly *allows* accepting
XML 1.1 requests. That's very different from not saying anything about it.

Accepting XML 1.1 requests as they are defined right now would mean that
property values and names (and labels, privileges...) can occur that can't
be marshalled to a client using XML 1.0. That's what I call a major interop
problem.

> On the subject of certain non-compliant clients, if we were to
> prohibit the accepting of XML1.1, what effect would you expect
> that prohibition to have on the decisions made by the vendors of
> such clients?

Hopefully they'd send XML 1.0.

> Eric, I don't recall if you stated the client requirements.  Did you
> expect that clients would be permitted to generate XML1.1?
>
> Note, the above should not be construed as an argument either for or
> against WebDAV support for XML1.1.

Understood.

Actually, it's a good thing that finally we've got a discussion thread about
this topic :-)

--
<green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760

Received on Saturday, 16 November 2002 18:19:23 UTC