W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > October to December 2002

RE: Impact of XML 1.1 and namespaces 1.1 on WebDAV

From: Julian Reschke <julian.reschke@gmx.de>
Date: Sat, 16 Nov 2002 10:57:50 +0100
To: "WebDAV" <w3c-dist-auth@w3.org>
Message-ID: <JIEGINCHMLABHJBIGKBCMEPFFMAA.julian.reschke@gmx.de>

I'm in violent diasgreement.

The principle you mention is one of the biggest factors leading to
interoperability. If servers do not reject non-compliant requests, clients
do not get fixed. If clients do not get fixed, other server implementors are
forced to implement "workarounds" as well. We already see the results of
this within WebDAV -- almost everybody has workarounds in the server code to
please the most-deployed but worst-supported client (Microsoft webfolders).

While doing this sometimes is unavoidable in real life, putting this into
the protocol is a completely different story.

In this particular case, allowing XML 1.1 in PROPPATCH requests will allow
creation of property names/values that can not be sent back using XML 1.0
(*). Furthermore, there's no reliable way how a server can find out whether
the client actually *understands* XML 1.1. So supporting XML 1.1 will
require new compliance levels or it *will* break communication with older
clients. That's why we really need to understand what's going on before
making changes in RFC2518bis. Right now RFC2518 normatively refers to XML
1.0, and we may want to clarify that this means that requests marshalled in
any other XML version *must* be rejected.

Julian

(*) Unless we allow XML 1.1 but explicitly forbid property names/values that
cannot be marshalled using XML 1.0 (in which case using XML 1.1 in the first
place is questionable).

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

> -----Original Message-----
> From: w3c-dist-auth-request@w3.org
> [mailto:w3c-dist-auth-request@w3.org]On Behalf Of Eric Sedlar
> Sent: Saturday, November 16, 2002 12:59 AM
> To: Julian Reschke; WebDAV
> Subject: Re: Impact of XML 1.1 and namespaces 1.1 on WebDAV
>
>
>
> In the general theory of being flexible with what you accept and careful
> with what you generate, it sounds like WebDAV servers should accept
> XML1.1 input, but only generate XML 1.0 output.
>
> I also recommend us looking into the IRI issue.
>
> --Eric
>
> ----- Original Message -----
> From: "Julian Reschke" <julian.reschke@gmx.de>
> To: "WebDAV" <w3c-dist-auth@w3.org>
> Sent: Wednesday, November 13, 2002 2:58 PM
> Subject: RE: Impact of XML 1.1 and namespaces 1.1 on WebDAV
>
>
> >
> > (here's a minor update for usage during the IETF meeting)
> >
> > Hi,
> >
> > as the W3C is finishing these specs, I'd like to make the WG
> aware of the
> > questions we'll have to answer about how these changes impact WebDAV.
> >
> > Changes that are IMHO relevant to WebDAV are:
> >
> > 1) The set of characters that can appear in an XML name has
> been extended
> > (it now includes more Unicode characters). Infosets using these
> names can
> > *not* be marshalled as XML 1.0 documents.
> >
> > 2a) The set of characters that can be marshalled as text data has been
> > extended to include the control characters 1..31 (although only as
> character
> > reference; null is still forbidden). Again: infosets containing these
> > characters can *not* be marshalled as XML 1.0 documents.
> >
> > 2b) The characters 128..159 (Unicode control characters) can not be
> > marshalled as is (only as character references). Infosets
> containing these
> > characters can be marshalled as XML 1.1 documents, but requires
> > serialization as character references (which will be XML 1.0
> compatible as
> > well).
> >
> > 3) XML 1.1 processors may reject documents if they contain
> non-normalized
> > Unicode (yes, this is optional).
> >
> > 4) Namespaces 1.1 allows IRIs rather than only URIs as namespace names.
> >
> > One possible position would be that we just ignore it. Nothing would
> change
> > immediately. However we lose the ability to marshall "any" kind of XML
> > through WebDAV properties.
> >
> > Allowing XML 1.1 request bodies for PROPPATCH *will* have a big impact.
> For
> > instance, properties are identified by QNames, and with XML 1.1 both the
> > valid character sets for the namespace name (can now be a IRI and thus
> > contain non-quoted non-ASCII characters) and the local name
> (just a bigger
> > subset of Unicode) will change. It may not be possible to marshall these
> > "extended" property names back to a client that only
> understands XML 1.0.
> >
> > Similar problems appear with control characters in property
> values -- once
> > they appear in a property value, the property can only be
> marshalled back
> to
> > clients with XML-1.1 compliant parsers.
> >
> > At this point, I don't have a recommendation how to treat this,
> but maybe
> > some more WG members should take a look at the current drafts. XML
> > namespaces 1.1 allowing IRIs is currently under debate on the W3C TAG
> > (Technical Architecture Mailing List) -- people interested in this topic
> > might want to check the mailing list archives.
> >
> > Julian
> >
> > --
> > <green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760
> >
> > > -----Original Message-----
> > > From: Julian Reschke [mailto:julian.reschke@gmx.de]
> > > Sent: Tuesday, October 22, 2002 10:04 AM
> > > To: WebDAV
> > > Subject: Impact of XML 1.1 and namespaces 1.1 on WebDAV
> > >
> > >
> > > Hi,
> > >
> > > as the W3C is finishing these specs, I'd like to make the WG
> > > aware of the questions we'll have to answer about how these
> > > changes impact WebDAV.
> > >
> > > Changes that are IMHO relevant to WebDAV are:
> > >
> > > 1) The set of characters that can appear in an XML name has been
> > > extended (it now includes more Unicode characters).
> > >
> > > 2) The set of characters that can be marshalled as text data has
> > > been extended to include the control characters 1..31 (although
> > > only as character reference; null is still forbidden)
> > >
> > > 3) XML 1.1 processors may reject documents if they contain
> > > non-normalized Unicode (yers, this is optional).
> > >
> > > 4) Namespaces 1.1 allows IRIs rather than only URIs as
> namespace names.
> > >
> > > One possible position would be that we just ignore it. Nothing
> > > would change immediately. However we lose the ability to marshall
> > > "any" kind of XML through WebDAV properties.
> > >
> > > Allowing XML 1.1 request bodies for PROPPATCH *will* have a big
> > > impact. For instance, properties are identified by QNames, and
> > > with XML 1.1 both the valid character sets for the namespace name
> > > (can now be a IRI and thus contain non.quoted non-ASCII
> > > characters) and the local name (just a bigger subset of Unicode)
> > > will change. It may not be possible to marshall these "extended"
> > > property names back to a client that only understands XML 1.0.
> > >
> > > Similar problems appear with control characters in property
> > > values -- once they appear in a property value, the property can
> > > only be marshalled back to clients with XML-1.1 compliant parsers.
> > >
> > > At this point, I don't have a recommendation how to treat this,
> > > but maybe some more WG members should take a look at the
> current drafts.
> > >
> > > Julian
> > > --
> > > <green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760
> > >
> >
> >
>
Received on Saturday, 16 November 2002 04:58:05 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:44:02 GMT