RE: Interoperability testing with Xythos DAV server

At 11:26 PM 7/20/99 -0700, Kevin Wiggen wrote:
>
>->2) the server does not seem to be preserving the xml:lang attribute
>
>I talked to Jim Whitehead on this.  He told me that I must keep the xml:lang
>attribute, but it is not intended to stripe the property.  In other words a
>propatch of
>
><K:foo xml:lang=us> foobar </K:foo>
>
>then propatch
>
><K:foo xml:lang=f> frenchfoobar </K:foo>
>
>Will result in only ONE value for <K:foo>.  The second value REPLACES the
>first even though it is a different lang.
>
>I think this is different than what you are telling me.

What Jim W said is the same as what I believe.  Sorry if I did not explain
it well.

>While I am at it.  Is the server required to keep ALL attributes sent to
>it???

Required?  No, at least not at present.  You'll note that the protocol
document is mute on the point.  SInce it says nothing, there's no requirement.

Some have stated that it *should* be required, but the list has never
reached concensus on the point. There are arguments both ways.  I won't
repeat them in full, but in brief they are:

NO:  The data model for WebDAV properties is a tree, where the leaf nodes
are string values, and the interior nodes have a label, BUT NO OTHER STATE,
except for the xml:lang attribute (and perhaps one or two other attributes
required by XML).  WebDAV uses XML as an on the wire format, but the data
model is not XML.  An additional argument against is the possible cost of
storing attributes.

YES: Since WebDAV uses XML, and since XML uses attributes, users will
expect it, so it should be required.

The list has never reached concensus on this point, and, as I recall, it
has not even been able to agree to place an explicit statement on the
non-agreement in the protocol document.  So the protocol is simply silent
on the point.

Received on Wednesday, 21 July 1999 13:41:42 UTC