W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > October to December 1998

Re: interoperability testing of mod_dav

From: Greg Stein <gstein@lyra.org>
Date: Wed, 18 Nov 1998 03:16:10 -0800
Message-ID: <3652AC7A.34C74A4E@lyra.org>
To: Jim Davis <jdavis@parc.xerox.com>
CC: w3c-dist-auth@w3.org
Thanx for the testing, Jim!

Jim Davis wrote:
> ...
> The client is a WebDAV client side library in Python.  Almost everything
> worked perfectly.

Interesting point that we're both using Python clients to do our
testing... :-)

> I report the following bugs, some of  which may actually be bugs in my
> understanding of the protocol
> 
> 1. mod_dav does not allow one to set the dav:getcontenttype property, or at
> least the value one sets is ignored.  I created a file

Actually, it should be returning a 500 right now (which in itself is an
error, but if it isn't returning 500, then I've got a different, more
subtle error occurring).

(actually, the 500 should be returned for DAV:getcontenttype; it
shouldn't have an issue with dav:getcontenttype)

> http://msdav.lyra.org/dav/foo.html and set the content type to text/plain.
> But the server is returning a content type of text/html.
> 
> In my opinion, this should not be a read-only property.  After all, how
> could a client set a property value for a type the server did not even know?

The spec is quite unclear here. mod_dav will consider DAV:displayname,
DAV:source, and DAV:resourcetype (for non-collections) to be read/write.
Properties in the DAV: namespace not specifically detailed in the
specification will be considered to be dead properties.

> 2. An attempt to set a read-only property such as DAV:creationdate returns
> a 500.  It should return 409 conflict I think.  (see 7.2.1)

Bug in my code. In fact, I already had a comment in there stating that a
500 would be returned, which is improper. There is a bit of design
refinement that I need to do in some of the property stuff to deal with
this and some other situations.

I agree that 409 is the proper response.

> 3. I was not able to set a property value that contained an attribute.  (At
> least the xml:lang attribute should be supported)
> 
> Request:
> <?xml version="1.0"?>
> <A:propertyupdate xmlns:A="DAV:">
>   <A:set>
>     <A:prop>
>       <B:author xmlns:B="http://example.com/" xml:lang="nl">h van der
> kaas</B:author>
>     </A:prop>
>   </A:set>
> </A:propertyupdate>
> 
> Reply
> 
> <!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN">
> <HTML><HEAD>
> <TITLE>400 Bad Request</TITLE>
> </HEAD><BODY>
> <H1>Bad Request</H1>
> Your browser sent a request that
> this server could not understand.<P>
> An undefined namespace prefix was used.<P>
> <HR>
> <ADDRESS>Apache/1.3.3 Server at kurgan.lyra.org Port 80</ADDRESS>
> </BODY></HTML>

There are several issues here:

1) mod_dav does not recognize the xml: prefix, so it barfed on the
xml:lang attribute. (BUG)
2) mod_dav's design states that attributes (other than namespace
declarations) on the element that names the property (B:author in this
case) are not saved. That is detailed on the mod_dav web page. Note that
all attributes on child elements /are/ saved.
3) it is arguable whether the xml:lang should be saved when it occurs on
the property name element, or whether a client is required to pop it
down to a child element (with the caveat of (1) where xml: is not
currently recognized)

> Other than these three errors, I was able to do PUT, MKCOL, DELETE,
> PROPPATCH and PROPFIND to good effect.

Excellent. Thank you for your testing.

> Of course, mod_dav has no locking,
> so I could not test that.

Yup. It is only a Class 1 right now. I believe the community will find a
Class 1 useful, so I want to complete and release that work first. Class
2 capabilities will follow in a future release, along with capabilities
from the other WebDAV specs/drafts.

Thanx,
-g

--
Greg Stein, http://www.lyra.org/
Received on Wednesday, 18 November 1998 06:15:48 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:43:48 GMT