Re: LOCK refresh on indirectly locked resource

Joe Orton wrote:

> On Sun, Sep 12, 2004 at 07:39:43PM +0200, Julian Reschke wrote:
> 
>>It's some time ago that somebody asked how servers implement this today. 
>>I finally got around to write a test case (attached JScript), including
>>
>>1) using a no-tag list If header,
>>2) using a tagged If header, specifying the lock root and
>>2) using a tagged If header, specifying the indirectly locked resource.
> 
> 
> Is your latter (2) behaviour defined by 2518 or 2518bis? I suppose both
> are equivalent by 2518 at least.

I don't think RFC2518 defines any of these. This is why we were trying 
to find out what servers actually *do* implement, so that it can be 
clarified in RFC2518bis. IMHO, nobody seems to be doing this in practice 
(otherwise the bug in mod_dav would have surfaced earlier); thus 
RFC2518bis probably should say that you can't do a LOCK refresh that way.

>>Results:
> 
> ...
> 
>>Apache2.0.50/mod_dav: *crashes*, see 
>><http://nagoya.apache.org/bugzilla/show_bug.cgi?id=31183>
> 
> 
> It looks like the code path for indirectly refreshing the lock had never
> been tested in mod_dav, it has never worked AFAICT. I can't find any
> reports of this bug being triggered before, either, so I guess this is
> not the most widely used protocol feature...

Yep.

> (I've made a new release of the litmus test suite which also includes a
> test for an indirect lock refresh for any others who are interested and
> can't run JScript: http://www.webdav.org/neon/litmus/)

Thanks, Joe. Your work on Neon and Litmus is a really a big contribution 
to WebDAV.

Best regards, Julian

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

Received on Monday, 13 September 2004 10:21:38 UTC