W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > April to June 2000

(unknown charset) Re: If: header and "parent" resource checking

From: (unknown charset) Greg Stein <gstein@lyra.org>
Date: Mon, 29 May 2000 17:33:29 -0700 (PDT)
To: (unknown charset) jamsden@us.ibm.com
cc: (unknown charset) w3c-dist-auth@w3.org
Message-ID: <Pine.LNX.4.10.10005291731200.15722-100000@nebula.lyra.org>
On Mon, 29 May 2000 jamsden@us.ibm.com wrote:
> Greg,
> I believe DAV4J behaves as you describe below.

Great. Well, the details below can also make a good step-by-step review
list for DAV4J then.

I think that I'll code up some regression tests for these scenarios. The
locknull stuff may be changed if consensus goes with Geoff, but heading in
that direction will be helpful.

Cheers,
-g


Greg Stein wrote:
> 
> I'm going through mod_dav right now and refining/verifying what resources
> it checks during If: header processing. I'm taking a particular approach
> that I want to bounce off the group to see if this makes sense and is
> "proper".
> 
> Specifically, it revolves around whether we need to check the "parent"
> resource to see if it is locked and the If: header contains those locks.
> Say you're doing a MKCOL. The spec states that you must satisfy the locks
> held on the parent resource. The following cases are a bit less obvious:
> 
> *) when using LOCK to create a locknull resource, this effectively creates
>   a new resource in the parent collection.
>   ==> parent resource should be checked
> 
> *) when using UNLOCK on a locknull, this effectively deletes a resource
>   from the parent collection
>   ==> parent resource should be checked
> 
> *) using MOVE on a resource: this removes the resource from the parent
>   ==> parent resource should be checked
> 
> *) MOVE and COPY create resources at the Destination
>   ==> parent resource of the Destination should be checked
> 
> *) MKCOL on a null resource
>   ==> parent resource should be checked
> 
> *) MKCOL on a locknull resource does not add any resources to the parent
>   collection
>   ==> parent resource SHOULD NOT be checked
>   (the [locknull] resource itself will be checked, however)
> 
> *) PUT operates similar to MKCOL w.r.t. locknull resources
> 
> 
> Note that this can be an interoperability item. If one server enforces
> that conditions be met for the parent, but another doesn't, then a client
> (that doesn't provide parent conditions) could fail.
> 
> Cheers,
> -g
> 
> --
> Greg Stein, http://www.lyra.org/
> 
> 
> 
> 

-- 
Greg Stein, http://www.lyra.org/
Received on Monday, 29 May 2000 20:33:39 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:43:54 GMT