RE: Binding loops and PROPFIND clarification needed (was Re: COPY and bindings)

> From: w3c-dist-auth-request@w3.org
> [mailto:w3c-dist-auth-request@w3.org]On Behalf Of Jason Crawford
> Sent: Tuesday, August 12, 2003 7:07 AM
> To: nnw3c-dist-auth___at___w3.org@smallcue.com
> Subject: RE: Binding loops and PROPFIND clarification needed (was Re:
> COPY and bindings)
>
>
>
> > I think the granularity of <process-208> is too fine.
> > I'd prefer to just allow the use of the DAV header as a request header,
> > and if the client says:
> >
> > PROPFIND
> > DAV: 1,2,bind
> >
> > and then the server can use 208 responses (i.e. because "bind" appears
> > in the request DAV header).
> This sounds good to me.

I still don't like it, primarily because this is *too* generic. It will
probably cause clients to send this header with each and every request, even
when it doesn't make a difference. It also opens the door to new
interoperability problems -- what if the set of extensions keeps growing and
different extensions create subtile incompatibilities. It may become really
painful to the server to decide based on the request header what to do.

I really really prefer to be *specific*. The problem *only* occurs with bind
loops and PROPFIND/depth infinity. Let us only do what's needed to fix that,
instead of inventing a whole new negotiation mechanism. In this case, just
marshall the flag as part of the request body.

> If we start having trouble managing all those values, we can explore
another
> approach.  In the meantime, the binding spec will need to define this DAV:
bind

But then it's too late, isn't it?

> header value and indicate what servers can do with a PROPFIND in it's
> absense.

Independantly of how we send the information: if the client sends a
PROPFIND/depth infinity to a collection that contains bind cycles (thus
requires the new format), just reply with 506.

Julian

Received on Tuesday, 12 August 2003 03:02:37 UTC