RE: Comments on 06 spec

The word for "publish" and "unpublish" is transactioning. The TIP group has
a nifty protocol which does this and I even wrote a spec for them on how to
use it with HTTP. It will provide you with what you require.

		Yaron


> -----Original Message-----
> From:	Dylan Barrell [SMTP:dbarrell@opentext.ch]
> Sent:	Thursday, January 29, 1998 1:09 PM
> To:	Yaron Goland; 'Jim Davis'; Fisher Mark
> Cc:	w3c-dist-auth@w3.org
> Subject:	RE: Comments on 06 spec
> 
> It doesn't deal with the case where a whole subtree should be deleted
> where one or more resources are already locked (because DELETE is not
> atomic and can be partially successful).
> 
> I think we should be explicitly optimistic in the spec as it will cause
> little problems in practice because if the user noticed the inconsistency
> and performed a reload it is very likely that the resource being accessed
> will have been deleted, moved, copied too.
> 
> I originally made a reference to this problem about six months ago when I
> asked for publishing methods to be included in WebDAV. This would allow
> the principle manipulating the namespace to "unpublish" the effected
> namespaces (an atomic operation), perform the operation and the "publish"
> the results (another atomic operation). Nobody took much notice of this at
> the time.
> 
> I suppose that your server could perform implicit "publish" and
> "unpublish" operations when the namespace is manipulated (although this is
> likely to lead to performance problems)
> 
> Cheers
> Dylan
> 
> -----Original Message-----
> From:	Yaron Goland [SMTP:yarong@microsoft.com]
> Sent:	Tuesday, January 27, 1998 2:13 AM
> To:	'Jim Davis'; Fisher Mark
> Cc:	w3c-dist-auth@w3.org
> Subject:	RE: Comments on 06 spec
> 
> Hold it, the spec does not state that GETs are unaffected by locks. It
> states that GETs are unaffected by WRITE locks. This is only one kind of
> lock. I know that a read lock spec will be released in the near future
> (mostly because I have to write it). Additionally our syntax allows for
> one
> to request multiple lock types simultaneous so one could, for example, ask
> for a read/write exclusive lock. This would create the sort of atomicity
> that has been asked for.
> 	Yaron
> 
> > -----Original Message-----
> > From:	Jim Davis [SMTP:jdavis@parc.xerox.com]
> > Sent:	Monday, January 26, 1998 10:42 AM
> > To:	Fisher Mark
> > Cc:	w3c-dist-auth@w3.org
> > Subject:	RE: Comments on 06 spec
> > 
> > At 09:36 AM 1/26/98 PST, Fisher Mark wrote:
> > >
> > >Maybe I am assuming too much, but if I was a user of a

Received on Thursday, 29 January 1998 17:12:40 UTC