W3C home > Mailing lists > Public > ietf-dav-versioning@w3.org > July to September 2002

Re: UPDATE semantics for checked-out resources

From: Stefan Eissing <stefan.eissing@greenbytes.de>
Date: Fri, 27 Sep 2002 16:25:42 +0200
Cc: ietf-dav-versioning@w3.org
To: "Clemm, Geoff" <gclemm@rational.com>
Message-Id: <00B3BE65-D225-11D6-9660-00039384827E@greenbytes.de>


Am Freitag, 27.09.02, um 16:13 Uhr (Europe/Berlin) schrieb Clemm, Geoff:

> Apologies, I misread the original message (i.e. the issue is the
> checkout of a member of the collection, not a checkout of the
> collection).
>
> This is just one way in which you can delete a checked-out VCR, and
> currently, deleting a checked-out VCR is allowed by the protocol.

I see where this is going...

I think DELETE on a checked-out VCR must be ok. However if I UPDATE
a collection and someone else checked out the contained VCR and
that VCR is silently deleted, all his/her changes are lost.

Which can happen anyway, unless she/he locks the VCR.

Therefore I change my mind and from now on to forver say that
explicit and implicit DELETE on checked-out resources are "OK".

//Stefan

> Stefan: Were you suggesting that it be disallowed just in this case,
> or disallowed in general (e.g. add it as a precondition to the DELETE
> method).
>
> For now, I'll just add this to the Errata document as an open issue,
> since disallowing deletion of a checked-out VCR would be a change
> to the protocol semantics.
>
> So, to get some initial feedback, who thinks we should disallow the
> deletion of a checked-out VCR?
>
> Cheers,
> Geoff
>
> -----Original Message-----
> From: Manfred Baedke [mailto:manfred.baedke@greenbytes.de]
> Sent: Friday, September 27, 2002 9:21 AM
> To: Clemm, Geoff; ietf-dav-versioning@w3.org
> Subject: AW: UPDATE semantics for checked-out resources
>
>
> This is true, but it does not apply to the more general case of an 
> UPDATE
> of a version.controlled collection containing a checked-out resource 
> which is not identified
> by the DAV:version-controlled-binding-set of the update source.
>
> Cheers, Manfred
> -----Ursprüngliche Nachricht-----
> Von: ietf-dav-versioning-request@w3.org 
> [mailto:ietf-dav-versioning-request@w3.org]Im Auftrag von Clemm, Geoff 
> T
> Gesendet: Freitag, 27. September 2002 14:00
> An: ietf-dav-versioning@w3.org
> Betreff: RE: UPDATE semantics for checked-out resources
>
>
> I agree with your conclusion, but I believe this follows from
> the DAV:no-overwrite-by-auto-update precondition for CHECKIN, i.e.:
>  If the DAV:auto-update property for the checked-out resource
>  identifies a version-controlled resource, at least one of the
>  versions identified by the DAV:predecessor-set property of the
>  checked-out resource MUST identify a version that is either the same
>  as or a descendant of the version identified by the DAV:checked-in
>  property of that version-controlled resource.
> If the VCR is checked-out, there is no DAV:checked-in version,
> which means this precondition would not be satisfied.
> Cheers,
> Geoff
>
>
> -----Original Message-----
> From: Stefan Eissing [mailto:stefan.eissing@greenbytes.de]
> Sent: Friday, September 27, 2002 4:48 AM
> To: Clemm, Geoff
> Cc: ietf-dav-versioning@w3.org
> Subject: Re: UPDATE semantics for checked-out resources
>
>
> While we're at this topic: we have a similar issue with auto-update of
> version controlled collections.
> - checkout a versioned collection with apply-to-version
> - remove a member from the working collection
> - checkout in-place the member of the versioned controlled collection
> - checkin the working collection
> -> the version controlled collection should be updated an remove
>    the binding to the checked-out resource.
> I think the checkin should fail in this case, as the removal of
> a checked-out member might cannot be permitted.
> Do you agree?
>
Received on Friday, 27 September 2002 10:26:37 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 13:57:43 GMT