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

FW: Problems with Delete of a version-controlled collection

From: Kirmse, Daniel <daniel.kirmse@sap.com>
Date: Thu, 17 Jan 2002 17:23:19 +0100
Message-ID: <59357A260E15D311B5A60008C75D3530068B476F@dbwdfx13.wdf.sap-ag.de>
To: "Ietf-Dav-Versioning (E-mail)" <ietf-dav-versioning@w3.org>

-----Original Message-----
From: Kirmse, Daniel 
Sent: Donnerstag, 17. Januar 2002 17:19
To: 'Clemm, Geoff'
Subject: RE: Problems with Delete of a version-controlled collection

>-----Original Message-----
>From: Clemm, Geoff [mailto:gclemm@rational.com]
>Sent: Mittwoch, 16. Januar 2002 17:50
>To: Ietf-Dav-Versioning (E-mail)
>Subject: RE: Problems with Delete of a version-controlled collection
>   From: Kirmse, Daniel [mailto:daniel.kirmse@sap.com]
>   suppose this:
>	   /foo
>	   |
>	   +- /bar
>		|
>		+- foobar.c
>   all the resources are under version-control. Now suppose this
>   sequence of requests:
>   CHECKOUT /foo/bar
>     DAV:apply-to-version
>   (working-collection: /wc/wc1  auto-update: /foo/bar)
>   CHECKOUT /foo
>     DAV:apply-to-version
>   (working-collection: /wc/wc2, auto-update: /foo)
>   delete folder /foo/bar:
>   DELETE /wc/wc2/bar   
>   or is it done via the real VCR? I hope not! The DELETE 
>should be hidden
>   until checkin!
>You are correct.  The DELETE should be done to the working collection,
>as you show here.
>   CHECKIN /wc/wc2
>   (applied to /foo -> the folder /bar and all its content is gone)
>   CHECKIN /wc/wc1
>   (must fail because the VCR the auto-update should go to is gone)
>That depends on the server.  A server could just delete the
>auto-update property when the VCR referenced by that property
>is deleted, and allow the CHECKIN of /wc/wc1 to succeed.
>You haven't really "lost" anything, because this version is
>now available in the version history that is associated with

So the VCR of /foo/bar is still gone (even so the VCR of /foo/bar/foobar.c)
but the version-hsitory of it would contain a new version created by the
last checkin.
Hmmm. What if the deleted VCR was the last one pointing to this
version-history? Is there a way of creating a new VCR pointing to a existing
VH? Rolling back the delete would be an option too, I think.

>   Questions:
>   1. Is this right?
>   2. If yes: Shouldn't the deletion be prevented or fail because a
>   is checked out and therefore some kind of "locked"?
>A server certainly could fail the CHECKIN of /wc/wc2 for this reason,
>but that would have to be a server-defined precondition (i.e. this
>CHECKIN is not forbidden by any of the standard preconditions).
Received on Thursday, 17 January 2002 11:24:01 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:55:48 UTC