[Date Prev][Date Next] [Chronological] [Thread] [Top]

RE: lastmod modif* attributes mishandled? (ITS#168)

perhaps a simple "netscape_broken_modify_behaviour" option in the
configuration file to keep people who have broken clients and servers to
work with?

Roy Hooper
Sr. Systems Administrator
Network Operations
Cyberus Online Inc.
(613) 233-0068

> -----Original Message-----
> From: owner-openldap-bugs@OpenLDAP.org
> [mailto:owner-openldap-bugs@OpenLDAP.org]On Behalf Of Kurt Zeilenga
> Sent: Monday, June 21, 1999 5:44 PM
> To: ajf_nylorac@acm.org
> Cc: openldap-bugs@OpenLDAP.org
> Subject: Re: lastmod modif* attributes mishandled? (ITS#168)
> Appologies for the delay in responding...
> >While configuring OpenLDAP-1.2.1 to support roaming by Netscape
> >Communicator 4.5, I determined that OpenLDAP does not provide
> >modifiersname and/or modifytimestamp for newly created entries.
> The server should NOT update modif* until it's been modified.
> >The failure to return a value for modifytimestamp causes Nu^Htscr^Hape to
> >fail to update the roaming entries after it initially creates them.
> This is their bug.  A client should NOT rely on getting these if
> if explicit asked for.  They may be ACL'ed away.
> >I have configured "lastmod" as "on" and the "creat*" operational
> attributes
> >are automatically added when an entry is created.  But, when Netscape
> >requests the value of modifytimestamp (which was not automatically added
> >when the entry was created), it is not available and not spoofed.
> Again, as designed.
> >My quick fix was to patch ldap/servers/slapd/add.c to create a
> >"modifiersname" attribute whenever a "creatorsname" attribute is created
> >and to create a "modifytimestamp" attribute whenever a "createtimestamp"
> >attribute is created.
> This is actually incorrect behavior.
> >My patch does not handle records previously created.  The better fix is
> >probably to make OpenLDAP return the value (if any) of the corresponding
> >"creator*" attribute whenever a request of a non-existant "modif*"
> >attribute is encountered.  But, I was unable to locate the appropriate
> >code quickly.
> I believe the client should not expect a "modif*" attribute to
> be available upon create.