Re: question on 'LAST-MODIFIED' property

[email protected] Mon, 19 Jul 2004 21:34:19 -0400
Newsgroups gmane.ietf.calendar
Message-ID <[email protected]>
First - I do not believe the intent of CAP is for true "synchronization" in 
the PDA/IntelliSync/SyncML sense. That would be more in the vein of the 
SyncML projects I believe.

Secondly, see 4.8.7.2 (DTSTAMP) in RFC2445 - I belive that this DTSTAMP, in 
conjunction perhaps with UID and/or SEQ, could be used to determine the 
update time of things on the other end.. It would be up to the client to 
maintain DTSTAMP in any local database containing iCalendar 
objects;  remember, however that storing calendar objects makes you a CS 
(Calendar Store) in addition to a CUA  and you might want/need to act 
accordingly.

Tim Hare
Interested Bystander, Non-Inc.



At 08:38 AM 7/19/04, you wrote:

>HI all,
>
>I have an implementation kind of question regarding the LAST-MODIFIED
>property. I am curious of thoughts for this property whether it
>is usefull to maintain in CAP clients (or intermediate-servers that
>act as clients).
>
>
>In our implementation of a CAP server we have added at the server side
>the LAST-MODIFIED property. The approach is; adding or updating this 
>property at the moment the component is inserted or updated in the
>calendar store. My problem is that the CAP client invoking the
>respective  action in the server will never know the value of the
>property unless the client is does invoke a SEARCH command for the
>specific UID/component.
>The CAP client can do a reasonable guess of the time only when the system 
>clock of the server and client are in sync. Not knowing the value
>of the LAST-MODIFIED makes it difficult for a CAP client that would only 
>like to retrieve updates made since the last synchronization (it does not 
>really know the LAST-MODIFIED value) and having been off-line
>(think along the lines of IMAP usage).
>
>Therefore, I was thinking that it could be useful to have a LAST-MODIFIED 
>property in the VREPLY also. If I look at examples in
>page 103/104 the UID is included.
>Maybe there are other apporaches I have missed so far, but always
>invoking a SEARCH command for a specific time and then in the client
>chekcing whether there was an update is also not attractive. In perticular 
>in clients with limited resources.
>
>
>Any ideas,
>
>Harrie

Tim Hare
Interested Bystander, Non-Inc.