[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Alarms - was RE: Question on calstore current time.
Just a thought along these lines:
First - how do vendors CURRENTLY handle alarms - i.e. can the vendors here
each speak up and explain how they handle it today... interoperability etc
will be greatly eased if we set a standard that is not too distant from
Second - fundamentally what are the REQUIREMENTS around alarms? Are ALL
alarms a user-centric condition - i.e. each attendee for a given event might
want DIFFERENT alarm conditions (including no alarm at all) - if this is the
case - do they belong in CAP at all?
[mailto:owner-ietf-calendar@xxxxxxxxxxxx]On Behalf Of Doug Royer
Sent: Wednesday, December 19, 2001 3:12 PM
Subject: Re: Question on calstore current time.
John Stracke wrote:
> >What if I want it to go to the CUA that I am connected with?
> >And not a specific CUA? As in if ANY CUA is connected, then
> >the CUA will send the alarm, else the CS sends it.
> I think there are problems with that.
> The big one: Suppose the clocks on the CUA and the CS are not in
> sync--say, the CS's clock is 1 minute ahead of the CUA's. I have an alarm
> set for 2:00. At 1:59:30, CS time (1:58:30, CUA time), the CUA connects.
> At 2:00:00, CS time (1:59:00, CUA time), the CS decides it doesn't need to
> send an alarm. At 2:00:10, CS time (1:59:10, CUA time), the CUA
> disconnects and exits. At 2:00:00, CUA time, the CUA is not running, so
> it doesn't send an alarm. The alarm is lost.
Yep. The same problem exists if several CUA's are connected and
you only want one to send the alarm.
> In addition, there are a couple of scenarios involving CUAs that don't fit
> the mold of a long-lived app with a continual connection to the CS. For
> example, you might have a daemon syncing your calendar down to a desktop
> PIM; it's got no UI at all, and shouldn't be sending alarms. Or you might
> have a Web-based app that makes transient connections to the CS; the only
> way for it to handle alarms would be if it queried the CS for them on
> every single page view.
We need MUCH more work on the alarms. I don't think this should
slow down or stop CAP. But it is unresolved.