[Cosmo-dev] Re: [commits-cosmo] (rletness) [4318] Serialize datetime as floating instead of throwing timezone exception for

[Randy Letness]

Brian Moseley wrote:
>
> i'm not sure that we should accept the caldav update if it sends a
> timezone we don't understand.

We would limit what we could accept via caldav if we rejected timezones 
we didn't understand.  And its not that cosmo doesn't understand the 
timezone, its that it can't serialize arbitrary timezones over eimml.

> you could make the argument that we should register and store the new
> timezone, but that would require a database-backed timezone registry,
> which we of course don't have at the moment.

Yeah but there is still the problem of how we serialize these new 
timezones for eimml.

> is it better to throw away the unknown timezone from the caldav update
> or to disallow the update altogether?

The timezone would only be thrown away if  Chandler updates the event.  
The vtimezone is stored with the event.  The change I made serializes 
dates with unknown timezones as floating in eimml.  So if caldav is the 
only protocol used there is no data loss.  The only downside is that if 
a caldav client updates an event with an unknown timezone, Chandler will 
see the event as floating because there is no way to serialize that 
timezone over eimml.

Ideally cosmo should attempt to translate the unknown timezone to a 
known one by matching the rules like Chandler does.  Currently we only 
rely on a set of known aliases.  There is an open bug for this.

-Randy