Re: [Tzdist] Fwd: [tzdist] #32 (service): managing historical data

[Cyrus Daboo <[e-mail]>]

Hi Doug,

--On December 12, 2014 at 11:22:26 AM -0700 Doug Royer 
<[e-mail]> wrote:

>> I am not proposing changing iTIP right now - even though I suspect
>> many clients will work fine without them. The goal is to get tzdist
>> deployed and in use and then eventually change iTIP.
>>
>
>      Are you saying that CALDAV compliant CUA implementations are already
> REQUIRED to ignore the VTIMEZONE included?
>
>          If YES, I see your point.
>          If NO, then your breaking them - correct?
>
>      Or are you saying that *ALL* CALDAV implementations currently ignore
> the VTIMEZONE (as in they are all broke) so the point is mute ?

This is all explained in the current draft: 
draft-ietf-tzdist-caldav-timezone-ref-00. Nothing currently requires 
clients to ignore time zones. The proposed extension will allow that in 
some circumstances.

Strictly speaking clients are not "ignoring" the time zone - they are 
matching the TZID with ones in their local cache and using the data in 
their local cache in place of the VTIMEZONE that was sent. As Mike has 
pointed out that is often a necessity since the transmitted VTIMEZONE may 
not be complete (only covers a very small range of time as applicable to 
the event it is attached to).

As the draft explains, we did a lot of interop testing and discovered that 
server could unilaterally decide to not return VTIMEZONE data in iCalendar 
objects read by the client (at least for those with "well known" TZIDs). 
Nothing broken as a result of that. The spec also makes it clear that this 
"timezones-by-reference" approach is only scoped to CalDAV interactions and 
does not apply to other protocols exchanging calendar (i.e., iTIP or iMIP).

-- 
Cyrus Daboo