[Ietf-caldav] Client property loss?

[Alexander Traud]

> There are several use cases that justify the statement we currently have:

Do not see a reason to point implementers on this. They figure out
themselves, when they need this (dangerous) feature. Additionally, I do not
say anything about MUST.


>> I would like to see a statement like "highly encouraging to" save keep all
>> unknown-to-client fields as far as possible here.
> Perhaps we could also mention that they need to do that if they want to
> 'hand-off' the iCal object to some other service (eg sync.).

What happens, when the server is not reachable? Let us tighten this a bit
more: Any client which (might) exchange a component with a third party
SHOULD load all properties. What about that? Yes I know, a SHOULD is rather
useless, however, better than the current situation. Perhaps the implementer
things twice about it then.