Ping again?

A colleague of mine is in the process of re-testing recurrence with 
the latest version of Kronolith and the Funambol Outlook client, I'll 
update when he has some results.

I can't reproduce anymore, because Funambol sends both the number of 
occurrences (calculated from the recurrence end) and the recurrence 
end date. Since number of occurrences has precedence now in the Horde 
code, such events are converted now to recurrence rules like: 
RRULE:FREQ=WEEKLY;COUNT=10;INTERVAL=1;BYDAY=WE



Beside that, the date stamp sent by the client is a perfectly valid 
ISO date time string. If it's not correctly translated to the local 
timezone, this has to be fixed either in the iCalendar library or in 
Kronolith, but definitely not in the Sync4j device handler.



Can you please try if you still see this with the latest code from CVS 
or the next RCs?

When synchronizing a recurring item with an end (either date or count) 
from a windows mobile PDA with the funambol 3.x plugin to kronolith, 
the last appointment in kronolith is one before the last appointment 
on the PDA.



The funambol plugin converts the enddate on the PDA to Universal Time 
(for both end date and recur count), and the date gets stored 
unmodified in the kronolith database. Because of DST, I'm now two 
hours ahead of DST, which means the enddate is set to two hours before 
midnight on the endday, so the day before. This causes the recurrence 
to end prematurely.



The attached patch converts the PatternEndDate to a timestamp using 
mktime, then adds the difference from UTC using date('Z');



I've so far only noticed this issue with the Funambol plugin. I've 
asked a colleague of mine to test this using a Nokia device, to see if 
the problem is with the import of the vcard (can't imagine so), and 
not funambol's implementation.



In the mean time, I've attached the patch for review. Please comment 
if this isn't the right way to fix this, so I can learn how to do it 
correctly within the Horde framework.