[Evolution-hackers] SyncEvolution + EClient API + EXDATE regression (Bug #655253)

[Patrick Ohly <patrick ohly gmx de>]

Hello!

I have started integrating a patch by Chris Dumez which switches the
SyncEvolution Evolution API from the legacy API to the new EClient API
in EDS 3.2.

I'm currently testing it in the nightly testing, with EDS compiled from
source on Debian Unstable. Backend is the normal file backend.

I see one regression: when adding a detached recurrence with
e_cal_client_modify_object_sync(CALOBJ_MOD_THIS), an EXDATE is added to
the parent event.

While semantically correct as far as I know (the detached recurrence
overrides the EXDATE, instead of the EXDATE canceling out the detached
recurrence), it is a side effect which makes it hard to undo the
addition of the detached recurrence. Calling
e_cal_remove_object_with_mod(CALOBJ_MOD_ONLY_THIS) won't be enough, the
caller would also have to know that it has to remove the EXDATE.

It also breaks SyncEvolution's automated testing:
http://syncev.meego.com/latest/head-unstable-amd64/5-evolution/Client_Source_eds_event_LinkedItems_0_testLinkedItemsParentChild.log


This seems to come from this code:
e_cal_backend_file_modify_object()
8631a8f2 (Milan Crha           2011-09-02 13:21:07 +0200 2394)          } else if (obj_data->full_object) {
8631a8f2 (Milan Crha           2011-09-02 13:21:07 +0200 2395)                  /* add exception for the modified instance */
8631a8f2 (Milan Crha           2011-09-02 13:21:07 +0200 2396)                  e_cal_util_remove_instances (e_cal_component_get_icalcomponent (obj_data->full_object), icaltime_from_string (rid), CALOBJ_MOD_THIS);

commit 8631a8f2e0c1ca71a48aeca5a44a11506ac77e33
Author: Milan Crha <mcrha redhat com>
Date:   Fri Sep 2 13:21:07 2011 +0200

    Bug #655253 - Delete of one occurrence of a repeatable event don't work

This was added in 3.1.91.

Milan, can you shed some light on why the patch solves #655253? I fail
to see what e_cal_backend_file_modify_object() has to do with deleting
one occurrence of a repeatable event.

If the EXDATE was really necessary to avoid having the original and the
detached recurrence show up, then IMHO adding the EXDATE only works
around the real problem. The real problem is more likely to be in the
matching against RECURRENCE-ID.

-- 
Bye, Patrick Ohly
--  
Patrick Ohly gmx de
http://www.estamos.de/