On 02/10/16 14:39, Patrick Ohly wrote:
On Sun, 2016-10-02 at 14:21 +0100, Graham Cobb wrote:
> activesyncd now works without crashing. But it is being sent empty data
> by the server. So some more work to do.
Please share your progress. I consider this a 1.5.2 release blocker.
Still working on it.
I am seeing a few different issues but having some difficulty working
out how to reproduce them.
The one I can most easily reproduce at the moment is a crash in
activesyncd when receiving some specific calendar data. The crash is in
icalproperty_set_value, called from
eas_cal_info_translator_parse_response handling unmapped data fields
(i.e. trying to create a X-MEEGO-ACTIVESYNCD- custom property).
I have not yet had time to build debug versions of all the code to step
through this. I will have another go tomorrow if I get a chance.
I have also seen a problem with sync keys: I have managed to get into a
mode where EAS is rejecting the sync key but we are not managing to
start again with a slow sync. I need to reproduce this again.
The "empty data" problem I saw earlier may or may not be real. Again I
need to try to reproduce it again.
I am beginning to suspect that these latter 2 problems may be affected
by the fact that my first tests were using "one-way-from-remote" syncs
(as I wanted to keep the first tests simple). I am beginning to wonder
if that mode prevents recovering from cases where activesyncd and
Exchange end up out of sync with sync keys. But this is just
speculation until I look into them further. I am concentrating on the
icalproperty crash for now.