It does not happen with -d 8.

I'm unable to reproduce using KE14 with pykolab 0 7.26 and setting :

This issue still exists as per the commit revert.

thanks for the tip, i just copy pasted your code which leads to one space too much

line 355 in /usr/lib/python2.7/site-packages/pykolab/xml/event.py is not properly idented. this forces wallace to die. check and correct identing.

Just a guess, maybe i cannot see log entries because wallace is unable to start?

nothing.
changed warning to debug, did a complete reboot and i am still unable to see something in the log (from wallace)?
I get error messages to stout when do a kolabd --help (unicodeerror) or a kolab acl-cleanup (CYRUSError Mailbox does not exist) but i think they don´t belong into this discussion

no lines in /var/log/kolab/pykolab.log ?

still unable to invite resources after this. Always get Undelivered Mail Returned to Sender...
I get no log entries in pykolab.log so i have no idea where to start searching

to those, who might be interested

This "ERROR event|todo_from_ical() exception: AttributeError("'list' object has no attribute 'year'",)" and duplicated DTSTAMP in itip were fixed. The other errors I don't know.

This may have already been fixed in GIT, but does not progress in Continuous Delivery because the unit tests keep failing.

I have now tried the same test case on Kolab Winterfell.

I have just done a clean installation of Kolab 16:

I checked a little bit.

D127 together with D123 are fixing unit tests. Now the output is:

[root@kolab pykolab]# nosetests ./tests/unit
...............................Error: /home/abuild/rpmbuild/BUILD/libkolabxml-1.2/src/objectvalidation.cpp 94:   event.start() is not set
Error: /home/abuild/rpmbuild/BUILD/libkolabxml-1.2/src/xcalconversions.h 1500:   Start date is missing, but is mandatory for events
Debug: /home/abuild/rpmbuild/BUILD/libkolabxml-1.2/src/kolabformat.cpp 94:   Error occurred while writing.
.......................unnamed app(16962) KCalCore::ICalFormatImpl::populate: No PRODID property found
unnamed app(16962) KSystemTimeZonesPrivate::readConfig: readConfig(): local zone= "UTC"
unnamed app(16962) KSystemTimeZonesPrivate::readZoneTab: readZoneTab( "/usr/share/zoneinfo/zone.tab" )
.....unnamed app(16962) KCalCore::ICalFormatImpl::populate: No PRODID property found
.....(20:39:09) timezoneconverter.cpp(51):      Guessed timezone and found:  "" 
(20:39:09) commonconversion.cpp(44):    invalid timezone:  "None" , assuming floating time 
.unnamed app(16962) KCalCore::ICalFormatImpl::populate: No PRODID property found
.Error: /home/abuild/rpmbuild/BUILD/libkolabxml-1.2/src/utils.h 114:   bad numeric conversion: negative overflow
Error: /home/abuild/rpmbuild/BUILD/libkolabxml-1.2/src/utils.h 114:   bad numeric conversion: negative overflow
schema exception
:25:25 error: value '' does not match regular expression facet '(\-|\+)?\d{4}\-\d{2}\-\d{2}T\d{2}:\d{2}:\d{2}(\.\d*)?Z?'
Critical: /home/abuild/rpmbuild/BUILD/libkolabxml-1.2/src/xcalconversions.h 1940:   Failed to parse calendar!
Debug: /home/abuild/rpmbuild/BUILD/libkolabxml-1.2/src/kolabformat.cpp 94:   Error occurred while writing.
........................................................................S.
----------------------------------------------------------------------
Ran 140 tests in 2.898s

Unfortunately, simple_bind_s() is used also in authenticate() and _list_domains(). Please, check my differential very carefully. Works for me in the same scenario, but my pykolab knowledge is very limited. How can I test code paths with authenticate()?

It would be, provided the _bind() call does not cause the connection to use credentials that the connection had not intended to use (i.e. from user John to service account or somesuch).

OK, this seems entirely reasonable.

In D81#1467, @machniak wrote:

kolab_auth options are used since rRPKcbd97e52cde

kolab_auth options are used since rRPKcbd97e52cde

Hmm, the only way I seem to be able to pop this off my todo list is to require changes :/

These settings being obsolete means... what exactly?