The discussion on Red Hat Bugzilla #1178686 uncovered another old, now obsolete, assumption in SIPE code:
- backends parse login name to pass login domain to core
- core just passes login domain around, but doesn't use it
- sip-sec modules receive domain, but not all of them use it
- sip-sec-gssapi needs to re-assemble the GSSAPI username using login domain and name. In case of gssapi-only setup this isn't needed, because GSSAPI should do the parsing.
Therefore login name parsing should be pushed down to the sip-sec modules:
- backends should not touch login name
- remove login_domain from
sipe_core_allocate()
- remove
authdomain field from sipe_core_private and others
- remove domain parameter from sip-sec acquire credential interfaces
- implement username parsing in those sip-sec modules that need it: gssapi (when
HAVE_GSSAPI_ONLY is not defined), ntlm, sspi
Implemented by git commit a1b2ed3