privilege: store roster cache in database: - rosters are now stored on database and restored on startup. This way, presence map can be restored without the need to wait for all contact to send presence again - roster version are checked, if a new version is received, presence map is updated accordingly - roster are not retrieved if presence are received in a too short delay (see ROSTER_TTL), to avoid using too much resources if a client connect/disconnect a lot The current behaviour works around XEP-0356 limitations. An update of the XEP will be needed to get roster pushes and roster version.

syntax: glob
*.pyc
*.pyv
*.swp
*.swo
tags
twistd.log
twistd.pid
bridge_constructor/generated
_trial_temp/
sat.egg-info
*.un~
dist
MANIFEST
build