The actor in this case will either be ourself (in which case users.Bot
already exists), or a services nick (which may not have a full hostmask,
and therefore cause hashing errors). Only try to add a user if we're
changing a channel mode. This may still break on chanmodes, needs more
testing in that regard.
* Fix inaccuracy in users._add docstring
Also don't track disconnected users by nick anymore. But that seems less
major than fixing the docstring.
* Fixes
* Fix docstring
Instead of trying to guess what format a token is based on its content,
hardcode a list of known tokens and what formats their parameters take.
Any unknown token will therefore be kept as a string, instead of
potentially blowing up parsing due to unexpected values (such as
multiple colons).
Also fix parsing for colon-separated lists. MAXLIST=beI:100 is *VERY*
different from MAXLIST=b:100,e:100,I:100 but our previous processing
made the two identical in terms of what the structure looked like.
Before, strong refs ensured that the user stayed "alive" but now we
remove user instances when they leave the last channel. As such, fire
the event before removing them from channels so the user instance is
still in our overall user set.
Specifically, the mode handling and the end of list modes are delayed until the end of the WHO reply from the server. When the end of the WHO reply is received, all queued operations on the channel, if any, are triggered at once in the form of relevant events.
This also does some more bits of rudimentary user handling, but most of it still remains to be done.
Note: This adds DEV_PREFIX to botconfig and LOG_PREFIX to settings (var), make sure to properly update your bot!
