Some people have real issues moderating their behaviour to be friendly or helpful. These are hard cases to deal with, and in the limit can result in permanent bans (some of the #ocaml crowd are haskellers in exile). The usual path is to talk to them, point them to the channel principles, then if they refuse to cooperate, exponentially increasing host bans.
Redirect ban to the op channel is good for wide-reaching bans. (e.g. tor or *ass*!*@*), if we expect the odd false positive.
This can also be done for people with problematic connections.
Controlling offtopic conversations
Never happened in #haskell, but it is possible to manage an off topic flamefest, with:
All of the offtopic conversations die off because no one can see any responses to them.
Kicks and parts
You can "remove" that forces a PART instead of a KICK. The difference being that many clients don't auto-rejoin on PART.
T +q ____ and +b %____ are identical and just silence the offender but don't prevent joins
Real name bans
Some persistent trolls will attempt to rejoin using all available means, they can be often stopped with realname bans, using +d.
I default to *!*@hostname bans, especially when I expect it to be a temporary ban or when banning an unregistered nick. Hostname specific bans against a dynamically assigned hostname should be cleared periodically.
I am much less lenient with someone that join/spams than with someone that has a history of productive behaviour who slips up. I have no problem kick/temp-banning a join/spammer while I'm likely to warn and chat with a regular user who violates the policy.
Many users aren't aware of what acceptable #haskell behavior is. We keep the channel noise level low to encourage productive, on-topic discussion. Private messages to a problem user explaining why a behaviour is not acceptable are often successful at neutralizing a situation before it escalates. Of course other users are intentionally disruptive, but even these are eligible to be saved. I will often ban the user in the channel without kicking (which mutes them) and immediately send a private message explaining the situation.
This situation hasn't really happened in #haskell before, but since I've dealt with it in other channels I'll document it here:
Sometimes the channel can become wildly off-topic with too many people to blame to point individual fingers. The most effective way I've found to deal with this problem is to +o a few of the channel moderators and to set the channel to +mz. This configures the channel such that only +o users can read the messages and respond to them. This off-topic conversation will die out and once someone asks a productive, on-topic questions you can set the mode to -mz and return to normal. -- glguy
Dealing with gateway abuse
Some people think that IRC gateways like Tor and Mibbit grant them enough anonymity that they can hassle IRC channels unchecked. If someone is reconnecting through such a gateway and proving difficult to ban, do the following (example using tor):
This will redirect all Tor users to #haskell-ops. If a legitimate user gets caught in this wide-reaching ban, you can add an exception for that specific user:
/mode #haskell +e nick!user@host