Is an option to reply in MUCs without being directly addressed being worked on?

Asked by fabulouspanda

The set of input the bot has answers to that aren't essentially 'huh?' is small enough that the bot could reply to those in a MUC without having to be addressed e.g. help rather than having to say bot help - is this in the works? Mine is configured to also act like an elizabot with basic conversational 'skills' but people won't ever use it if they have to type the bots name first.

Question information

English Edit question
Ibid Edit question
No assignee Edit question
Solved by:
Last query:
Last reply:
Revision history for this message
Stefano Rivera (stefanor) said :

I suspect that the set of input you describe is bigger than you think. Probably not with particularly useful responses, but most plugins do match a wide range of input. People can create very broad factoids, too.

We have a design principle that bots shouldn't speak unless spoken to:

13:57 <~Vhata> Spinach: rule #1 for bots
13:58 < Spinach> Vhata: rule #1 for bots is Speak Only When Spoken To

This can be done, by setting the addressed attribute on processors to False. (IIRC this is possible from the configuration file, too). But I wouldn't expect great results.

Revision history for this message
fabulouspanda (fabulouspanda) said :

Thanks, that works, although by changing just that the bot responds to itself repeatedly so I had to tell it not to respond to it's own id in groupchat - can you point me in the right direction as far as not replying when the response is 'huh?' or *blink* etc. please?

Revision history for this message
fabulouspanda (fabulouspanda) said :

Sorry don't worry about it I just removed the 'nonsense' replies in the plugin. Thanks for your help, and all your work so far!

Revision history for this message
Stefano Rivera (stefanor) said :

Yeah, it's the core.Complain processor that you'd want to have not respond to unaddressed messages.

As to replying to its own ID, that sounds like a problem specific to Jabber MUC. We should filter the bot's own messages at source level, I think. Please file a bug.

I see no such problems in IRC:
20:44 <@tumbleweed> tibid: say tibid: hi
20:44 < tibid> tibid: hi