On non-blocking sockets, recv() raises BlockingIOError instead of blocking when there's no data to be read.
The correct behaviour is to wait and try again instead of breaking the connection.
- Remove "8P#" as a sidrange default - it does not leave adequate room for larger networks
- Move the TS6 example to near the end; it has less demand than Unreal and P10
- Consistently use "must" to describe P10 cloaking options, where any misconfiguration will lead to serious desyncs
- Bumped example autoconnect times from 5 to 10 seconds
- Add the 'netname' option to server blocks that previously didn't include it
- Comment out the 'channels' setting on each server example by default
- Various wording clarifications
- Extend CTCP replies to all service bots - closes#468.
- Use a generic hook handler instead of wrapping around the commands handler (#407).
This code takes advantage of the hook suppression feature introduced in 2e66b9bde6 (#547).
This behaviour really isn't consistent across IRCds though...
- Unreal, Hybrid, and charybdis mangle messages at the server side to use the lowest prefix
- InspIRCd throws a "No such nick/channel" error
- Nefarious silently drops messages with multiple prefixes?
So far, this code only knows about changing prefixes while keeping mode characters as-is.
A complete but longer solution would be to actually go through irc.cmodes, but I don't
think doing so is necessary given how little STATUSMSG is actually used in production.
- Clarify the project's goals of being an IRC services *framework*
- Briefly mention in the FAQ that the relay plugin is needed for...well, relay!
[skip ci]
This fixes various issues including:
- Networks going missing (the server list is read in a non-deterministic order)
- world.started never being set, causing relay to never work!
This is a very specific hack that shouldn't be extended across the protocol module spec. So far, all other protocol modules ignore the endburst_delay option anyways.
