Age | Commit message (Collapse) | Author |
|
|
|
Thanks to @KoraggKnightWolf for reporting this.
|
|
|
|
|
|
|
|
|
|
This requires all messages to be UTF-8. If you are using a UTF-8
incompatible text encoding then you can disable this by setting
<websocket:sendastext> to false.
|
|
This makes more sense for WebSocket clients than CR+LF termination
as the protocol already understands the concept of messages.
|
|
|
|
- Remove the deprecated config tag checker. This checker is neither
exhaustive nor is it particularly accurate. Instead the new docs
site will contain a page detailing the breaking changes between
releases.
- Remove the insulting <die> tags in inspircd.conf.example and add
some descriptive ones in files that it is critical for the user
to edit correctly.
- Show the message from the <die> tag so the user actually knows
what they did wrong.
|
|
|
|
- Remove the CHANMODES, USERMODES, and PREFIX tokens in the 1205
protocol. These have entirely been superceded by the CHANMODES
and USERMODES CAPAB commands.
- Only compare the 1202 PREFIX tokens if the remote server did not
send CAPAB CHANMODES. This replicates the existing behaviour
used for the CHANMODES token.
- Fix checking whether the remote server has an appropriate case
mapping. If it is not sent we assume it is the same as the local
server as always. This should prevent issues with people using
services packages that do not send this token yet.
- Fix checking if the user modes are mismatched and then promptly
overwriting the error message with mismatched channel modes.
- Fix servers not being able to tell whether a mode on the remote
server is a prefixless prefix mode. Requires the 1205 protocol.
- Fix servers that receive CAPAB CHANMODES or CAPAB USERMODES not
checking the type of those modes. Requires the 1205 protocol.
|
|
|
|
|
|
The appropriate numeric response when a user has the censor mode
enabled is ERR_CANTSENDTOUSER not ERR_CANTSENDTOUSER.
|
|
|
|
|
|
Fixes #1281.
|
|
Fixes ircdocs/irc-defs#37.
|
|
|
|
The require_compiler("GCC") directive is required to work around
some GCCs ignoring the GCC pragma for disabling warnings.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Closes #331.
|
|
This fixes a regression introduced in 97a1d642.
|
|
|
|
|
|
Closes #1433.
|
|
This has the format a:nick!user@host+gecos.
e.g. a:*!*@abusive.host+*MegaBot?v2.0*
Closes #1023.
Closes #1026.
|
|
Closes #1202.
|
|
|
|
|
|
Closes #1427.
|
|
|
|
|
|
|
|
- Allow the ident for ident type hosts to be set in the config.
- Consistently check clone counts, connect classes, and xlines for
all users regardless of their cgiirc host type.
- Remove some obsolete code from when gateway clients had their IP
changed in OnCheckReady.
- Don't allow a gateway client to match both a WebIRC and an ident
host.
|
|
- Get rid of <host> and merge the prefix and suffix values into the
<hostchange> value.
- Remove the 'suffix' action. This has been replaced by the 'set'
action.
- Add the 'addaccount' action. This adds the users account to their
hostname.
- Use the <hostname:charmap> characters when removing invalid chars
in addnick mode.
- Allow multiple entries on each hostmask so that they can fall
through if a certain value is not acceptable.
Closes #816.
Closes #1001.
|
|
This restores previous behaviour which was lost when the original
line parameter was removed.
|
|
Co-authored-by: Attila Molnar <attilamolnar@hush.com>
|
|
The library code used by this module is licensed under a non-free
license which is incompatible with the GPLv2. Combined with the
fact that it has been superceded by better algorithms like bcrypt
I have decided to remove it.
An alternate implementation of this algorithm is provided by the
m_hash_gnutls module if people *really* need to use it.
|
|
Closes #1305.
|
|
|
|
- Switch to a module-specific snotice character (f).
- Redo the format of the message to be easier to read and include
the filter that the user matched.
Closes #490.
Closes #841.
|