| Commit message (Collapse) | Author | Age | Files | Lines |
| | |
|
| |
|
|
|
| |
This rename happened before we had config compatibility and it doesn't
make sense to keep it anymore.
|
| | |
|
| |
|
|
|
|
| |
This allows restricting an oper account based on other data such
as TLS fingerprint or services account but without logging them in
automatically like autologin.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
This option stores the user *mask* not the user host so the old
name is incorrect.
|
| |
|
|
|
|
| |
Now the module only implements the IRCv3 capability this is a more
accurate name. If users want the protoctl method they can install
the protoctl module from contrib as before.
|
| |
|
|
|
| |
"ERROR" is apparently defined by more than just Windows. Let's
pick a different name which is less likely to cause collisions.
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
This makes things easier for modules like require_auth in contrib
which has a GA-line.
|
| | |
|
| |
|
|
|
|
|
| |
The custom codes that InspIRCd exits with are not very useful and can confuse init systems like systemd which assume that certain
exit codes mean certain things. INSPIRCD_BINARY_EXIT was a workaround
for this in v3 but considering thatsers have to check the logs anyway
so we may as well just use EXIT_SUCCESS and EXIT_FAILURE.
|
| |
|
|
|
|
| |
- Fix invalidating invoking undefined behaviour when a file is not
already cached.
- Fix memory corruption caused by assigning a bool to a string.
|
| |
|
|
|
|
|
|
|
| |
This supports more platforms (e.g. Haiku) and is actually still
maintained unlike the former.
All of this code should really be cleaned up for release (maybe by
adding something like Anope's LOG_CONSOLE) but for now I've just
replaced it with the fmtlib equivalent.
|
| | |
|
| | |
|
| |
|
|
|
|
| |
All fields are now ordered in the most memory efficient way and all
public fields use the correct case. Every member has been redocumented
to match the current documentation style.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
This introduces the concept of a real username. This value comes
from either the initial USER message or from an ident lookup. Doing
this allows us to use it for bans through vidents and cloaking web
client users using their remote username.
While changing this I also changed all of the uses of "ident" other
than RFC 1413 lookups and some compatibility cases to refer to
usernames as user(name) instead of ident. Our use of ident in these
places was incorrect as that only refers to the RFC 1413 response
and is not commonly used in the way we used it by any other IRC
server implementations.
|
| |
|
|
|
| |
This fell back to the correct value but created a spurious message
in the server log.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |\ |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
- Move core connect class checks and <performance:clonesonconnect>
to the core_user module.
- Add pre-change and post-change events for when a connect class
changes.
- Split explicit class changing out into its own method.
- Remove the need to almost always call CheckClass after SetClass.
- Add use counting to the connect class instead of relying on the
shared_ptr use count.
|
| |\| |
|
| | | |
|
| | | |
|
| | | |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
- Allow overriding privileges from the <class> blocks in the <type>
and <oper> blocks.
- Separate oper types from oper accounts in the code. This enables
moving some core stuff out of the config tag later.
- Merge the config tags together to make a synthetic tag that can
have getXXX called on it instead of using getConfig and then
converting it.
- Move the details of Have*Permission into the oper type class.
- Improve oper events to allow modules to easily hook into the oper
system.
|