* Simplifies the export declarations for module developers. While
MAPI III utilized a flexible key-value vector to eliminate positional
arguments in a header initializer, now the developer simply makes
a list of pointers to what they want to export for injection into
IRCd. Example:
mapi::header IRCD_MODULE
{
"mymod",
"My module adds a command, a hook, and a CLICAP",
&my_cmdtab,
&some_hook,
&clicaptab
};
* Distributes the handlers for items passed to the above vector.
Anyone can add a type-handler to the module system from anywhere in IRCd
(and other modules?) When your type is encountered a handler is called
providing the symbol name to read out of the module. Example in parser.cc:
mods::add_loader<Message>([]
(mod &loading, const std::string &symbol)
{
auto &msg(get<Message>(loading, symbol));
add_command(msg.name, msg);
});
* To benefit from the precompiled-header (PCH) it MUST provide "the first C token."
Advantages:
Never worry about the include stack again. Remember, this means one less thing
for random module developers, community people learning C++, and new developers
to deal with. It should reduce the learning curve and barrier for participation.
Disadvantages:
Makes overall compilation a bit slower, especially without any additional
work to improve it again. There are several opportunities, places where the
PCH is probably being ignored, etc that can be addressed.
* librb is no longer a separately configured subproject.
* charybdis is now a standalone directory with a binary.
* Include path layout now requires a directory ircd/ rb/ etc.
channel mode classification which is required by RPL_MYINFO indicating arity,
and RPL_ISUPPORT indicating an enumerated class. The content of these replies
had previously been generated by hardcoded strings of some letters.
Channel modes require classification which corresponds to the
CHANMODES= data in RPL_ISUPPORT. Classes A,B,C can then be
listed in the unary column of RPL_MYINFO. cflag_add() is updated
for this. Additional cleanup of chmode.h and channel.h
circularity is also proffered within.
Submitted-by: Jason Volk <jason@zemos.net>
There are two important caveats here, however:
1) Aliased commands have more than 8 parameters will be truncated;
there's nothing I can do about this.
2) Parameters with colons will not be handled as you expect. Again,
nothing I can do about this.
This also lays the groundwork for the netjoin batch type, but that isn't
implemented yet. I don't like how some of this is implemented but it'll
have to do for now...
Compile tested, needs more testing.
It's a bit of a hack, but better than before. Rather than rehashing
(which could get us into an endless loop), we now segregate the
configuration phase (creating entries ircd-side in case we restart authd
later) and sending phases (when configure_authd() is called). Since we
have to call configure_authd() no matter what (to send timeouts etc.)
and we have to send this data to configure authd anyway, and sending
duplicate data is bad, this is the only way I can think of for now.
It seems to come from an era where long long didn't exist and 64-bit
machines weren't common. 32-bit machines are still common but I can't
imagine this will have much performance impact there.
This "fixes" #179 in title only, but see comments within.
There's no reason to really have these in the main ircd anymore, static
modules are dead and aren't coming back.
To ensure people don't do something hopelessly retarded, this is a core
module.
This also does a lot of surgery on the conf system to reconfigure authd.
/!\ WARNING! ACHTUNG! ADVERTENCIA! ATTENTION! AVVERTIMENTO! /!\
This code has not been run-time tested yet (though it compiles)!
now connid's are allocated on demand and clients may have as many connid's as necessary.
this allows us to build chains of helpers while ensuring the ircd properly tracks and GCs the resources.