parameter values.
Created js_timeval() function for converting numeric jsargs to struct timeval.

flaw... on Linux systems, it probobly never worked anyways.

This should fix the "Sync mysteriously died" after a slew of signal 13s
received on FreeBSD.

H:Line list for any QWK leafs connecting.

the C:Line would always cause it to match, thus never auto-connecting.  Added
an option to searchbyserver() to tell it to ignore wildcards.

* the default scope of the executed module is now the current scope (this)
  rather than the global object - solves the problem with redefining const vars
  if load()ing same include module more than once in a context. This will
  likely defeat modules (like S3 Shell) that use this old behavior to create
  session-persistent objects (another technique will need to be utilized).
* an optional scope object parameter can be specified as the first parameter
  specifying the context in which to execute the module (default = this)

global socket_select() and Socket.poll() methods.

accessing the high-resolution system clock.

* new bool write argument (defaults to false - for readability checking)
* timeout is specified in seconds (with fractions supported) ala Socket.poll()
* socket array can be array of socket descriptors or Socket objects
* arguments can be given in any order and all but the socket array are optional
* updated method description for documentation

(gethostname) work on Win32.

(This will change soon - don't use this ver.)

This interface is scheduled to be changes "Real Soon Now"

sockselect(array of socket descriptors, timeout in usecs)

  was set /AWAY.
* The wrong cmd[] arg was being checked on a network /WHOIS.

* Drastically changed behaviour of 'this.quit()' to always imply broadcasting,
  and to fix how it handles netsplits.
* A server wasn't broadcasting SQUIT in some conditions.
* Opers were being restricted to 'max_who' responses from /WHO.
* Reduced the /WHOWAS and nick change history, 10000 was just way too big.

* Allow servers to dynamically connect via a network PASS command for QWK auth.
* Introduced new N:Line flags because of this: 'w' defines the local QWK-master
  for forwarding PASS messages to, and 'k' is used for defining which N:Lines
  should have their passwords checked by the QWK-master uplink.
* WHO now supports a Bahamut-style complex searching syntax.  If the user
  doesn't use the syntax, we do more intelligent searching based on the
  information passed to us.
* Allow opers to see all channels that a user is on via WHOIS.
--- Thus, ircd.js is now over 150k in size, 5000+ lines of code.

filename for a specific instance, in one common place.

clients, servers, bots, or other neat stuff.  This is very much still in its
infancy.  Commited as per DigitalMan's sword. ;p

  N:Line flag.
* Fix SQUIT and netsplits so that multiple SQUIT commands aren't propagated
  across the network.  Netsplits now show the *actual* servers being split.
* Only process SQUIT commands coming from the appropriate *uplink* server,
  otherwise pass them forwards for processing.
* A user leaving the network via QUIT wasn't being propagated all the time.
* Cleaned up /STATS u as per DigitalMan's recommendations.
* Don't allow opers without the correct O:Line flags to /SQUIT remote servers,
  among other things.
* Deny incoming NICK commands from any server which has a server mismatch,
  and send out a KILL message for that nick.
* PINGs and PONGs weren't being propagated (as per /SPING in many clients.)
* Fixed client_add and client_remove calls, at the expense of a Synchronet
  port to the C64 and VIC-20.  Sorry guys, we tried :(

other callback functions).

other callback functions).

read/writable (when available).

with bbs object (alias left for backwards compatibility).

* Slice off the : on a multi-server PING.
* Support /TRACE, finally!  What a pain of a command this was to implement,
  I assure you.  We take the Bahamut approach and never allow dumping of whole
  user lists.  This was the last unimplemented command.
* Added all of the last remaining unimplemented numerics.
--- As of now, ALL the numerics as defined in RFC1459 have been implemented
    one way or another in the Synchronet IRC daemon!  Furthermore, the command
    set as defined in RFC1459 is also complete, there no longer exist any
    unimplemented commands.

Services accept system shutdown control messages now.