'finnw_pre-broken-dvorak'.

Changed the way scripts work slightly, so that they will quite happily run
off the talker - although output from SAY etc.. are ignored, and WHEREAMI
always returns "-1" - that sort of thing.

---

Changed checkonoff's to use this ability, so you can receive checkonoff
codes while on the board (ie, if you leave the talker, you will get your
own checkonoff code). Normal scripts of course, still cant be run from
the BBS, and IPC/RPC messages are currently ignored if off the talker.

---

Radically, completely rewrote the entire checkonoff system to give more
information, and a much more logical code system. Scripts executed using
this method now recieve certain arguments:

$0 = "CheckOnOff"
$1 = <code>
$2 = <method>
$3 = <is_quiet>
$4 = <sourceuser>
$5 = <version>


Where <code> can be one of:
   0 - leave the talker
   1 - enter the talker
   2 - leave the board
   3 - enter the board

This keeps backwards compatibility the same - users can still detect each
of these settings, although the extra features - such as detecting zods -
will never occur using this code alone


For each of the <code> values, <method> has several meanings:

for leaving the board (2):
   0 - normal (by user)
   1 - timeout
   2 - connection dropped
M  3 - get mrodded
S  4 - get banned
S  5 - crash / error
S  6 - deleted


for leaving the talker (0):
   0 - normal (by user)
Z  1 - zodded
S  2 - chatmode -c
   3 - straight to shell (!quit etc...)


for joining the board (3):
   0 - normal (by user)
   1 - straight to talker (-autochat)


for joining the talker (1):
   0 - normal (by user)
U  1 - summoned
S  2 - chatmode +c


The extra letters in front of the <method> indicate certain privs are
needed:

   M - needs mrod priv to detect
   S - needs SU to detect
   U - needs summon to detect
   Z - needs zod to detect


So, for example, if a user were to log on using "-autochat", two messages
would be created:

<code=3> <method=1>
<code=1> <method=0>


These are the main two values that most people will use - along with
"USER <variable" to get the user that caused the checkonoff message.


Added to these two values, <is_quiet> simply returns:
   0 - user logged on normally
   1 - user logged on with -quiet to supress messages


This allows script writers to only display their own messages if a user
is logged on quietly if they so wish.


<sourceuser> is more tricky. This is the person that caused the initiation
of the checkonoff code in the first place. ie, for most messages, it is
the same as the user the broadcast came from. However, in cases such as
zod, mrod, or chatmode +/-c alteration, <sourceuser> will be the person
that performed that operation.

ie, SUPERUSER bans USER
    <code=0> <method=3>
    <code=2> <method=4> <user=USER> <sourceuser=SUPERUSER>


and finally, <version> is mainly for being able to work out which version
of checkonoff is used. this ranges from "1" to "3" currently, but if you
wish, you can simply check if any of the arguments is left blank "" and
work out what needs to be done from there.

---

Added a "force_status" to incoming.c, and moved the 'ban' and 'delete'
detection code to there instead of in the main loop - necessary for the
new checkonoff information.

---

Fixed a bug with 'ExecEvent2' where it only accepted up to 3 arguments

---

Removed the now-obsolete "zodded" script


Chris

Force and Shutdown functions have to be bound now using:

force <function>
shutdown <function>

like checkonoff's

Chris

Corrected error in help for msg and postinfo that said they were the
same command - they're not.

More tasks

Chris

strdup().  This fixes another phantom memory leak.

the wrappers.  Changed dupstr() to use this function instead of malloc()
so that readline() can safely free these blocks.

Okay, all of the snprintf's have been updated again to remove all of the
extra null characters added. However, each of them have now been checked to
make sure the lengths are okay, so that no overruns can occur.

Chris

The wrapper functions were complaining when script variables set by
the INPUT command were destroyed.

Corrected spelling of "receive" (was spelt "recieve" eight times)

Updated the who lists (!who, .who, mw -who) to detect screen width and fill
the correct width with ---------- characters, rather than always 80. This
means that screens less than 80 characters wide are easier to read.

Modified the protection level stuff to show "?" instead of "9" if a user
has too low a protection level to see other peoples values, as this is more
intuitive to show you dont know what level it is.

Chris

Fixed memory leak in zod/mrod messages if the entire reason was spaces:
  ie, ".zod <user>          "


Added command ".linewrap <on|off>" to enable/disable the milliways talker
linewrapping, since several people prefer it without, as it splits up pastes
such as web addresses. *LINEWRAPPING ENABLED BY DEFAULT TO KEEP MW THE SAME*


Chris

Fixed a mistake I made in expand.c using snprintf

Chris

Realised that 'snprintf' and 'vsnprintf' dont terminate a string with a \0
if it reaches the end of the buffer. Had to make quite a few corrections to
fix this, but scripts seem quite a bit more stable now!

Chris

Found a fault with the STRISTR script command, and fixed it

Chris

Fixed a problem with -c chatmode, where the "Sulking" message wasnt
displayed, and where I accidently changed the users chatmodes inside a
chatmode changing function (doh! :/)

Chris

Removed 'talkmode' variable, since this is an exact duplicate of querying
the 'CM_ONCHAT' chatmode flag, and it wasnt consistently updated in a few
places.

Due to the changes above, fixed a problem where a user though they were
still on the BBS prompt, but couldnt log onto the talker. This was caused
when a superuser did a "!user chatmode <user>" and added "+c".

Also fixed a similar issue in scripts, where some scripts could be run
from the BBS when they shouldnt be!

Added new checkonoff broadcasts "-6" and "4" to show chatmode "-c" and "+c"
being done to a user. Before, a user just got told "log off" and "log on"
but it was put forward that these should be different. Easy to change back
if it is deemed completely unuseful.

Neatened up the 'new user' code where it queries if a user is a member of
sucs, for autoregistration.

Removed some unused 'extern'd variables at the start of a few C files.

Chris

Removed the reference to me that Dez keeps fscking adding in again!

Corrected a typo in help for "script"

Removed some dumb comments about me.

Updated the SAFE_FREE macro in bb.h with finnw's suggestion of using
do/while to avoid errors using the macro

Chris

Problems with tab-completion fixed (".unbind <tab>" and ",<tab>" used the
same function in different ways causing ".unbind" to break).

Added extra tab completion for ".raw", ".e" and ".emote"

Chris

Fixed another memory leak in the room changing code. I cant find any more
leaks other than the "apparent leak" caused by using 'strdup' to return
strings to the readline tab completion functions. Since readline frees this
memory, the wrappers cant detect it and report a leak.

Okay for a release now I think

Chris

Fixed another couple of memory leaks - one in .restart, and a couple
in alias.c and rooms.c

Chris

Corrected spelling mistake in script command descriptions.

Added help for isedall and sedall.

Oops, I really oughtn't to check in Makefile with wrappers enabled

Chris

Found two more memory leaks in tab-completion code (memory from 'FindLinks'
was never freed)

Chris

Fixed a small error that allowed checkonoffs to be intercepted while on the
BBS and not the talker, resulting in initialisation problems

Chris

Quiet information now given on board logon messages (checkonoff code 3)

Chris

Written a second "ExecEvent" function (ExecEvent2) that now takes an array
of arguments instead of just the one bit of text. The preamble still works
on argument 1 as normal.

Changed 'CheckOnOff' to use this new event function. There is now an
optional second argument to checkonoff events, that indicates whether the
user is logged on silently or not.

NOTE: this will cause any current users to experience a slightly different
      checkonoff argument format - namely "code[,quiet]". eg:
      "1,1" if the user logged on to the talker quietly.

Forgot to mention it in the last checkin, but summons now ignore multiple
similar arguments. eg ".summon 0 test1 test2 test2 test2" will only summon
the users "test1" and "test2" once each. Before, it used to repeat all of
the "test2" users...

Chris

Fixed the new random function so that it returns a value (min,max) inclusive,
rather than (min,max-1).

Chris

Reworded the 'split' script help file since it implied the first argument
was a variable, or could be a variable, instead of being a value input
(ie, "hello", or $test etc..)

Chris

Corrected error in file split

Okay, thats the last of the current bugs out of the way!

Chris

Moved ".alias" and ".unalias" to "!alias" and "!unalias" since all commands
are affected by aliases, and not just talker commands.

Fixed a problem with displaying "---more---" in the wrong place in aliases
in the help (!help etc..)

Chris

Updated command help (!help) to take priveledges into account on commands.
The 'wizhelp' directory is now used for more help on the same message.

For example, "emote", and "wiz" would go into the "help" directory since
these rely on priveledges alone. However, "ls" will probably have an extra
file also present in the "wizhelp" directory to explain about being able
to read files you dont have permissions for etc... - there are quite a few
commands like this!

Chris

Updated todo list

Chris

Updated the 'special' help to include the 'o' flag (for blocking use
of 'contact').

Added 'emote' command

Chris

Updated the really old 'user' help

Chris