New release version 2.6

within a release.

New version string display on startup.

script command 'version' to retrieve the new string into a variable,
format  'D 2 5 0 arthur 20000620'

D for devel version, R for public release
2 = major version number
5 = minor version number
0 = revision number or 'tweak'
arthur = user that compiled this build
20000620 = compile date in standard YYYYMMDD format

More 'devel' flag conversions

Fry

Did the other devel function

script function: scr_devel_msg(struct code *pc, char *fmt, ...)
    This allows a script function to print an error text, and it
    automatically adds where the problem occured, and the instruction run.

normal function: devel_msg(char *func, char *fmt, ...)
    Allows a function to print a function name, and an error text, and it
    gets output in the correct format.

Also, both of these functions handle the developer flag, so you just need to
call the relevant function.

Fry

Added 'devel_message' function for scripts, to allow users to print
developer messages to screen. This is intended for non-fatal warnings, eg,
as I've already done, a warning when you use 'jump', and cant find the
function. These messages only show if you have the Developer chatpriv (D).

The point of this is just for developers to know simple problems, and it
isnt confined to scripts. For example, the 'invalid incoming message type'
message in incoming.c is now Devloper only, so normal users dont get
spammed.

I plan to convert all non-fatal (or non-useful) warnings to developer mode,
and will be writing a function for that too.

Fry

Fixed strmid. if end pos < start pos, now returns '', rather than something
fairly random.

Fry

Oops. left in a debug text, sorry

Fry

Fixed a hell of a lot of memory overflow bugs. Thanks to rohan for pointing
out stuff :)

Fixed some other bugs in mrod/zod reasons.

'args' in DoCommand (Parse.c) now works all the time, rather than the
extremely random segfaults that people might have got when typing in a
command (only happened to me once since I changed the format of 'wiz',
'whisper', 'tell' etc.. to allow spaces/quotes - might have happened to
other people more?)

Fry

Updated the duck/bunny gag _slightly_.

Fixed notsayto to work with different rooms/globals - like normal text

Fry

Fixed, and fairly radically updated all the ignorelist stuff. Now quicker,
with less memory leaks (surprised they didnt show up in -exitinfo - must
check this), and doesnt need to store user PID anymore.

Username detection also now works (ie, doesnt allow '.ignore fry',
'.unignore f'), and where possible (ie, user detected online), displays the
username in correct capitalisations :)

Fry

Fixed 'zod', and 'mrod' commands. People can now correctly specify reasons.

Fixed 'display_message' somewhat - still got a slightly broken linewrap :(

Talkpriv script command now returns 'B' too to allow users to see whether
they need to 'beep' or not for alternative text

Fry

Fixed some segv's in zod/mrod/kick/remove, made 'reasons' work better in
them too, and now prints messages to board if you zod yourself. I forgot to
take this check out when I fixed a bug elsewhere :)

Fry

probably what was causing the segv with strmid.  Also added an extra
sanity check in the hashtable iterators.

Made 'invalid variable name' errors appear even with debugging turned
off.  If you get it your script is definitely broken, probably building
a variable name with strcat which contains reserved characters such as
"$".

Added code to prevent malloc(0) when building the room link arrays.
This made efence abort if you entered an undefined room.  The count is
now set to 1 if there are no exits in a room.

Added the -D__NO_STRING_INLINES flag when debugging, to make it easier
to track down segv's in string operations.

Fixed just a _few_ bugs 'n' things that made mw go bump in the night :)

ie, it now works again :)

NOTE: Threads are totally shafted now, as it seems hash tables, and
variables
completely break threads. This is good, because the shutdown function
sometimes still breaks threads, and now any JUMP's in scripts will shaft
threads too. Oh well. perhaps radically update it later? :)

Fry

in the heap).

Make comparison counters 'local' now, so a function jump within a funciton
doesnt change the comparison counters. eg:

function a
    IFCLEAR
    EQ "a" "b"
    JUMP b
    IFALL c
    PRINT "match"
    RETURN
  d:
    PRINT "fail"
    RETURN
endfunc

function b
    IFCLEAR
    RETURN
endfunc

If function 'a' is run, returns the text 'match'. If it didnt work, it would
return 'fail', of course :)

Fry

Commented the 'set_defaults' routine. Was gonna change it, but couldnt be
bothered :)

Fry

Fixed print/nprint - no beeps now

Fry

Ooops. Should have CVS'd this a while ago. no idea what I changed :)

Fry

debugging enabled and the variable was not found.

variables, as a string.  This list is compressed using the same
algorithm as the unix 'sq' filter, and separated by spaces.  There's
no built-in decoding yet.  I may add a variant of 'split' that does
that.  It is quite simple to decode it with a script though.

non-existent variables.

Removed the 'magic' variable ${?} for return values.  'RETURN' without
an argument now stores an empty string in the result variable.  The
result is still passed in a variable called '?' but no script functions
ever see this.

Modified incoming.c so that the 'invalid mesg: 21' etc.. only appears if you
have the 'developer' priv (chatpriv D) on.

removed the 'shift' command from scripts, as obsolete, and can break script
arguments. (or used to)

Fry

Fixed a problem with the 'log' function, when lines got too long. Now logs
will be neater... trust me. :)

Fry

fixed broadcast RPC logging.

Fry

Fixed the autoexec stuff, so you now appear in the wholist.

Simple really, after setting the 'ONCHAT' chatmode, only updated whofile
*after* the autoexec. Kinda silly :)

Fry

Fixed Broadcast IPC's and RPC's

Now actually sends the correct string..

Usage:

SENDIPC <user> <message>
SENDIPB <message>

SENDRPC <user> <type> <message>
SENDRPB <type> <message>

Fry

RPC calls now work.

Usage: SENDRPC <user> <type> <message>

Any user may specify a function to call on a certain message type. these
have to be 'bound'.

In scripts put:

rpc <type> <function>

-------------

Example:

[~/.mwrc]
rpc INIT rpc_init

[~/ipc.mws]
SENDRPC Fry INIT "Test"

etc...

Any type not bound will not be called.

Fry

(This is not done automatically)

(or to have 'E' yourself) to see them on the who list.  They also never
recieve '.notsayto' messages.

This flag is intended for IPC 'daemons', to make them less annoying.
It should not be given to normal users.

Recommended flags for 'daemon' users:
status: rnip
specials: CnTt
chatmodes: pf
chatprivs: sA

Usage: CALL <function> <variable> [arg...]

This command calls the given function and stores the value it returns
in <variable>.

RETURN now accepts an argument.  If none is given, the content of the
called function's local variable '$?' is returned.

The variable is always overwritten, possibly with a null string if
the function is not designed to return a value.

This is an alternative to returning values in a global variable, but
without the compatibility problems of combining scripts using different
variable names.

Now allows user to set the 'ipc' function as you can with events, by using
the 'ipc' command in scripts.

eg:

ipc incoming_ipc

Fry

New message type FORCE_IPC, which can be sent uding the SENDIPC script
command.  When a client receives FORCE_IPC it calls the ipc script function
if it exists.  i.e.

to send "fish" to a user:

   SENDIPC user fish

that user will then have a function called 'ipc', whcih will be executed
the same as any other event - the text (in this case, 'fish') will appear
as $1 and the USER command will return the name of the sender.

Changed 'non-existant variable' from {?} to ""

Fry

by deleting variables).  This seems to work ok with both lists and
hashes but is very inefficient compared to the basic list-freeing
functions (VAR_FREELIST).  Perhaps this could be replaced by a new
'selective purge' function in vars.c (with 2 implementations).

Modified the syntax for variable expansions:
1. Arithmetic expressions should now use "$[..]" instead of "$(..)".
   This is for consistency with shells.
2. The array expansions are disabled (They are likely to cause name clashes,
   and are not useful at the moment).
3. The '/' character is allowed in identifiers, for now.  This is to
   avoid breaking certain scripts.  I still suggest avoiding using it.

of args to a function.

Other special chars (including "$(..)") will probably be changed soon.

Added some more checks for parse errors.  Most should be reported at
runtime now.

Blame whoever did this:

script_init()
{
#ifdef MW_HASHTABLES
        hash_setup();
#endif
        VAR_NEWLIST(&var_list);
}

and this:

### uncomment for hash tables instead of lists
DEFS+= -DHASHTABLES

Replaced a couple more macros with functions

A new ISED function to do a case insensitive SED.  This has not been tested
yet because finnw's latest changes have totally broken scripts (as soon as
it tries to exec any script it SEGV's).

Finnw: please fix this and in future test it before committing.

 - Steve