LambdaMOO server patch release 1.8.0p3

• Date: Sun, 7 Apr 1996 18:22:10 PDT
• From:
  Pavel Curtis <pavel@parc.xerox.com>

I have just released what I'm hoping will be my last patch release of the
LambdaMOO server to the usual place:

	ftp://ftp.parc.xerox.com/pub/MOO/LambdaMOO-1.8.0p3

This is primarily a bug-fix release, though a couple of tiny new features made
it in, too.

>From what I can tell, this release is pretty stable.  After a few days to see
if there's some new stupid bug, it would probably be safe for use in any
situation.

Here's the ChangeLog.txt entry for this release:
-------------------------------------------------------------------------------
Version 1.8.0p3, 7 April 1996
-- Fixed panic when a line of input existed from a logged-in but invalid object
   (as is made possible by force_input()).
-- Fixed `usage' message to include the new `-e' flag.
-- Fixed bug where it didn't work to rename a property changing only the case
   of the letters.  (Thanks to Brack for reporting this.)
-- Exported `get_server_option()' to `server.h', for possible use by server
   extensions.  (Thanks to Richard Godard for suggesting this.)
-- Added `-l LOGFILE' command-line option, to allow specifying the destination
   for logging messages separately from redirecting the standard-error stream.
   This is most useful when intending to `continue' from emergency wizard mode.
   (Thanks to Seth Rich for suggesting this.)
-- Moved test for a built-in function being `protected' and the possible call
   to $bf_FOO() before the test for the correct number and type of arguments.
   This allows a $bf_FOO() overriding verb to take different arguments from the
   built-in it's overriding.  (Thanks to Brian Buchanan for suggesting this.)
-- Added built-in function `flush_input(CONN [, SHOW_MESSAGES])' that does the
   same thing as the user CONN typing their defined `flush' command.  That is,
   it removes all pending lines of input from CONN's queue and, if
   SHOW_MESSAGES is provided and true, prints a message to CONN listing the
   flushed lines, if any.  This makes it possible to flush an input queue from
   within verb code.  (Thanks to Seth Rich for pointing out this issue.)
-- An un-logged-in connection can now only `log in' (via $do_login_command()
   returning a valid player object) if the network connection is actually still
   open.  In particular, the following cases, which used to cause panics, now
   have no effect aside from the obvious calls to $do_login_command():
	1) Something opens a connection, sends input that would log it in, and
	   then closes the connection before the server has had time to run the
	   $do_login_command() task that would log it in.
	2) Someone executes `force_input(CONN, LINE)' where CONN is negative
	   and LINE would log in an un-logged-in connection.
   (Thanks to Dave Kormann for reporting this bug.)
-- Changed the definition of value_bytes() to make it more intuitive to users.
   Now, value_bytes() always returns a non-zero value, since *every* MOO value
   requires *some* memory in order to be represented.  (Thanks to several
   members of the MOO-Cows mailing list for the discussion that led to this.)
-- Fixed disassemble() to properly print the mnemonic for the PUSH_LABEL
   instruction.  (Thanks to Brian Buchanan for reporting this bug.)
-------------------------------------------------------------------------------

• Prev: Re: Killing the Server. What is the best signal to send?
• Next: LambdaMOO stalling on startup?
• Index: MOO-cows
• Thread: MOO-cows