MOO-cows Mailing List Archive


Patch release 1.8.0p2 of the LambdaMOO server

The flow of bug reports on 1.8.0p1 (never very high in the first place) has
apparently stopped, so I've released another patch release of the LambdaMOO
server in the usual place:

If things keep up as they have, maybe this is my last release...

Here's the ChangeLog.txt entry for this release:
Version 1.8.0p2, 18 March 1996
-- Fixed off-by-one bug in set_verb_code() that failed to type-check the last
   element of the CODE list.  (Thanks to Kipp the Kid for reporting this.)
-- Reordered printing of start-up messages and the registration of built-in
   functions to allow function registrars to more aesthetically print out their
   own such messages.  (Thanks to Richard Godard for suggesting this.)
-- Minor configuration change to improve chances of finding an ANSI C compiler
   on SGI IRIX 5.2 systems.  (Thanks to John Wilson for the suggestion here.)
-- Changed the behavior of calls to built-in functions made wiz-only via
   $server_options; for a call to a protected function FOO from an object other
   than #0, if $bf_FOO() exists, it is always called instead of the built-in,
   *even for wizards*.  (Thanks to Roger Crew for helping me beat this horse
   thoroughly dead.)
-- Fixed conversions of floating-point numbers into strings.  The default
   conversion, used in tostr() and toliteral(), uses the exact number of
   decimal digits of precision supported in the underlying representation
   (i.e., DBL_DIG in ANSI C).  The writing of the database file uses four more
   digits to be certain of extracting all of the available information.  The
   floatstr() function allows specification of any non-negative number of
   digits up to four more than DBL_DIG.  (Thanks to Brian Buchanan for
   reporting the problem and to H. Peter Anvin for suggesting the appropriate
-- Fixed task_stack() to test for the given task ID being valid (i.e., naming a
   suspended task) *before* testing that the caller's permissions are
   sufficient.  (Thanks to Judy Anderson for reporting this.)
-- The idle un-logged-in connection timeout can now be set in the database.
   For a connection accepted by a listening object L, the server checks for the
   property `connect_timeout' being defined on either L.server_options or, if
   L.server_options is not defined, on $server_options.  If the
   `connect_timeout' property exists and is a positive integer, that is the
   number of seconds to use for the timeout.  If `connect_timeout' exists but
   is not a positive integer, then no timeout is used.  Finally, if
   `connect_timeout' does not exist, the old default of 300 seconds is used as
   the timeout.  (Thanks to Richard Godard for prodding me into this.)
-- Setting the .programmer or .wizard bit on an object no longer raises
   E_INVARG if the object is not a player.  (Thanks to Roger Crew for
   suggesting this.)
-- Fixed `;' command in emergency wizard mode so that such tasks now have
   legitimate task IDs.

Home | Subject Index | Thread Index