Fri Dec 12 12:10:04 2003 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b14
* made changes to fix building on Mac OS X 10.3 (Panther)
* fixed logon() in backend.c to check for the object's
destructed status before making the apply [reported by
Arren@Anarres]
* moved the #include of std.h to before the HAS_UALARM
check in ualarm.c [reported by Loriel@Lima]
* fixed pluralization of some words ending in f [reported
by Loriel@Lima]
* fixed pluralization of pants [reported by Malic@Lima]
* fixed std_incl.h to avoid varargs.h if __GNUC__ > 2
[reported by Loriel@Lima]
* fixed reference counting in f_bind [reported by Beyond]
Sun Sep 15 17:53:16 2002 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b13
* added a newline after the BIGENDIAN define in edit_source.
* fixed extra_ref accesses in the parser package when DEBUG
is not defined [reported by Tigran@Lima Bean]
* #167: Sprintf missing vars [reported by Andrew@Nanvaent]
* #170: Fixed up some reference problems where dangling
pointers could be left in any number of places. [reported
by Uranus]
* #168: Buffer overflow on doing sprintf("%*d", 300, 1);
[reported by Andrew@Nanvaent]
* #171: Crasher caused by array overflow [reported by
Andrew@Nanvaent]
* #172: Crashers due to outbuf_* failing [reported by
Andrew@Nanvaent]
* #173: Crasher when heart_beats() is called [reported by
Andrew@Nanvaent]
* #180: copy() on a buffer performs no useful action
[reported by Beyond, fixed by Andrew@Nanvaent]
* #181: Memory allocation success not checked properly
[reported by Andrew@Nanvaent]
* #182: Telnet handling flaws [reported by Andrew@Nanvaent]
* #185: allocate_mapping() not freeing input arrays
[reported by Reio Remma]
* #186: check_memory() leaving a string with bad refs
[reported by Reio Remma]
* #178: Oddity in unlink_string_svalue() [reported by
Andrew@Nanvaent]
* #184: Crasher in debug mode while calling clean_up().
[reported by Reio Remma]
Thu Feb 7 18:39:16 2002 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b12
* #152: Fixed a crasher using nested references. Fix supplied
by Avenger@AtP.
* #153: Fixed crasher in debug builds with an invalid assert.
[reported by Pickett@Sumu]
* #159, #162: Fixed a crasher in repeat_string() when the max
string length is exceeded [reported by Eleinor@Arkansia]
* Ported to Mac OS X (Darwin)
* #166: this_player() will now always be this_object() when
the net_dead apply is called from remove_interactive(). This
should fix the write() oddities reported by Uranus. I'm not
labelling this as a COMPAT BUSTER because previously there was
no guarantee what this_player() would be.
* #165: pluralizing -ff words gives incorrect -fves suffix
[reported by Avenger@AtP]
* #161: everyone can see hidden object in named_livings() and
heart_beats(). Applied the fixes suggested by Ben Hoogterp.
[reported by Eleinor@Arkansia]
Fri Mar 30 17:14:07 2001 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b11
* #143: Fixed crashes when destructing various objects involved
in APPLY_INIT calls. Based on patches from Mark G. Adams
* Removed the definition of FP_THIS_OBJECT from
include/function.h since it is not used anywhere and conflicts
with FP_HAS_ARGUMENTS [reported by Arren@AnarresII]
* #148: Disallow load_object() with a filename containing a #
(indicative of a clone). [reported by Pickett@Sumu]
* #149: Crash when removing a function from the simul_efun when
that function is also defined by another object that is then
inherited and recompiled the second time after the function is
removed from the simul_efun object (whew!) [reported by
Avenger@AtP]
* #141: __FILE__ cannot be included multiple times [reported by
Arren@Anarres]
* #140: Fixed a crasher in kill_ref(). Problem occured using a
ref in foreach on a string [reported by Uranus, test case that
finally exposed the problem reported by Pickett@Sumu]
* Fixed foreach() in LPC2C generated code. Updated it to add
the "new" support for the ref keyword in foreach.
* #48: sprintf() from master::error_handler() dumps int instead
of string in 'locals'. Should be fixed up now [reported by
Javelin@Vincent's Hollow]
Mon Feb 12 16:09:42 2001 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b10
* #129: Added SLC_NAMELIST to the telnet.h included with MudOS
for broken systems that are missing <arpa/telnet.h>
[reported by Andrew@Nanvaent]
* #120: Fixed compilation problems when NO_RESETS is defined.
[reported by Arren@Anarres II]
* #139: Fixed a crasher when using a foreach ref within another
foreach loop [reported by Pickett@Sumu]
* #136: Fixed a problem with compiling a program after the
previous program had a compile error inside of a function
pointer [reported by Avenger@AtP]
* #131: Fixed some oddness with to_int(), in particular, a
single digit followed by a whitespace character would return
undefined which is incorrect [reported by
kriton@gameworkshop.com]
* #135: Macro names must begin with either an underscore or a
letter. Successive characters may be digits [reported by
Arren@Anarres II]
Mon Nov 6 18:48:25 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b9
* #104: Corrected the spelling of "ninth" which is used by
the parse_sentence() efun. [reported by lundberg@vr.net]
* #97: More fixes for get_char() when GET_CHAR_IS_BUFFERED is
enabled [fixes from Pelle Johansson]
* #111: Plural of "lotus" is "lotuses" [reported by
Turrican@Discworld]
* #109: Fixed inline program reference increments to use the
reference_prog() function. [Reported by Andrew@Nanvaent]
* #110: Fix a bad call to free_string_svalue() in
f_write_buffer(). The call should be free_svalue() instead.
[reported by Beyond]
* #103: Runtime errors in heart_beat will no longer stop the
entire process. The next heart_beat will run, call_outs
will run, and the time will continue to advance. This is a
bug that is older than old [reported by Scatter]
Wed Sep 27 08:20:05 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b8
* #83: Fixed get_char() when GET_CHAR_IS_BUFFERED is enabled.
Make sure that the string returned is null terminated and
TELOPT_SGA is handled properly [patch from Pelle Johansson]
* #84: Fixed a ref count problem with the connect() master
apply when 0 is returned.
* #86: Using default access modifiers, do not issue an error
if a conflicting modifier is used [reported by Mystic]
* #77: Fixed a crasher when calling map_delete() from filter()
on the mapping being filtered. [reported by Andrew@Nanvaent]
* #79: In certain cases generating a loop, a flag indicating
the code that follows is not reached was being erroneously
set, causing the code to not generate/compile properly
[reported by Uranus]
* #80: If a function is declared but not defined, don't enter
it in the function table when generating C source for LPC.
[reported by Uranus]
* #90: Fixed a problem with inheriting multiple objects that
each define classes and have functions with arguments using
those classes. Properly adjust the types for validating
arguments [reported by Randor@Aurora]
* #81: Filled in missing definitions for reference support for
LPC_TO_C [reported by Uranus]
Thu Sep 21 14:16:49 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b7
* #75: Fixed a typo in C_TRANSFER_LOCAL macro added in v22.2b6
[reported by Uranus]
* #63: Fixed heart_beat errors for optimized builds.
[reported by scatter@thevortex.com]
* #71: For MySQL support, all string/char/text/blob field types
will now be returned as strings by default unless the field
has the BINARY modifier applied to it, in which case a buffer
will be returned.
* #76: Filenames containing '//' are now allowed -- multiple
slashes are stripped down to a single slash [patch from
Skullslayer@RoD]
* Loading a binary now checks the simul_efun date to see if it
has been modified (in addition to driver, includes, config,
etc.) [from Bugs]
Mon Sep 18 20:19:20 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b6
* #57: Fixed a problem with single quotes inside of a macro
expansion [reported by Mystic]
* #64: Fixed unresolved symbol c_transfer_local for LPC_TO_C.
Only used when LPC optimization is on [reported by Uranus]
* #63: Fixed an error in heart_beat causing heart_beat() to
get called repeatedly [reported by scatter@thevortex.com]
* #65: Added some intelligence to build.MudOS on FreeBSD to
try and determine if -export-dynamic or -rdynamic will work
for compiling. Old versions of FreeBSD do not support
them [reported by darren@nighttide.net]
* Enabled #pragma optimize
* #73: Fixed memory_summary() efun to handle recursive data
structures [reported by Avenger@AtP]
* If NO_BUFFER_TYPE is defined, MySQL BLOB fields will be
treated as TEXT fields.
* #50: Fixed a crasher with recursive inherits deeper than the
maximum call depth [reported by Beyond]
Sat Sep 9 16:16:54 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b5
* #16: 'ref' works in foreach again [originally reported by
Uranus]
* #54: Added a new search location for mysql.h to satisfy RPM
installs on Linux from MySQL.com. Also fixed edit_source to
search for libmysqlclient.so properly [reported by jytong]
* #49: Fixed oldcrypt() to actually be oldcrypt() when
CUSTOM_CRYPT is defined [reported by Beyond]
* Fixed a crasher when heart beats and the command_giver is
shadowed [reported by Andrew@Nanvaent]
* Removed Beek's expand_all_defines hack introduced in v22.2a3.
This caused an undesirable compat buster from older versions
of MudOS. Instead, fixed macro expansion to work according
to the ANSI C standard -- almost. The only major difference
now is that comments are not considered whitespace everywhere
that they should be.
Sat Sep 2 09:40:38 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b4
* updated with respect to v22.1pre4
* #40: Fixed compilation error in add_action.c [reported by
Loriel@LimaBean]
* Don't call the first heart_beat until after saving the
initial error context, otherwise an error in heart_beats
or call_outs (set by preloads) will cause a crash.
* #47: When using OLD_ED with a broken client that does not
send CRLF for end of line, don't inadvertantly strip off the
first character of the next command if there are multiple
commands in the input buffer [reportd by fchen@uno.edu]
* #45: Added 'an' as a special word equivilent to 'a' and 'any'
Also, change ordinal for 'a', 'an' and 'any' to 1 from -1.
I'm not sure of the original intent in supporting -1, but the
end result is the same -- at least it's supposed to be. 1
works, -1 doesn't. [reported by Zifnab@Red Dragon]
Mon Aug 28 08:16:50 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b3
* updated with respect to v22.1pre3
* #15: Picked up the fix from v22.1pre2 that was missed in the
last revision for some reason
* #17, #26: Removed a chunk of old dead code for binaries
support that has been obsoleted since function tables are
always sorted and compressed now. Other new code had since
been written assuming this and not handling the (unnecessary)
binaries cases, causing the driver to crash in certain
instances [reported by ackers@coolnet.net, Beyond@Lima]
* #30: If an object has had its program replace, mark the
string that holds its name. Also make sure that it gets
freed properly when the object is deallocated [reported by
Uranus]
Wed Aug 23 05:08:05 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b2
* updated with respect to v22.1pre2
* #4: fixed a problem with using the third defined class in an
object as an argument to a function getting treated
incorrectly as void [reported by Loriel@LimaBean]
* #8: Fix compile problem when both DEBUGMALLOC_EXTENSIONS and
PACKAGE_UIDS are defined [reported by Skullslayer@RoD]
* #16: With COMPAT_32 defined, '&' is now a synonym for 'ref'
everywhere that is appropriate [reported by Uranus]
* #19: Made fixes so that RUNTIME_LOADING will work on FreeBSD
[problem reported by Uranus]
* #23: Corrected arguments passed to compile_object() master
apply when cloning a virtual object [reported by Uranus]
Sun Aug 13 22:25:28 CDT 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.2b1
* updated with respect to v22.1pre1
* fixed a memory leak with MUD port external ports leaking
ref counted objects received (arrays, mappings, etc.)
* fixed up check_memory() code to do proper checks:
. mark_apply_low_cache now bumps extra_ref for each
cache entry's oprogp and progp as appropriate
. dangling object list is checked for missing objects,
and if found, a more appropriate error is reported.
. TAG_SIMULS is used for master applies, so changed the
test in md.c to 3 rather than 2 uses for this tag.
. added marks to objects on the command_giver stack
* PACKAGE_DB compiles again
* properly free an added reference when establishing a new
user connection after APPLY_CONNECT is called.
* fixed a memory leak when updating the master object. The
master apply table reallocated memory without freeing the
previous table.
* 'void' is now acceptable as an argument list for a function
that takes no arguments. If used, it must be the first and
only argument, and it must not be named. Also updated the
testsuite to handle this.
* COMPAT BUSTER: SERVER_IP is no longer supported. Use the
runtime config option 'mud ip' to bind to a specific address.
addr_server and portbind have both been updated to accept a
parameter to specify a dotted decimal ip address to bind to.
Examples:
. portbind -i 192.168.1.1 -p 4000
. addr_server 7334 192.168.1.1
* COMPAT BUSTER: FD6_KIND and FD6_PORT are no longer supported
as compile time options. Set 'fd6 kind' and 'fd6 port' in
your config file to use them. You may use 'telnet', 'mud',
'binary' or 'ascii' as the kind. Additionally, you must now
always specify -p to portbind to specify the port to bind to.
* fixed various crashers in handling eof conditions while
compiling, particularly in unexpected locations (comment,
array block, text block, etc.)
* when DEBUGMALLOC_EXTENSIONS is enabled, free up some compiler
scribble space that is normally reused (cleans up errors that
check_memory() reports)
* fixed a crasher involving BINARIES being defined, functions
being inherited and prototypes, and all sorts of other weird
assorted nonsense [reported by Mystic]
-----------------------------------------------------------------------
Wed Aug 30 18:58:56 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1pre4
* Backported a fix from v22.2a24 for add_slash() to not prepend
a slash when the filename is '<function>'
* #42: Added \n to the two new error messages introduced in
v22.1pre3
* #34: Fixed another instance of regular expression memory
leaking using OLD_ED and an exit function passed to ed()
[reported by Skullslayer@RoD]
* #30: Fixed a problem with marking references for objects in
sentences with get_char() and input_to() [reported by Uranus]
Mon Aug 28 08:11:42 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1pre3
* Check for INADDR_NONE is now done in edit_source rather than
on os specific check in portability.h. [suggested by Beek]
* #27: Fixed socket_write() to return EESUCCESS if no data is
passed to be sent (i.e., socket_write(fd, "")) [reported by
Uranus]
* #33: Fixed a crasher when destructing command_giver from a fp
set to notify_fail was executed for command failure.
* #28: Do not call LPC code when a too deep recursion or eval
cost exceeded error is raised. Especially in the former
case, there is no stack space to run the code and it results
in an unstable environment after the error is supposedly
recovered from. LPC code means the mudlib error handler or
the object_name() apply from (s)printf().
[reported by Pinkfish@Discworld]
* #35: Do not allow lfun or functional binding after to an
object after replace_program() has been called on it, but
before the current thread of execution has completed.
Prevents bad code from creating dangling references to
programs [reported by Skullslayer@RoD]
* Backport from v22.2a27 fixes to contrib efun replaceable() so
that it works properly.
* #34: sscanf() efun was leaking memory in some cases if using
regular expressions [reported by Skullslayer@RoD]
* #32, #38: Fixed a crasher problem with add_action weirdness.
Involves using disable_commands() or destructing an object
in the sentence chain for command_giver and returning 1 from
the function. [reported by Loriel@NS and Uranus]
Wed Aug 23 04:59:44 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1pre2
* #1: fixed compilation error on Solaris where INADDR_NONE is
not defined in system headers [reported by Dvarsk@Nightmare]
* Fix a braindead check in socket_write_select_handler() for
S_LINKDEAD. Also properly close and flag the socket as
linkdead in socket_write() when an error occurs in sending.
[reported by Eddy@MidnightSun]
* #7: Fix a crasher on some platforms in debug_message() where
variable arguments are reset between calls to vfprintf()
[fix from reetnem@bigfoot.com]
* #10: Remove completely (brain)dead code in the socket_accept
efun that called gethostbyaddr on the accepted socket, which
would cause the driver to possibly block unacceptably. This
information was then never used anywhere at all again.
[reported by Eddy@MidnightSun]
* #11: Make sure the accepted socket is set to be nonblocking
in the address server [reported by Eddy@MidnightSun]
* #5: Fixed a memory leak when LPC optimization was turned on
using either #pragma optimize or by setting PRAGMA_OPTIMIZE
in DEFAULT_PRAGMAS in local_options/options.h [reported by
Skullslayer@RoD]
* #15: Fixed a crasher in implode() efun with some older
versions of gcc (reported specifically against 2.7.2.3).
Compiler optimization error [reported by Mystic]
* #2: children(), livings(), and objects() have been rewritten
to better handle destructed objects and things of that
nature. Particularly objects().
* #22: Backport a fix from v22.2a27 to make the (s)printf efun
re-entrant. Allows the no-(s)printf in object_name() master
apply restriction to be lifted. Fixes error handler crashes
[reported by Zifnab@Red Dragon]
Wed Aug 13 22:03:44 CDT 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1pre1
* patched a memory leak with MUD type LPC sockets
leaking arrays/strings/mappings/classes received.
* fix a bad stack crasher when 0x2 passed in the mask
to check_memory() PACKAGE_DEVELOP efun
* fixed the testsuite to work when:
. NO_ENVIRONMENT is defined
. NO_ADD_ACTION is defined
. SENSIBLE_MODIFIERS is defined
. OLD_ED is not defined
. any form of explode() is configured for use.
. any default max eval limit is set (for automated tests)
* fix up rename and cp efuns so that strings tucked away are
not reported as leaking with check_memory efun [backported
from v22.2a37]
* if OLD_ED is not defined, cleanup ed buffer when interactive
object goes netdead (from remove_interactive()).
* more fixes to socket efuns to help prevent hangs on FreeBSD
when a socket closes unexpectedly.
Sat Jul 29 21:30:19 CDT 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1b32
* fixed a crasher with reclaim_objects() and function
pointers introduced in b31 [reported by Skullslayer@RoD]
* rewrote hname_handler to properly handle fragmented response
packets from addr_esrver.
* When read fails in get_user_data(), flag the user as net_dead
before calling remove_interactive(). Don't flush messages in
flush_message() if NET_DEAD is set. Eliminates hangs on BSD
when a user connection is lost [fix from Eddy@LimaBean]
* If an LPC socket is blocked and the connection is unexpectedly
lost, clear the blocked state so that it doesn't hang in limbo
indefinitely. Similar problem to above.
* display a more appropriate error message when the global include
file cannot be included for some reason [reported by Mystic]
Fri Jul 7 20:43:00 CDT 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1b31
* objectp() efun now returns 0 for a destructed object
[reported by Ehan@MedievalTimes]
* reclaim_objects() now reclaims destructed objects
that own function pointers
* fixed some compiler warnings
* fixed ordinal handling in parser package added in b29
[reported by Dvarsk@Nightmare]
* added checks to prevent string concatenation from
exceeding the maximum string length [patch by Meph@LimaBean]
* process_efun_callback: save st_num_arg because it can get
blown away by the call to object_visible() [fixes crashers
reported by Ehan@MedievalTimes]
* FP_LOCAL function pointers now hold a reference to the
program of the object that they're bound to. this results
in replace_program() and swapping being prevented on the
object [fixes crashers reported by Avenger@AtP]
* backed out earlier changes to dont telopt_linemode. much
more support in the driver is necessary to properly support
this. it's too late to put it in now. it's in v22.2a36.
* fixed crasher when MAX_NUM_OBJECTS objects are found by the
parser using plural modifier [reported by Dvarsk@Nightmare]
Tue Jul 4 19:36:09 CDT 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1b30
* virtual objects can be cloned again
* moved STACK_INC into interpret.h from interpret.c so
that it is accessible throughout the driver
* use STACK_INC instead of sp++/++sp throught the driver
* made sure that all destructed objects are actually
const0u (undefined) rather than just const0. done to
be consistent and also to match the documentation
* added stack overflow checks throughout the driver
* fixed a potential crasher in allocate_mapping() if
a function is specified as the filler and an error
occurs evaluating it.
* fixed several locations where the stack pointer was
incremented but unset before an error check. if the
error occured, a garbage svalue would be popped in the
handler
* made sure all swapped object checks are done by checking
O_SWAPPED rather than prog==0
* fixed a crasher that reared its ugly head in various forms
caused by a bad call to free_object() when dereferencing
a class member that held a destructed object
Sat Jul 1 13:14:32 CDT 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1b29
* various fixes from Alaron@RetroMud:
. make db.c compile
. fix an infinite loop problem in db code
. add support for more mSQL datatypes
. miscellaneous typos (error messages and the like)
. correct ambigious if/else in md.c
. move #include of dlfcn.h to std_incl.h
. cleanup some compiler warnings
. check return from fork() in external.c
* the 'say' command in the testsuite will now compile
with #pragma strict_types
* Cloning a virtual object now ensures that O_CLONE is set
on the new object.
* Loading a virtual object that would result in a duplicate
object name now fails with a runtime error rather than
silently allowing the operation and corrupting the object
name hash table (or crashing in a debug build)
* Parser fixes:
. object count passed with ERR_ORDINAL was one too high
[reported by Zifnab@Red Dragon]
. can now access any number of objects in an inventory,
not just the first nine. above 9 requires 10th, 11th,
etc. and will not accept tenth, eleventh, etc.
[reported by Reflection]
* fixed up error messages for defining variables with an
undefined class name and attempting to create an instance
of an undefined class with new()
Tue Jun 20 18:40:18 CDT 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1b28
* fixed the changes made to ed in the last revision. Turrican's
patches were merged in wrong.
* Changes to the pluralize() contrib efun:
. Add an exception for bonus -> bonuses
. Add exceptions for datum -> data and forum -> fora
. Remove the rule *um -> *a (see comment in code for why)
* Cleanups to previous change to add slashes in error tracebacks
* prepend a slash to filename passed to the write callback when
OLD_ED is defined
* correct potentially fatal references to freed arrays and mappings in
F_MEMBER, F_INDEX and F_INDEXR cases by doing destructed object checks
before pushing onto the stack.
* fix crasher when socket_error(-33) is done [reported by Megaboz@Lima]
* added locale support [based on patches from Naebator@Remedy]
* reset command_giver when current_interactive is reset in the backend
to fix random crashes/fatals caused by use of set_this_player() and
command_giver not holding a reference to the object
* raise class redefinition error only if member names and/or types differ
[reported by Pinkfish@Discworld]
Thu Jun 15 17:07:32 CDT 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1b27
* if ARGUMENTS_IN_TRACEBACK or LOCALS_IN_TRACEBACK is defined, save
error context in dump_trace() so as to maintain a consistent state if
svalue_to_string() errors (due to object_name in master_ob generating
an error)
* if LOG_CATCHES is defined, bump num_error before dumping trace on a
catch so that we don't loop indefinitely in the event of an error in
dump_trace().
* fix a crasher when write permission is denied for the file specified
for dumpstat() [reported by Mystic]
* removed redundant code in smart_log() in file.c [reported by Mystic]
* if multiple address server replies are received in a single socket
read, handle all of them, not just the first and throw the rest away
* set nonblocking on newly accepted sockets for all platforms - not
just Linux [numerous reports / various platforms]
* fixed another crasher in (mapping) x *= x [reported by Avenger@AtP]
* removed -fomit-frame-pointer from default optimization flags in
build.MudOS due to problems with stack checkers
* Fixes from Turrican@Discworld:
. fix telnet negotiation when using fd 6 and port 23 with portbind.
. fix bad references in ed if the exit callback function uses
call_out() or input_to()
. remove an inappropriate warning when negative indices are used on a
mapping
* fix a crasher when a class is defined in the simul_efun object and
another object generates a compiler error referencing that class type.
[reported by Pinkfish@Discworld]
* fix another crasher returning from a catch { ... } block [reported by
Mystic]
Mon Jun 12 08:46:33 CDT 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1b26
* don't shut down the driver if a new connection fails to set
non-blocking on linux, just dump the connection and move on
* PACKAGE_EXTERNAL will compile without PACKAGE_SOCKETS now.
* if EWOULDBLOCK is not defined by the OS but EAGAIN is, make
EWOULDBLOCK equivilent to EAGAIN.
* in shutdownMudOS, use OS_socket_close instead of close
* set non-blocking on newly accepted sockets in the socket_accept
efun.
* in socket flushing code, ignore EWOULDBLOCK as well as EINTR.
* error when number of global variables reaches 256 to prevent
overflow in various opcodes [reported by Zifnab@Red Dragon]
* all conversions from lpc float to string now use %f as the
format specifier passed [eventually] to (v)(s)printf.
* restoring a numeric value from a string (restore_variable or
restore_object) will count the last character being a decimal
as a 0 rather than erroring. this is for backwards compatibility
in loading existing .o files that have data saved this way.
[reported by Zifnab@Red Dragon]
* correct index to local variable for warning about unused local
variables [reported by Andrew@Nanvaent]
Mon Jun 5 13:05:02 CDT 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1b25
* several fixes to the terminal_colour() contrib efun:
. bad stack/crash when string contains only %^'s
. reallocate parts array only if needed (when NSTRSEGS + 1
instead of when NSTRSEGS)
. properly support tabs in wrapping [reported by Tigran]
. fix crasher in wrapping caused by uninitialized variable
[reported by Zifnab@Red Dragon]
. allocate the proper amount of space for the temporary wrapping
buffer when constructing the final string.
* fixed crasher in (mapping) x *= x [reported by brett@stummies.org]
* several parser fixes for problems reported by brett@stummies.org
. bitvec_count() only counts the bottom half of each bitvec_t element
. local_error is unused in we_are_finished()
. DEBUG_DEC missing before return in parse_rule() and we_are_finished()
. miscellaneous cleanups in debugging code
. don't reset debug_parse_depth in free_parse_globals() (causes DEBUG_P
to break if used in do_the_call() if it is and is set in
f_parse_sentence() anyway)
. ERR_ORDINAL shouldn't be negative
* fix crasher when input to parse_sentence() exceeds MAX_WORD_LENGTH.
[reported by Zifnab@Red Dragon]
* prepend slashes to filenames for display in ed (f, w, x commands)
[reported by Avenger@AtP]
* added exceptions for plum and virus in the contrib efun pluralize().
[reported by Avenger@AtP]
Wed May 24 12:00:20 CDT 2000 (marius@mudos.org (Marius))
* raised patchlevel to v22.1b24
* updated contact information for reporting bugs
* remove undefined status with += when rhs is real (completes
the fix made in v22.1b23)
* cleaned up a handful of compile-time warnings
* fixed crashers when trying to return from a catch { ... } or
time_expression { ... } block
* corrected the filename and added the line numbers for overlapping
range cases in switch blocks
* moved sentence cleanup from dealloc_object() in object.c to
destruct2() in simulate.c because a sentence with a function
pointer callback could (and likely does) hold a reference to the
object being destructed [reported by Belgarat@Realms of Sorcery]
* add slashes to program and file name in error traces and also for
mudlib error handler
* corrected OS_socket_read and OS_socket_write on Unix to use recv
and send respectively. All socket io is now done with send/recv
instead of mixing with read/write
* fixed crasher in (mapping) x += x [reported by Xavier@AtP]
Tue Jan 12 17:50:55 EST 1999 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b23
* fix += and -= and clear undefined status [reported by Gorta]
* fixed crash in functions() on objects that define no new
functions [reported by Beyond, fix from Alaron@RetroMud]
Mon Nov 9 12:47:38 EST 1998 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b22
* fixed signed variable not to be declared as unsigned in sprintf.c
[reported by Maarten D. de Jong]
* remove unnecessary WIN32 crypt() prototype in port.h; evidentally
it causes MS VC5 builds to fail.
* in read_bytes() make sure the file gets closed when errors occur
[reported by Hamlet]
* fixed remove_shadow to remove all shadows on an object, like
the documentation says [reported by Skullslayer@Realms of the Dragon]
* added support for string[idx] -= ... [reported by Donky]
* fixed traceprefix() to use paths with leading slashes [reported
by Randor]
Wed Feb 25 21:12:30 EST 1998 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b21
* renamed STACK_CHECK to STACK_INC, since it increments the stack
pointer too
* add { } to make binding of else clearer (simulate.c, ed.c,
eoperators.c)
* remove implicit int from prototype (lex.c)
* fixed crasher when terminal_colour() string contains only %^
[fix from Skullslayer@ROD]
* Fixes from Belgarat:
. fixed crasher in explode() when result array was too large
. __cdecl funkiness in backend.c, comm.h, etc; why can't MS
support standard C? No other compiler needs anything this gross.
. remove WIN32 specific code in time() efun ... it is wrong, and
will cause call_outs to misbehave
. fixed off-by-one error in WIN32 code for file_size()
. push_ and push_refed should use STACK_INC not sp++
. fixed crasher in terminal_colour
. added UINT32 define to WIN32 configure.h
Tue Nov 11 12:47:22 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b20
* in portbind.c, change our gid before our uid, while we still
have permission to do so [reported by Radagast]
* slightly clearer error message for end of file while looking for
#endif [reported by Aragorn]
Mon Nov 10 17:56:57 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b19
* added support for classes in != and == [reported by Emeradii]
* fix for float += int [fix from Symmetry]
Tue Nov 4 22:55:49 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b18
* fixed program size check to check correct size
Tue Nov 4 15:57:10 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b17
* fixed typo in grammar.y.pre
* fixed restore_object() to choose the nonstatic version when more
than one var has the same name
* fixed compilation with NO_ADD_ACTION off
Sat Oct 4 23:17:32 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b16
* pluralize() fixes from DrFeelgood: skiff, sniff, gaff, etc.
Handle staff as an exception.
* fixed crasher in function() {} [reported by Kaolin]
* Enforce the 64k bytecode limit
* fixed unterminated comment in options.h
Thu Oct 2 01:21:02 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b15
* typo fix in addr_server.c [from Cloud@IdeaExchange]
* fixed the type of '((mixed)class)->member' [reported by Mystic]
* fixed parsing of code like (: identifier identifier :)
* took filter(string, ...) back out of func_spec.c since it isn't
implemented yet
Wed Sep 24 23:38:36 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b14
* fixed resolve() to pass (0, "255.255.255.255", key) and not
(0, 0, key) to the callback when the lookup fails
* resolve() now calls the callback even when inet_ntoa() fails
[both reported by ekusters@wabi.curinfo.an]
Fri Sep 5 00:40:03 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b13
* added prototype for allocate_empty_class_by_size() to class.h
[from Turrican@Discworld]
* fixed 'alot' -> 'a lot' everwhere it occurs
* compilation fixes for IRIX 6.2; <malloc.h> handled by configure,
"my_malloc.h" is now "my_malloc.h"
* fixed read_file() to only strip '\r's before '\n's [reported by
Burty]
* fixed unused local variable errors to report the correct variable
name in the presence of nested blocks of variables (e.g. {int x;
{ int y; } } )
[reported by Aragorn]
* some unused local variables that were previously overlooked are now
reported
* the parser now accepts "my xth obj" [reported by Lathander]
Tue Sep 2 20:40:31 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b12
* really fixed those invalid declarations this time
* added Dworkin's patches for RUNTIME_LOADING support on NetBSD
* updated Credits.MudOS a bunch. If you have submitted anything,
or your entry is incomplete or inaccurate, email me.
Sat Aug 30 19:14:34 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b11
* fixed invalid declarations in the named_livings() contrib efun
[reported by Qualin and others]
* added Maarten de Jong's updated README-gcc for Amiga compilation
* added missing fclose(f) to read_file()
Wed Jul 23 17:43:48 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b10
* updated with respect to v22pre10
* more fixes for set_hide disabled (named_livings() and find_living())
[patch from Alaron@RetroMud]
* fixed crasher in (array) x += x [reported by Qualin]
* added leading slash to log_error() filename [reported by Jihad]
* fixed func_spec to allow filter(string, ...)
* rewrote read_file(). New version is:
1. readable
2. platform independent (\r translation is now done on UNIX too)
3. catches \0's in all cases
4. doesnt ignore start when called with 2 args [reported by Nameless]
* fixed compilation warnings about S_IS*()
* fixed set_inc_list not to mangle the include list, so get_config()
reports the right value
* fixed array stat bug when copy() is used on classes [reported by
Alaron@RetroMud]
* foreach (x, y in z) { ... } no longer warns about x,y being unused
[reported by Qualin]
* fixed unsafe uses of push_svalue() in link()
Thu May 15 14:08:11 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b9
* added a missing newline to the end of trace lines when
MUDLIB_ERROR_HANDLER is off
* added Burty's fix for the macro expansion bug
* fixed compilation with set_hide disabled and add_action enabled
[reported by Mystic]
* fixed compilation with input_to and get_char disabled [patch from
Mystic]
* typo fixes in the break_string() simul [reported by Randor]
* some minor changes to ChangeLog.beta
* some changes to IT_CRASHED
* fixed compilation of tail() with COMPAT_TAIL on
Wed Apr 16 07:34:13 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b8
* updated with respect to v22pre8
* fixed type checking for functions that don't use all there arguments
[SUPPRESS_ARGUMENT_WARINGS off; fix from Wodan@Discworld]
* added missing #ifdef DEBUG in simulate.c [fix from Jesse@Doom]
Sun Apr 13 23:36:52 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b7
* added Belgarat's Win32 compilation fixes
* fixed compile error in simulate.c [Ailima, Aragorn, Avenger, etc]
* fixed dumpstat() not to crash on recursive structures
Sat Apr 12 14:57:49 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b6
* updated with respect to v22pre7
* fixed the max_eval_cost error check in call_out.c to pop the
error context correctly [fix from Ceda]
* Don't crash if an object which is snooping another object gets
destructed
Sun Apr 6 17:55:11 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b5
* updated with respect to v22pre6
* fixed a bug in the (mixed)->y code that caused it to never find
a class member. Code like: mixed x; x->y or mapping m; m["foo"]->y
should work now assuming it is unambiguous.
Wed Apr 2 14:15:48 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b4
* updated with respect to v22pre5
* check if my_string is allocated before trying to mark it
[reported by Ceda]
* mark the master_user_list in the parser package
[reported by Ceda]
* 255 -> LEX_EOF in lex.c [reported by Avenger@CoreDump]
Sat Mar 1 23:32:01 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b3
* updated with respect to v22pre4
* fixed the error context for error messages at the end of the file
[reported by Jihad]
* don't treat directories as source files, even if they end in .c
[reported by Jihad]
* fixed a bug that allowed the type of a global variable to be omitted
if a modifier was used
* fixed crasher in ref code [reported by Qualin]
* fixed #undef SENSIBLE_MODIFIERS not to claim static variables are
also nosave in the variables() output [found while tracking a bug
report from Aragorn though I'm not entirely sure this is what he
meant :-)]
* fixed leak in errors during fprintf() in save_object
* moved init_instrs (was init_num_args) earlier so it gets the names
of internal efuns (_call_other, etc) correct for LPC->C. Moved the
name fixing to be part of query_instr_name().
Fri Feb 7 15:31:47 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b2
* added Burty's fix for something involving backslashes, quotes,
and #defines [please describe patches ;-)]
* fixed crasher in mapping composition (reported by Andrew)
* fixed foreach () with a global as the loop variable [reported
by Jihad]
* fixed crashes related to errors in object_name()
Fri Feb 7 15:22:52 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22.1b1
* updated with respect to v22pre2
----------------------------------------------------------------------------
Sat Aug 30 19:24:51 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22pre11
* fixed ed_start() with no args to set the type of its return value
[fix from Jalh@LimaBean]
* fixed 'buffer[i] = 0' to work again
* fixed save_object() to save '0.000' as a float and not as int '0'
[reported by Descartes; patch from Aragorn]
Wed Jul 23 17:39:48 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22pre10
* added workaround for Ultrix sh bug (returns error code on if
without else) [patch from Hamlet@IdeaExchange]
* disabled function argument checking for foo::bar() since the patch
that was added could find the wrong function entry if the function
was overloaded, and the fix isn't trivial. 'private' is now handled
as a special case. [bug reported by Qualin]
Thu May 15 13:58:13 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22pre9
* when NO_ADD_ACTION is defined, set O_LISTENER for objects which
define a receive_message() function so message() calls it
Fri Apr 25 13:58:36 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22pre8
* using line numbers with quit in ed ('12q') no longer clears the
buffer [reported by Jihad]
* fixed replaceable() not to consider the internal function
#global_init# [reported by Donky]
* added Wodan@Discworld's fixes for PACKAGE_EXTERNAL
* another fix for file operations when the master object isn't loaded
yet or doesn't have a valid_{read,write} function
* Use 128 instead of SOMAXCONN for listen(2) since some BSD derived
operating systems have a ridiculously low value for SOMAXCONN.
[fix from Aule@The Two Towers]
* clear in_mblock in the ed indentor in case a block was unterminated
during the last use [fix from Aule@The Two Towers]
Sat Apr 12 05:25:31 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22pre7
* __OPTIMIZATION__ and __COMPILER__ are defined again; lex.c needed
to include cc.h [reported by Decker]
* fixed false alarm ("Bad stack after evaluation") when
time_expression { ... } was used with DEBUG defined
* fixed foo::bar() not to allow all sorts of nasties:
. type checking is now done
. private functions can't be called
. (: foo::bar() :) is not a bindable function pointer
Sun Apr 6 17:45:49 EDT 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22pre6
* fixed terminal_colour() to add indentation to lines which just barely
didn't fit (i.e. were one character short of fitting) on the
previous line [reported by Jihad]
* fixed a bug that caused synonyms for OBJ rules not to work
* removed a #include that caused some malloc packages to fail to work
on Solaris [reported by Aragorn]
* fixed an incorrect typecheck in socket_connect() which disallowed
function pointers as one of the callbacks [reported by Deathblade]
* fixed the LARGEST_PRINTABLE_STRING checks to allow printing of
strings with length == LARGEST_PRINTABLE_STRING
* added a missing newline to the Inherit chain too deep error message
Wed Apr 2 13:31:44 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22pre5
* fixed leak in check_valid_path() [reported by Fermat]
* fixed crashers in function pointer add_actions() when the driver
is run in -d -d mode [reported by Aragorn]
* ported back the socket close callback fix
* disabled #pragma optimize
Tue Apr 1 23:53:07 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22pre4
* Added Leslie@StyleGates' fix for a leak in REVERSIBLE_EXPLODE_STRING
* added Eggjon@Nameless Sorrows' fix for a typo in WIN32 code in file.c
* fixed crasher in using functions() on an object with no functions
[reported by Valentino]
* fixed compilation error in PACKAGE_EXTERNAL
* fixed save_object() not to call valid_read() [typo in src]
* added Deathblade's fix for filter_mapping() when the result mapping
is large
Tue Feb 25 21:30:57 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22pre3
* fixed bug in rename() efun [used Turrican@DW's fix (essentially);
tons of people reported this one]
* fixed crasher in PACKAGE_EXTERNAL (reported by Turrican@DW)
* fixed livings() not to return hidden objects (the check was
backwards)
Sun Feb 2 14:09:47 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22pre2
* custom_crypt() now never returns NULL, and can deal with
arbitrarily long salts; also a number of compilation fixes
* fixed crypt() to use the password as the password, not the
salt; also fixes a crasher when salt == 0
* added the old version of crypt as an oldcrypt() efun; useful
for converting old passwords to the new format, or on systems where
the new crypt is not backwards compatible (FreeBSD)
* fixed edit_source to catch options that are #undef'ed in options.h
but not set in local_options
Sun Feb 2 14:08:33 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22pre1
* added documentation and setup the release directory structure
Sat Jan 25 03:40:07 EST 1997 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b27
* increased the salt length to 8; salts from the mudlib are no
longer truncated at 2 characters
* added Fermat's crypt package; use CUSTOM_CRYPT to enable it
* fixed problems with cp() and rename() [reported by Jihad]
* fixed crasher in restore_object()
Mon Dec 30 03:57:55 EST 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b26
* fixed restore_variable()/restore_object() not to lose precision on
values between 0.01 and 0.0001
* fixed sprintf of integer types not to ignore the precision field, and
fixed the '+' option in connection with 'f' and a precision (e.g. "%+.5f")
* added two patches from Ceda for Win95 crashers
* added ARCH entry for NetBSD
* valid_read/write now get passed the real name of the save file for save/restore_object
(e.g. including the .o)
* save and restore_object no longer tack on a .o if the filename already has one
* the value returned by valid_read/write, if a string, is used as the filename again
[allows ~ expansion, etc]
* added dchen@sei.xjtu.edu.cn's fix for a problem with STR rules on HP/UX
* fixed bug in 'nomask' error message
Sun Dec 8 22:54:36 EST 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b25
* fixed crasher in trying to call_out() from a destructed object
* fixed read_file() not to read \0's into strings
* fixed a number of builtin operators not to return undefined values
* added Marius' patch to not warn about '\"'
* fixed reclaim_objects() not to leak destructed objects found as
mapping keys
Fri Dec 6 17:36:10 EST 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b24
* fixed typo in INLINE_STATIC changes
Fri Nov 8 11:45:33 EST 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b23
* function_exists() no longer finds #global_init#
* don't include #global_init# in functions()
* fixed a bug that would cause (mixed)[<anything] to be of
type int, not mixed
* added Ceda's fixes for file operations on directories when
using Win95
* mention ./build.MudOS debug in IT_CRASHED
* fixed unique_mapping() [it broke last time the mapping hash changed]
* ported back from v22.1a5
. fixed crasher in using the name of a file that doesn't exist
as the object in an efun callback (e.g. filter(({}), "foo", "bar"))
* minor change to mud_status() formatting
* removed the restriction disallowing converting 0 to network byte
order using write_buffer()
* fixed erroneous too few argument error messages when the last
pattern was %*s followed immediately by another pattern.
* fixed delayed crasher in errors with PROFILE_FUNCTIONS defined
(this is the infamous one that causes string constants to
unexplainably start pointing to random memory locaitions)
* fixed a crasher in refering to yourself when more than
MAX_NUM_OBJECTS are around
* bumped MAX_NUM_OBJECTS up a bit
* fixed a crasher in doing moving or destructing an object, then
using command() in an add_action() that returns zero.
Sat Nov 2 17:41:50 EST 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b22
* fixed a bug in the parser that would prefer "There is no x ..."
to a better error message in some situations
Mon Oct 28 16:38:55 EST 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b21
* fixed crasher in 'ed_start()' [found by Beyond@LimaBean]
* fixed a bug in save_object that would insert lots of spaces into
the temporary file name [reported by Aragorn]
Mon Oct 21 17:40:33 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b20
* increased the save_object() temporary file name buffer from
80 to 256 bytes, and made sure it can't overflow
[reported by Avenger@CoreDump]
* fixed crasher in recieving a mud-mode packet which claimed to
have a negative length
* fixed leak in errors during call_outs
* fixed CLOSING sockets with data pending for reading not to cause
the driver to busywait
* tighter compile-time type checking for indexing. Disallow:
. mapping[<anything]
. function[anything]
* fixed crasher in parsing a string constant while the scratchpad is
full (most commonly caused by unterminated strings; reported by
Jihad)
* sending a SIGUSR2 to the mud will now cause it to eval out; temporary
fix until eval_cost gets rewritten.
* fixed delayed crasher in errors during the loading of the master or
simul_efun object
* fixed crasher in long @@blocks
Fri Oct 11 14:08:45 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b19
* fixed miscalculation of result length in terminal_colour() when
the string ended with a newline followed (only) by color codes
[reported by Descartes, LordSutch]
* bugfix in the flushing of 'CLOSING' sockets [from Turrican@DW]
* fixed read_file() to return "" and not 0 for zero-length files
* fixed crasher in words longer than 1024 bytes in the parser
[reported by Descartes]
Mon Oct 7 16:28:04 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b18
* another bug fix in OLD_RANGE_BEHAVIOR
* cleaned up the disk full logic a bit; also added error checking
on the fclose(), which apparently the call that usually errors.
* changed non-void functions using 'return;' to a warning instead
of an error, since it doesn't actually affect generated code
* more debugging code to the terminal_colour() length check
* Get the left and right hand side in the right order in
"Bad assignment" error messages [reported by Aragorn]
* fixed crasher in calling an undefined (and not prototyped) function
from an inherited program [reported by Aragorn]
* fixed wrapping in terminal_colour() when there are colour codes
between the last space and the end of the line
[reported by Jihad]
Thu Oct 3 18:12:42 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b17
* fixed socket_close() to not close sockets immediately when data
is buffered internally; this adds a new socket state 'CLOSING'
which will transparently change to 'CLOSED' when the data has
been written.
* call reset_error() at the top of parse_sentence() so that everything
is set up in the event of an early return if the number of words
is greater than MAX_WORDS_PER_LINE; prevents a crash trying to
make a call based on out of date match information. [reported
by Ra@Deepest]
* fixed a bug in type checking that caused many type errors not
to be caught at compile time. Highlights:
. Type errors in +=, relational operators, unary -, ~, array
indexes now caught at compile time
. 'return;' in a function with a non-array, non-class return value
is now caught
[based on a bug report from Avenger@CD]
* fixed bugs in various operations that would not clear the
'undefinedness' of the result. This would cause undefinedp()
to return 1 if the value was zero, e.g. ((undefined) & 1)
or (!!(undefined))
[reported by both Avenger@CD and Ciao@Team-Aye]
* fixed a bug in OLD_RANGE_BEHAVIOR that would cause indexing off
the start of the array with [..<x] to wrap around to the end
[reported by Aragorn]
* fixed a crasher in ambiguous parses; thanks a bunch to
Descartes for tracking down this one [reported by at least
10 different people; this is the query_the_short() one]
Tue Oct 1 19:07:42 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b16
* fixed crasher in ending a sentence with an article, adjective, etc
in the parsing package [reported by Aragorn]
* small fix for compilation on machines which have MAXSHORT but not
USHRT_MAX
* bug fixes in terminal_colour():
. don't wrap 1 char early when breaking lines
. don't add indentation after the last newline in a string if
nothing follows it
. get line breaks right inside words that are longer than the
wrapping width
Mon Sep 30 21:20:26 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b15
* fixed crasher in using something other than a string or object
for an object name in efun callbacks, like:
filter_array(({ 1 }), "func", ([ ]))
[reported by Jihad]
* fixed crasher in map_delete() with COMPAT_32 defined
* fixed infinite loop in parse_remove() [reported by Aragorn]
Wed Jul 10 15:14:11 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b14
* Fixed a bug that would cause the length of the return value
of terminal_colour() to be set incorrectly if the string
contained a newline (reported by Sibbis@Suvangi)
* also fixed indentation for terminal_colour()
* fixed typo in ed.c that cause the exit_function to be used
as the write function (or crash if a write function was
specified with no exit function) [fix from Earendil]
* fixed compilation with sort_array(), map_array(), tell_room(),
or sizeof() commented out in func_spec.c
* fixed a crasher in $(anything) outside of a function pointer
* fixed a typo that could cause errors to be ignored while writing
save files
Wed Jul 3 14:46:41 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b13
* Fixed several compilation errors involving the_null_array
* Fixed code generation for expressions involving 256 or more
consecutive simple pushes
* Fixed sscanf() not to give a "Format string cannot end in %"
error if it ends in (a legal) "%%"
* Fixed sscanf() to not give "Too few arguments to sscanf()"
when the last argument is a "%%"
* Fixed a bad error message when the format string ends with an
isolated "%"
* A couple of testsuite changes
Sun Jun 30 13:52:01 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b12
* removed null_array() since the_null_array is now explicitly checked
for during deallocation, so its refcount is irrelevant (idea from
Avenger@CoreDump)
* fixed some crashers in errors before function calls with implicit
catch()'s; for example call_out's to destructed fps when
PROFILE_FUNCTIONS was defined.
* fixed crasher in replaceable(something, ({})) [reported by
Avenger@CoreDump]
* fixed crasher in ERR_IS_NOT, ERR_NO_LIVING, ERR_NOT_ACCESSIBLE
errors from the parsing package [this is the Lima 'pat <obj>'
crasher]
* Reduced the weight of rules where OBJ matches due to an error.
Should eliminate cases where the parser erroneously reports
'There is no ... here' instead of a more sensible error message.
* socket_error() message change: EEBADF is now 'Socket is closed'
and not 'Descriptor is invalid'
Thu Jun 27 20:46:06 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b11
* Fixed a crasher in updating a simul_efun which inherits an
unloaded object (this also happens to be the Lima Bean
'loadall /secure' crasher)
* Cleaned up destruct_object() a bit
* fixed more ARRAY_STATS problems
* fixed an unitialized use of version_buf when a config file
isn't specified
Sun Jun 16 18:08:32 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b10
* Fixed a compilation problem with ARRAY_STATS undefined; reported
by a number of people
* Added the ChangeLog entries for v21.7b14 back in
* fixed a typo in the Makefile: -rm instead of rm in a shell section
* fixed a crasher in destructed function pointers in call_outs when
PROFILE_FUNCTIONS was enabled; should also fix a number of other
possible crashers in errors thrown in odd situations before calling
functions
* Check if wait3()'s return value is positive, not just nonzero so
we don't loop on errors (most notably ECHILD). Fixes
external_start() to not hang the MUD on BSDish systems
* The two characters recognized by terminal_colour (defaults to "%^")
is now configurable near the top of the routine
Fri Jun 7 11:13:14 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b9
* Fixed typos in GNUmakefile, efuns_main.c, grammar.y.pre, simulate.c
* Added someone's patch for printing out the version and architecture
before the 'You must specify a configuration filename' error message
* Note: static clean_up() functions are now called again; same with
catch_tell() when NO_ADD_ACTION is on
Tue Jun 4 18:00:31 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b8
* The third argument passed to valid_override() now has a leading
slash
* The filename passed to compile_object() has a leading slash
[fix from Descartes]
* Made the "Program size miscalculated" debug message non-fatal,
since it doesn't really cause anything to break.
* function_exists(string, object, flag); if flag is nonzero, private
and static functions are checked too.
* Fixed two bugs in terminal_colour():
. if wrap is specified and is zero, do no wrapping
. pull indent and wrap off the stack in the right order if both
are specified [fix from Descartes]
Tue Jun 4 16:26:49 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b7
* fixed a few compilation problems in interpret.c
Tue Jun 4 12:19:09 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b6
* Fixed a typo in the BSD readdir() configure check
* Added slashes to the front of many object/program names. The most
noteworthy are: deep_inherit_list(), inherit_list(), call_stack(),
names passed to author_file(), domain_file(), privs_file(),
creator_file(), a few error messages and tons of debug messages
* fixed the parser package to mark it's error messages when
CHECK_MEMORY is on
* fixed a bug that caused " and \'s on the line of the terminator
of an @ block to have \'s inserted before them causing syntax
errors
* call_out, input_to, socket callbacks and add_action can no longer
be used to call ##global_init##
* a few more testsuite tests
* it is now illegal to insert the character '\0' in to the middle
of a string
Mon Jun 3 15:03:16 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b5
* fixed pluralization of "gum"
* added Blackheart's fix for a compilation error when CALLOUT_HANDLES
is turned off
* fixed various bugs with string statistics
* fixed a few bugs bug where classes would cause num_arrays to be
wrong
* added an ARRAY_STATS option for those who want to turn array
statistics off
* added DEBUG checks for allocd_bytes and allocd_strings
* added a debugging check to make sure FREE_MSTR isn't called on
a string with more than one reference
* fixed crasher in switch() on a non-shared string
Sun Jun 2 20:10:59 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b4
* fixed a bug in epilog that allocated space for A_FILE_INFO
in the program structure, even though that data isn't stored there,
wasting approximately 100 bytes/program
* fixed the configuration checks to be a bit more robust with respect
to various files not existing or being outdated ('configuration'
in particular). Fixes compilation errors on OSF reported by Yaynu
and onSun 4.1.4 as reported by Pace
* When NO_SHADOWS is off, and INTERACTIVE_CATCH_TELL is on,
add_message() doesn't need to call catch_tell() since the shadows
already got a chance via tell_npc(). Also fixes a bug reported by
Cire that caused two deep recursion errors when a user was shadowed
with those option settings. shadow_catch_message() also moved to
comm.c, since it is only called from there.
* typo fix from Darrin Wilson for compilation with DEBUG not defined
* a few small optimizations to switch()
* changed POINTER_INT back to a signed type; the fact that it was
unsigned caused switch to misbehave on negative numbers and/or
other cases
* program_info() now calculates class size as well; also fixed a
bug in the calculation of function table size.
* changed an instance of "catch_tell" to APPLY_CATCH_TELL
* characters with the 8th bit set are no longer ignored
Thu May 30 21:11:57 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b3
* added Kaolin's fix for a crasher in terminal_colour()
* fixed another crasher in the prototype-before-inheriting fix
(from Ceda)
* added some more Windows 95 fixes from Ceda
* made all the calls qsort() strictly conforming
* configure now checks for 'const' and 'inline'
* debugmalloc now fills freed blocks with 0x78 so using a block
after it has been freed will crash/misbehave
Thu May 30 15:12:42 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b2
* fixed a crasher when more than 65535 instances of a shared string
are allocated
* minor fixes to various pointer hashes to avoid warnings on 64-bit
architectures
* turned RUNTIME_LOADING off by default since not all architectures
support it
* Fixes for terminal_colour() with illegal/negative widths/indents,
also fix for crasher introduced last patchlevel
Wed May 29 01:39:20 EDT 1996 (tim@wfn-shop.princeton.edu (Beek))
* raised patchlevel to v22b1
* fixed typo in backend.c in Windows 95 code
* Bumped the maximum byte transfer setting in the example config
file up to 200k
* Added some Solaris compilation fixes based on information from
Aragorn, Darrin Wilson, Descartes, et al
* fixed terminal_colour() to do wrapping even if there are no %^'s
in the string
* mud_status() now notes when STRING_STATS is off
----------------------------------------------------------------------------
(v21.7b22 promoted to v21.7)
Thu Apr 11 21:04:27 EDT 1996 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b21
* fixed a bug in replace_program() which would sometimes zero out
too many/the wrong variables
Sun Mar 10 04:49:28 EST 1996 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b20
* removed abort() prototype from lint.h; SunOS defines it as int
abort(), which is very odd since it never returns ...
* made a zero return value from domain_file() legal again
Fri Mar 1 12:20:12 EST 1996 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b19
* set the telnet flag for anything that responds to telnet queries,
even if it refuses terminal negotiation
* fixed debugmalloc compilation problem (had old name for CFG_MAX_EFUN_SOCKS)
* fixed crasher in author/domain_file() returning bad types
Thu Feb 22 10:38:57 EST 1996 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b18
* Fixed ./build.MudOS to create comptest.o so it exists for the ar
test
* tried a new fix for the namespace collision between automatic
options.h defines and include/runtime_config.h; the options.h
names are now in the CFG_* namespace
Sat Jan 27 16:36:52 EST 1996 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b17
* added Quango's patch for the file_length() efun
* fixed clear_notify() to not assume this_player(); this means
destructing an interactive no longer cancels notify_fail() messages
for whoever this_player() really is
* tidyed up remove_interactive() a bit
* fixed include/runtime_config.h to not define some things which
conflict with predefines from options.h
Sat Jan 20 15:03:54 EST 1996 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b16
* fixed a crasher in using set_this_player(0) from inside a player
command
* fixed a crasher in loading an object while another object was
compiling
* fixed compilation on systems where GNU make isn't present; RANLIB
wasn't being passed to the subdirectories
* removed the restriction that a non nomask function can't be
overloaded and made nomask
Wed Jan 3 14:24:12 EST 1996 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b15
* fixed add_action() to check if it's flag is in the legal range;
prevents crashes when bit 4 is passed, since that is used internally
to mean 'function pointer'
* In the parsing package, fixed a crasher on sentences of the form:
"<verb> <valid object> <garbage> prep <valid object>"
when a rule of the form:
"OBJ prep OBJ"
existed
* When checking includes, include file_incl.h as well to check for
conflicts there. Fixes a compilation problem on BSDI where
<sys/dirent.h> is _NOT_ protected from multiple inclusion
* fixed the 'q' and 'Q' commands to not clear the buffer it there
is a syntax error
Tue Jan 2 17:43:55 EST 1996 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b14
* ported back the fixed call_out leak fix from v22a17
* fixed leak in illegal patterns to reg_assoc()
* fixed a string reference bug in parse errors very near the end
of a function
* fixed a leak in using F_TRANSFER_LOCAL on a variable containing
a dested object
Sat Dec 30 20:50:17 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b13
* Fixes from ngreen@plains.nodak.edu:
. crasher in lpc_info() when one of the objects was swapped
. implemented time_expression() from compiled code
* Fixed crasher in defining a function, then inheriting a copy
of it (as opposed to inheriting first) with #pragma warnings on
[reported by Lima Bean, Vonotar, Aragorn, Jihad, ...]
* Linux compilation fix: check for SOMAXCONN in <sys/socket.h>
Tue Dec 5 19:24:06 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b12
* really corrected the order of arguments to fread() in file_length()
* fixed a crasher in the removal of heartbeats from objects.
Mon Dec 4 15:52:24 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b11
* fixed a misplaced parenthesis in the file_length() contrib efun
Mon Nov 27 22:45:28 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b10
* fixed a optimizer bug that would clobber parse_command() calls
* Two fixes from Raistin@IE:
. chop off y before adding ies in pluralize()
. initialize verb_entry->flags to zero so that the parsing packages'
hash tables aren't unnecessarily set up.
* Fix from Aragorn: pass args in correct order to fread(), and use
memchr() instead of strchr() to handle binary files correctly in
the contrib efun file_length()
* fixed the ranlib check to work correctly
Wed Nov 22 12:47:40 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b9
* fixed a crasher in x->y->foo() when x->y was a destructed object.
* fixed a crasher in the parser error message "The foo isn't bar."
Tue Nov 21 15:07:27 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b8
* fixed a // comment in generate.c [gcc 2.7.1 allows it *roll*]
* Following patches from Quango's DW version of v21.7b7:
. added Linux/m68k to arch.h
. "thief" -> "thieves" in pluralize()
Wed Nov 15 14:30:44 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b7
* #include "file_incl.h" in addr_server.c so that <sys/filio.h>
is included and SVR4 systems are happy.
* check for the existence of ranlib on build.MudOS, so systems
that don't have it (Solaris) don't complain
* Added a missing newline to the end of the 'Q' help in ed.
* Fixed a crasher when a trailing slash was included in the first
argument to rename(), and valid_read() didn't strip it off
* Fixed a misoptimization of code where local variables died
inside arguments to a function pointer.
* Fixed the initialization chain to be made even if the object
being inherited defines no functions, just initialized variables.
* Fixed a bug that would cause (: private_inherited_var :) to
either go unnoticed, or print a garbage variable name in the
error
* Changed the 'internal disaster' regexp error message to
'unexpected ]' which I think is right. Let me know if it
shows up in other cases.
Tue Nov 14 12:47:48 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b6
* Don't crash when buggy telnet clients send an IAC SE (end suboption)
without ever having send a IAC SB (begin suboption)
[as a side note, there is a windows 95 client that sends:
IAC WILL TELOPT_TTYPE "vt100\0" IAC SE
(1) (2) (3)
1. should be SB, which differs by 1 from WILL
2. missing TELQUAL_IS
3. terminating null isn't part of terminal types
more high quality PC software, I guess ...]
* fixed #pragma optimize to not kill variables that only die along
one branch of an if/switch
Tue Nov 7 17:46:21 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b5
* fixed a crasher with #pragma save_types
* fixed indentation errors in ed when OLD_ED was off to report correct
line numbers and send errors to the correct place
* fixed a crasher in using an expression of type (class foo *)
as the left side of ->
* fixed a crasher if a socket had a file descriptor greater than 32.
Also, function pointers as callbacks should get called/not crash now.
Fri Nov 3 16:24:49 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b4
* check for overflow of suboption buffer when recieving quoted IAC's
* fixed crasher in __TREE__ on blocks containing (opcode_2) nodes
* fixed a bug in (int constant)/(real constant)
Sun Oct 8 19:52:11 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b3
* added Kalinash's fixes for FreeBSD compilation
* fixed some crashers in parse_command_users(): returning the
environment of a user, returning an object unknown to the
parser, or destructing objects involved in the parse
* added a possible workaround to get around a possible non-feature
of FreeBSD/NetBSD make commands (can't find rule for malloc.o)
* added a patch from Yakko so bsdmalloc compiles on Linux
* fixed a crasher in the verbose version of functions() on objects
with global class variables
Fri Oct 6 20:26:51 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b2
* fixed a crasher in disabling a heartbeat from a heartbeat
(usually only if a large number were deleted, like most of
the heart_beats on the mud)
* reorganized the signal handling to be a bit simpler
* fixed some compilation problems with the matrix package
Fri Oct 6 20:26:51 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.7b1
* updated with respect to v21.6
* fixed the handling of call_outs (especially to function pointers)
in the ref count checking code
* take objects on the destruct list into account in the ref checking
code
* fixed a crasher when the master object self-destructs during
connect()
* fixed a number of warnings and improved -pedantic support
* errors in get_save_file_name() will no longer cause zeros in
users()
* fixed a memory leak in the cached ids of interactive objects
* fixed a crasher in replace_program()
* more verbose error message for inherit chain too deep
--------------------------------------------------------------------------
v21.6b7 promoted to v21.6 with the following changes:
* fixed a bug that would cause the driver to busy-wait when an
object had it's heart_beat enabled, but had no such function.
* trap SIGFPE to avoid going down on floating point exceptions
* run runlib on the archives for systems that need it
* Actually use the limits.h configure info
Thu Aug 24 06:53:08 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.6b6
* crasher fix in loading objects with a .c.c suffix. The current
fix entails stripping multiple .c's off the end instead of just one.
* edit_source -configure now checks if the configure.h file needs to
be remade due to new config checks
Mon Aug 21 03:18:45 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.6b5
* fixed a crasher in parse_command() [reported by Magician]
* fixed a few bad free's in parse_value() [reported by Aragorn]
* added Symmetry's fix for Tensor's 'Couldn't find enabled object in
heart_beat list' crasher
* fixed a crasher with #pragma optimize on
* added pluralization for do, go, was
Mon Aug 21 03:16:33 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.6b4
* make configure.h early enough so that it can be included in
func_spec.c, so that the rusage() efun can be included
* added in Deathblade's fixes for remove_call_out()
Thu Aug 3 11:29:56 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.6b3
* added REALLY verbose parsing mode for those really hard to find
parser bugs; use flag > 1 to parse_sentence
* LPC->C fixes:
. fixed a crasher in compilation of control jumps [break, continue]
. fixed an error in adding labels that could write one byte
past the end of the program block
. fixed a syntax error in code for objects using function() {}
. don't generate typechecks after the 4th efun arg
. fixed incorrect argument number in errors from above type checks
. varargs support
. fixed several bugs in foreach handling
. reordered code to fix forward references to functionals when they
occur in global initializers
. fixed continue and break branches for files which generate > 32k
of source
. added a lot of prototypes to cfuns.h so LPC->C code compiles more
quietly
. fixed a crasher in __TREE__ { /* something involving foreach */ }
Sun Jul 30 11:11:12 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.6b2
* fixed error() to add a \n to the end of the string if one is not
present
* remove_call_out() [no args] now removes ALL this_object()'s call_outs
* added pluralize() prototype
* parser fixes:
. remote livings are always within reach
. fixed a bug that caused direct/indirect objects to not get
callbacks
. fixed a crasher when parse_my_rules tried to return an error
Sat Jul 29 20:12:46 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.6b1
* updated with respect to v21.5
* added next_bit(). next_bit(string str, int foo) returns the
next bit set in str after foo, or -1 if foo is the last one
(str is a set_bit/clear_bit string). For example, to loop over
set bits:
int bit = -1; while ((bit = next_bit(str, bit)) != -1) { ... }
* Added a warning for (: string constant :)
* fixed another crasher in filter_mapping
* fixed crashers in promote_to_float and promote_to_int
* clarified the error message for passing x... to a non-varargs
function
* parser stuff:
. added parser support for 'me' and 'myself'
. fixed 'the' and 'a' support to not inappropriately gobble them
. added a debug mode turned on with an optional flag to
parse_sentence: parse_sentence(foo, 1)
. Added parser support for visibility:
ob->inventory_accessible() can objects inside be used?
ob->inventory_visible() can objects inside be seen?
as usual, they are cached, so call parse_refresh() if they change
. Added extra attributes for tokens; the for is XXX:wxyz
where w, x, y, z are one of:
l - must be living v - need not be accessible (visible only)
so "OBJ:l" is the same as "LIV", and "LIV:v" is someone you can
see.
. fixed a bug introduced in the get_single optimization that
erroneously reported ambiguous parses
. fixed STR to return the original text, not the sanitized version
. the 'verb' passed to verb_* rules are also no longer sanitized
. same with 'real names' of objects
. finished implementing the error messages, including 'There is no
X here'
. optimized STR rules a bit
. fixed a few bugs in finding interpretations with multi-word
tokens
. parse_sentence() can legally be called from do_* callbacks now
---------------------------------------------------------------------------
(v21.5b10 promoted to v21.5 with the following fixes)
* fixed a crasher in deallocating classes with PACKAGE_MUDLIB_STATS
defined
* fixed a crasher in read_file() when reading just a portion of the
file
Tue Jul 18 01:39:53 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.5b10
* made push_lvalue_range() non-static; it's used in LPC->C code
* fixed a crasher in the F_SHORT_STRING fix
* fixed a crasher in ed messages when <varargs.h> is used
Sat Jul 15 14:30:37 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.5b9
* Fix from Marius: define the crdir_fopen() function even if BINARIES
isn't defined
* fixed an LPC->C code generation bug that would change the third
character of the first function to a ')' in some files
* in LPC->C compilation, generate 'case x:;' and 'default:;' instead
of 'case x:' and 'default:' to satisfy picky compilers.
Thu Jul 13 01:52:40 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.5b8
* another F_SHORT_STRING string disassembly fix from Sym
* fixed a bug that caused classes to fail to restore from inside
classes/arrays/mappings
* fixed a bug that caused 'Inherit chain too deep' errors when
load_object() was called repeatedly on non-existent objects
* fixed the alpha crasher fix (I changed the wrong #ifdef merging
the changes into my local version)
* (: foo :) when foo is both a local function and a simul now gives
a pointer to the lfun, not the simul in line with the normal
visibility rules.
Thu Jul 6 03:29:37 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.5b7
* fix from Symmetry: use EXTRACT_UCHAR when disassembling
F_SHORT_STRING
* Alpha crasher fix: make sure there isn't any padding at the end
of the shared string and malloc'ed string structs so that they
overlap correctly
* don't use bison on AIX when compiling with xlC; xlC won't grok
bison's use of alloca
Fri Jun 30 16:02:44 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.5b6
* fixed a typo in socket_efuns.c
* fixed a bug that caused socket callbacks not to be called
(introduced with the v21.5b4 fix)
Tue Jun 27 01:01:27 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.5b5
* fixed MUD mode sockets to not abort the driver if they can't
allocate enough memory for a packet (due to garbage for the
length field, for example). The connection is killed instead.
Thu Jun 15 09:00:09 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.5b3
* fixed a LPC->C compilation error in range switch tables
Mon Jun 12 04:03:58 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.5b2
* fixed a bug that generated incorrect code for string switches
* fixed a typeclash in grammar.y.pre
* better fix for the obj/grammar.tab.o problem
* fixed a crasher in some assignments in LPC->C code
* fixed a crasher in casting a void valued efun, i.e.
return (int) destruct(ob)
Sun Jun 11 04:04:49 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.5b1
* updated with respect to v21.4
* fixed a bunch of warning messages and unsafe casts for DEC Alpha
* added a call_stack() efun: for each internal as well as external
frame:
. call_stack(), call_stack(0) : an array of program names
. call_stack(1) : array of current objects (in that frame)
. call_stack(2) : array of function names
. call_stack(3) : array of what origin() returns in that frame
all the arrays are the same length; i.e. element 5 of call_stack(0)
is the same frame as element 5 of call_stack(2)
e.g. printf("I got called from %s in %s", call_stack(2)[0],
call_stack(0)[0]);
---------------------------------------------------------------------------
(v21.4b11 promoted to v21.4 with the following fix)
* fixed a fatal error in compiling switches to C
Sat Jun 3 05:24:08 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.4b11
* added Magician's fixes for pluralize() for words starting with 'b'
and 'c'
Thu Jun 1 14:28:38 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.4b10
* added support for w->x[y] = z, w->x[y]++ and w->x[y]-- which was
accidentally omitted (generated 'Illegal lvalue' errors)
* fixed a crasher in strings longer than 64k
* fixed a compilation error crasher in x->foo when x was not a class
Thu Jun 1 01:52:29 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.4b9
* removed a kludge to help older TMI-2 libs run. It was causing
delayed crashers after errors in add_actions(), as well as
screwing up the previous_object(-1) lists. I also think it was
the cause of the unresolved backend crashers around v21 or so.
* COMPAT_BUSTER: due to the above, previous_object() is now zero
in add_action()'ed functions
Mon May 29 21:24:55 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.4b8
* fixed a bug that caused str[n..] to return garbage if n > strlen(str)
* fixed a compilation problem with LPC_TO_C off
* fixed a bad free() in edit_source that would cause defines to
mutate
* fixed a crasher when a socket which had a function pointer as a
callback was closed and then reused
* if a unbound socket is marked as needing a read, print an error
message and close it. This shouldn't ever happen, but evidentally
does and causes the CPU to race
Sat May 27 00:39:07 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.4b7
* fixed a crasher in set_privs(ob, 0)
* fixed a crasher in message() when the destinatination was a string
that doesn't correspond to an object
* fixed a crasher in calling a function pointer in a swapped object
* fixed a bug that caused ./build.MudOS not to find 'make' if GNU
make wasn't installed
* fixed a bug in disassembling F_BBRANCH_LT and F_NEXT_FOREACH
(print as backwards, not forward branches)
* removed the switch stack; explicit jumps are used instead
* fixed a crasher in exiting foreach() via break or return
* cleaned up duplicated code in code generation for loops
* added support for verifying string lengths in md.c
Tue May 23 17:59:12 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.4b6
* fixed a bug in the makefile dependencies that would cause the
wrong system libs to be used
* fixed a bug that when using older versions of gmake would give the
'can't find rule for obj/grammar.tab.o' error
* updated the Makefiles, fixed a uid compilation problem
* removed some extra newlines at the end of compiler error messages
* fixed a crasher in calling functions() on a swapped object
* fixed a crasher in pluralize()
* fixed a bug that would cause parse errors/wierd error messages
if (: was followed by the end of a line
Mon May 22 15:54:58 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.4b5
* fixed a read_file() to set it's return string size correctly
* fixed a bug that would kick you out of ed if you went to line -1
* fixed pluralize() to make sure the string is zero terminated
* swapped two include files to prevent a definition of SE in
<sys/sema.h> from overriding one in <arpa/telnet.h>; it causes
terminal negotiation to freeze output to the user
* fixed a bug that caused type errors when inheriting from a
program with #pragma save_types on which was saved as a binary
* upped the default optimization to -O3 on xlc
* fixed a crasher in assigning to indexed strings/array ranges
in LPC->C
* added LPC->C support for parse_command() and evaluate()
* put the O_CLONE flag back on replace_program() objects; makes
memory accounting efuns return slightly more intuitive values
and makes some sense. This means you can't clone from such an
object, but that's probably good since create() isn't what you'd
expect anyway
* fixed a bug that generated illegal LPC->C code if the last case in a
switch was empty (it's legal in LPC but not ANSI C)
* fixed a few typos in LPC->C code generation (i -> lpc_int)
* made the error messages for trying to compile something with a
functional more friendly (LPC->C)
Sun May 21 09:17:34 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.4b4
* fixed a bug that caused void and int to be compatible
* modified svalue_to_string and sprintf() to use outbuffers
* modified bad_argument() to use an outbuffer
* modified *_IN_TRACEBACK to use an outbuffer
* fixed outbuffers to truncate if they accumulate > 64k of data
* fixed a bunch of lines in interpret.c which had a space
before the preprocessor directive
* cleaned up some more uses of MSTR_SIZE/COUNTED_STRLEN to fix
more bugs with > 64k strings
* fixed a typo in the string switch type error message
* optimized STRING_STATS a bit (only do COUNTED_STRLEN() once
in places)
* set up an error context before calling flag() in master in case
there is an error
* check_memory() updated to know about the new malloc'ed strings
* optimized/cleaned up break_string()
* fixed a memory leak due to functionals keeping programs loaded
* fixed a memory leak in throw()
* fixed a memory leak in [0..n] array ranges on ref 1 arrays
Sat May 20 10:21:58 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.4b3
* fixed a syntax error with PACKAGE_MUDLIB_STATS defined
* fixed a name conflict between mudlib_stats.c and
packages/mudlib_stats.c; moved all mudlib_stats and uid
code into the appropriate package file
* removed some remaining DEALLOCATE_MEMORY_AT_SHUTDOWN code
* edit_source -configure now checks for POSIX getcwd()
and <crypt.h>
* fixed a problem where edit_source would abort if a define
in local_options and options.h had different values (it died
with a redefinition warning)
* fixed a lot of compilation warnings under SGI cc -fullwarn
* fixed a bug where __PORT__ was zero when specifying sockets
as external_port_x
* fixed a crasher in socket_address()
Wed May 17 15:30:02 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.4b2
* fixed a typo in cfuns.h
* fixed a problem with parsing of line specifiers which was causing
search in ed to not return an error message; changed the message
for search failing
* fixed a bug in SVALUE_STRLEN that would cause crashes involving
strings longer than 64k
Tue May 16 16:24:21 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.4b1
* fixed a typo in dumpstat.c
* added Aragorn's patch to allow %x to specify hexidecimal in sscanf
* partially rewrote the contrib efun pluralize() to handle 'X of Y'
and clear up some crashers, as well as be a bit more robust/efficient
* cleaned up some gcc -Wall warnings
* include configure.h as well as std_incl.h when testing for existence
of header files. This fixes misconfiguration bugs with respect to
rusage() on some systems (SunOS, ...)
* fixed outbuf_t compilation problems in comm.c on some systems
* fixed a crasher when explode() returned an array that was truncated
due to max_array_size
---------------------------------------------------------------------------
(v21.3b5 promoted to v21.3)
Mon May 15 02:46:10 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.3b5
* fixed a bug in code generation for addition; would produce incorrect
code when LHS was a number or real and RHS was type string or mixed
Sat Apr 29 23:59:18 CDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.3b4
* fixed a badly indented line in the GNUmakefile
* fixed some OBJDIR rules to not cause recompilation of a lot of
things when using GNU make
* fixed generation of package Makefiles
Sat Apr 29 23:59:18 CDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.3b3
* fixed some compilation problems with LPC->C off
* put in a better fix for the function pointer to void valued efun
problem; a debug check was erroneously misfiring on the old
version
Sat Apr 29 20:58:34 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.3b2
* changed some #ifdef 0's to #if 0's
* added a check to use -m486 when compiling using gcc on a 486
* fixed a crasher in self-destructing objects with call_outs
* fixed a crasher with calling reload_object() when call_outs to
function pointers existed
* fixed TELOPT_NAWS to properly handle zero bytes; caused the
reported screen size to be in the 17000 range
* fixed a typo in the missing type for argument fix in v21.3a7
* fixed a macro that xlc complained about in lex.c
* fixed a crasher in calling reload_object() on a swapped object
* fixed a crasher in reclaim_objects() with function pointers with
no arguments
* fixed a crasher in edit_source
* #ifdefs are now legal in packages/*_spec.c
* updated 'Install' with the new build procedure
Sat Apr 29 20:58:34 EDT 1995 (tim@handel.princeton.edu (Beek))
* v21.3a7 promoted to v21.3b1
---------------------------------------------------------------------------
(v21.2b8 promoted to v21.2)
Sun Apr 23 21:02:52 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.2b8
* fixed a crasher in efun pointers to void valued efuns
Fri Apr 21 01:17:24 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.2b6
* fixed a bug that caused all the login ports to try to use the
same port number
* fixed a crasher in PORT_ASCII objects which dest themselves
from inside process_input()
* fixed a crasher that would occasionally generate the wrong
code for default arguments
* Added a missing \n in the Function is no longer a simul error
message
* Fixed a crasher in short style efun pointers (: efun, x, y ... :)
* changed a bunch of references to 'vec' in the source to 'arr'
since we no longer refer to arrays as 'vectors' :)
* removed the port_number variable internally, the external_port
array is used everywhere now
* changed the -p switch to change the port of external_port_1
* added some missing commas in the %O output of function pointers
* the refs() contrib efun was updated to know about classes
Tue Apr 18 17:30:51 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.2b5
* fixed a crasher in bind() on lfun/efun/simul function pointers
with no arguments
* fixed a crasher in call_out_list() when a destructed object
had a pending call_out
* fixed socket_address(object) to convert the port number back
to host byte ordering (from Angus@Styx)
* fixed a crasher after compiling 256 break statements
* fixed a typo in the GNUmakefile
* sockets are now initialized before preload() is called, so it
is possible to preload socket objects
* name the spec files in packages/ foo_spec.c instead of foo.spec
to avoid a gcc bug
* include all include files found up to a given point in include
tests; <sys/resource.h> needs <sys/time.h> etc.
* make sure TELNET_NAWS is in <arpa/telnet.h>, otherwise use the
one in the MudOS distribution
* fixed a crasher in set_hide() when the valid_hide() check failed
* fixed a crasher in anonymous functions with empty blocks
function(...) {}
* removed some unnecessary prototypes from lint.h; the configuration
process now finds the appropriate includes
* mark the config strings as being correctly allocated for
check_memory()
* fixed a bug where Linux 1.2.1 sometimes returns ECONNREFUSED
from recvfrom() for UDP sockets when an ICMP_PORT_UNREACHED
is generated internally
* fixed a compiling problem with pluralize() in the contrib packages
Tue Apr 18 17:30:51 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.2b4
* fixed a crasher in calling process_input() for PORT_BINARY and
PORT_ASCII ports
* fixed a bug that would cause the first efun (alphabetically) to
do nothing
* fixed a typo in a call to debug_message_with_location() in simulate.c
* fixed a mis-optimization of typeof() that caused it to always
return 2
Mon Apr 10 23:41:33 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.2b3
* added Symmetry's fix so OSFLAGS gets propagated to the makefiles
for AIX
* fixed a typo in the v21.1b2 regexp fix
* removed an unused definition of find_status()
Fri Apr 7 12:57:56 EDT 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.2b2
* reorganized the v21.1 changelog entry to be more concise/clear
* fixed a bug that would loop forever on partial matches when
finding a substring in a string using strsrch() search from the
right
* fixed regexp.c on Linux to define CHARBITS correctly; won't work
on systems with non-8 bit chars, but the old code seems to be
incorrect.
* Next two fixes from Aragorn:
. Fixed the warning about too many file descriptors to print the
right number
. Don't free 'use_name' until after we're done using 'file' since
'file' may point at it
Fri Mar 24 20:59:08 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.2b1
* v21.1b7 promoted to minor version status (see v21.1)
* Changed the error message for arr[<x] where x > sizeof(arr)
to 'Array index out of bounds' from 'Negative index passed to array'
* fixed the extern definition of control_stack[] to use MAX_TRACE
instead of being hardcoded at the default
------------------------------------------------------------------------------
Fri Mar 24 20:59:08 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.1b7
* fixed a crasher in call_out(function, ...) when NO_SHADOWS
was off
* removed a check for destructed objects in get_all_call_outs()
which:
(a) didn't remove destructed objects
(b) removed call_outs to function pointers
* added in the config file crasher fix from v21.2a5
Sun Mar 19 21:02:18 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.1b6
* fixed a small optimization to backend.c introduced in v21.1b5
* fixed a typo in the v21.1b5 type check fix
* Robocoder: in config files, restrict sscanf() matches to
the current line
Sat Mar 4 16:04:43 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.1b5
* some amiga compilation fixes from Robocoder
* added some smalloc accounting fixes from Robocoder; keep track
of wasted memory as well as some more overhead
* cast IAC to signed char so it is compatible with (char *)
* fixed a bug in type checking that made (class foo *) and (mixed *)
incompatible
* fixed a typo in the prototype fix from v21.1b4
* defining two classes with the same name is now an error
* compilation fix for IRIX 5.3 (needs <unistd.h>)
Sat Mar 4 16:04:43 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.1b4
* fixed a ref count error/crasher in inherits()
* fixed a ref count bug/memory leak (programs referenced by
non-bindable function pointers weren't being freed)
* added code to check function references to programs in md.c
* added in some porting fixes for CX/UX 6.1 (from DrFeelgood)
* fixed a memory munger in regular expressions that end in '\'
* fixed inherited functions which have a different number of
arguments than the functions they replace to not error
Sat Mar 4 16:04:43 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.1b3
* removed some debugging code in lex.c
* fixed a typo that prevented #undef OLD_ED compilation
* fixed compilation with #undef NO_MUDLIB_STATS
* fixed a crasher in type errors in division
* fixed a division error message that erroneously indicated
multiplication
* fixed a type error in locals used in anonymous functions
* fixed a bug in smalloc accounting
Wed Mar 1 18:14:01 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.1b2
* brought up to date with respect to v21c1
Wed Mar 1 18:14:01 EST 1995 (tim@handel.princeton.edu (Beek))
* raised patchlevel to v21.1b1
* added three new contrib efuns:
. int replaceable(object) - returns 1 if the object consists only
of a create() function and inherited functions
. void program_info() - prints out memory usage of programs
divided up by sections
. string upper_case() - similar to lower_case(), but ... :)
* fixed a crasher in the compilation of switch tables to C code
* rewrote/optimized some of the compile time type checking
* void efuns no longer return a value on the stack. If the value
is used (c.f. x = write("hi")) a F_CONST0 instruction is inserted
* removed the errorp() efun
* added a customized allocator for mapping nodes. This should reduce
memory usage and increase speed of additions/deletions
* removed RUNTIME_LOADING from options.h
* fixed a crasher involving finding duplicate objects in the hash
table with LPC objects precompiled and DEBUG on
* changed program ref counts to shorts