Version 1.0.0, 5 February 1991 -- First public release. Version 1.1.0, 8 February 1991 -- recycle(x) now calls x:recycle() before destroying the object. -- the little-used and functionally-dubious update() function has been removed. **** NOTE: Technically, this change should make this a major release, since **** it can break old code that depended on update() existing. In **** practice, however, only $prog:@update used it (in LambdaCore.db, **** anyway). Thus, to install this server, you should first @rmverb **** $prog:@update, since its code would otherwise be syntactically **** illegal. -- added renumber() and reset_max_object(), to facilitate automatically generating new versions of LambdaCore.db from the current LambdaMOO db. -- made it so that only wizards can set the .name property of players; code in the database now checks that players keep unique names. -- added set_player_flag() and boot_player() to enable registration-style player creation and @toad-ing. Version 1.1,1, 12 February 1991 -- Much rearrangement of #include's to come as close as possible to conformance with ANSI C and POSIX. -- Many changes to improve portability, especially to sites with only 'nearly' ANSI-compliant compilers. -- Added backtrace-printing for tasks aborted for running too long. Version 1.1.2, 13 February 1991 -- Fixed nasty infinite recursion problem on server panic. -- Fixed the implementation of renumber() to really do the job. It's now much more expensive (linear in the size of the database rather than constant time), but at least it's correct. It's only intended for use on a tiny DB anyway. Version 1.1.3, 11 March 1991 -- Fixed benign unparsing bug that always parenthesized unary-minus expressions. -- Changed move() to insert the newly-moved object as the last item in the contents list, rather than the first. -- Fixed a less-than-benign bug in the unparsing of conditional expressions, where 'x ? y | (z ? w | v)' was being unparsed without the parentheses. -- Fixed a bug whereby an extra checkpoint timer was created every time the '.dump' command was used. -- Added printing the value of 'this' to error backtraces whenever it is different from the verb location. -- Fixed bug in propagation of property ownership on object creation. -- WARNING: I believe that there is a storage leak somewhere in the server. If so, it is not very fast. With any luck, I'll find it soon. For safety's sake, check on the size of your server's address space and reboot the server every few days if it's getting large. Version 1.2.0, 11 April 1991 -- Fixed bug in parser that made 'after' a reserved word; this was part of an old (and never completely implemented) alternative syntax for the 'fork' statement. -- Added memory_usage() function, to help in tracking down any memory-leak bugs, among other things. -- Removed ancient restrictions on players recycling themselves. -- Changed recycle() so that children of recycled objects are not themselves recycled. Instead, they are adopted by their former grandparent. -- Fixed bugs in which errors are returned from move() in certain cases. -- Removed useless error value E_BADMOVE. -- Fixed bug in move() that allowed the enterfunc to be called even if the object being moved was no longer valid. -- Changed several uses of E_INVIND to E_INVARG; E_INVIND is now only used when an attempt is made to fetch or store the value of a property or verb from an invalid object. -- Changed several functions to improve consistency in the order in which certain kinds of errors are checked for. -- Made the allocator keep track of statistics concerning memory usage by type of object being allocated. -- Changed queued_tasks() to return the value of 'this' in each task as well. This new value is returned at the end of the sublist for each task, so old code should continue to work unless it depended upon the exact length of the sublists. -- Changed set_task_perms() not to return E_PERM if the argument is the same as the current permissions of the verb (i.e., if the change in permissions would be a no-op). -- Made tonum() and toobj() more robust in the face of extra spaces before or after tokens. Version 1.3.0, 7 June 1991 -- Removed the now-useless `.help' and `.mem' commands, and all code concerning the also-removed CHECKMALLOC option. -- Fixed allocation bug in inheritance of 'dobjstr' and its ilk, when the current value of the variable has the wrong type (e.g., 'dobjstr' is a list when another verb-call happens). -- Fixed the timers implementation not to try to free storage inside an interrupt procedure. -- Changed move() so that :accept is called even if it's a wizard doing the moving; of course, the wizards gets to do the move even if :accept returns false. -- Added 'off/off of' as a new preposition. -- Fixed the built-in parser to return #-1 ($nothing) for the empty string, rather than #-2 ($ambiguous_match). -- Fixed a few bugs in how the allocator keeps track of types. -- Fixed two memory leaks: 1) If you returned or aborted from inside an iteration over a list, the list was never freed. 2) If you killed a task, its associated environment was never freed. -- Added ticks_left() and seconds_left(), for determining how much longer the current task will be allowed to run. -- Reduced the maximum number of seconds for each task from 60 to 15. -- Removed restriction that the first argument to notify() and boot_player() must be a valid player object. -- Removed the .shutdown and .dump commands. -- Added shutdown() and dump_database() functions to replace their respective built-in commands. The shutdown() function takes a single string argument that is printed to all players as their connections are closed. -- Added output_delimiters() as a way to discover the current PREFIX and SUFFIX strings on a given connection. -- Completely rewrote and modularized the network interface (now in bsd_network.c), the main loop, connection-management, and checkpointing code (now in server.c), and the task dispatcher (now in tasks.c). -- Eliminated the notion of `clocks' associated with tasks. All tasks, including forked tasks, begin execution with a full complement of ticks and seconds. To make this reasonable, the task dispatcher maintains a separate queue per player, with command tasks and ready forked tasks intermingled. Thus, a malicious player who forks many tasks can only clog his own queue. -- Eliminated the `command burst' heuristics from the dispatcher; this always seemed like just a patch for having a fair dispatcher anyway. -- Added a timeout for closing idle un-logged-in connections. -- Removed the QUIT command; the boot_player() function can now be used for that purpose. boot_player() now does a normal disconnect, including running the disfuncs. At the same time, I made it allow players to boot themselves. -- Because clocks have been eliminated, the lists returned by queued_tasks() have changed somewhat, though in a mostly-compatible way. The clock ID and clock ticks fields are both constant now, with the ticks equal to 20,000. -- ****************** All processing of commands typed by un-logged-in players *** VITAL NOTE *** has been moved into the database. Every un-logged-in ****************** connection is assigned a unique negative player number. The functions notify() and boot_player() can be used by wizards to send text to and to terminate such a connection, respectively, using that player number as the connection identifier. Each line of input on such a connection is first parsed into words in the usual way and then these words are passed as the arguments to #0:do_login_command(). For example, the line connect Munchkin frebblebit would result in the following call being made: #0:do_login_command("connect", "Munchkin", "frebblebit") If #0:do_login_command returns a valid player object, then the connection is considered to have logged into that player. When the connection is first established, the null command is automatically entered, resulting in a call to #0:do_login_command with no arguments. This signal can be used by the verb to print out a welcome message, for example; this service is no longer provided by the server. Because of this change, it will be **NECESSARY** to add an implementation of #0:do_login_command to your database *before* beginning to use version 1.3.0 of the server. The Minimal.db database and the version of the LambdaCore database released concurrently with 1.3.0 already provide such an implementation. The following code can be used as a simple implementation that preserves most of the functionality of version 1.2.0: if (callers()) "This code should only be run as a server task."; return; endif if (args == {}) args = {"help"}; endif command = args[1]; args = args[2..length(args)]; nargs = length(args); if (command == "connect") if (nargs < 1) notify(player, "Usage: connect "); else name = args[1]; password = nargs >= 2 ? args[2] | ""; for p in (players()) if (p.name == name && (typeof(p.password) != STR || (length(p.password) >= 2 && crypt(password, p.password[1..2]) == p.password))) p.last_connect_time = time(); return p; endif endfor notify(player, "Unknown player or wrong password."); endif elseif (command == "create") if (nargs != 2) notify(player, "Usage: create "); else name = args[1]; password = args[2]; for p in (players()) if (p.name == name) notify(player, "That player name is already in use."); return 0; endif endfor new = create($player_class, $nothing); set_player_flag(new, 1); new.name = name; new.aliases = {name}; new.programmer = $player_class.programmer; new.password = crypt(password); new.last_connect_time = time(); move(new, $player_start); return new; endif elseif (strcmp(command, "QUIT") == 0) boot_player(player); else msg = $welcome_message; if (typeof(msg) != LIST) msg = {msg}; endif for line in (msg) if (typeof(line) == STR) notify(player, line); endif endfor endif return 0; Of course, the whole point of bringing this code into the database is to enable easy experimentation with new facilities for un-logged-in users, such as multiple Guest players, friendlier coping with similarly-named players, provision of a `who' command, etc. -- The network interface now copes more nicely with running out of buffer space for output to a player. Before, excess output was simply dropped on the floor; now, it instead attempts to push it out the network socket and only flushes output when that doesn't work. I was able to have a single command send me over 60,000 characters of output without flushing any. -- The PREFIX and SUFFIX strings are no longer printed either for the .program command or for any of the lines of program input. -- Recycled players with active connections are more promptly noticed and their connections closed. -- The numeric ID associated with forked tasks is now guaranteed to be the same as the value of task_id() in that task when it executes. -- The create() function now invokes the :initialize verb on the newly-created object before returning it. It is not an error if there is no :initialize verb defined on the object. -- Changed listinsert() and listappend() never to generate E_RANGE errors. Instead, the following expressions are now always equivalent: listinsert(list, element, index) listappend(list, element, index - 1) {@list[1..index - 1], element, @list[index..length(list)]} -- Changed the matching algorithm for verb names so that the old behavior of `*' matching anything is generalized to `foo*' matching anything beginning with `foo'. -- Changed eval() and set_verb_code() to require programmer permissions. -- Added a timezone abbreviation onto the end of ctime()'s result. -- Added min(), max(), and abs() functions. -- The messages printed for running out of ticks and for running out of seconds are now disinguishable. Version 1.4.0, 22 August 1991 -- Fixed a bug whereby running out of seconds might be misreported as running out of ticks. -- Fixed a bug whereby using `@' on a non-list in a verb's argument list could end up passing E_TYPE as the value of `args'. -- Fixed parsing bug that would ignore an extraneous comma at the front of an argument list. -- Changed random() to raise E_INVARG on non-positive numerical arguments, rather than E_TYPE. -- Fixed bug in handling of the case where a verb disappears between a `.program' and the corresponding `.'. -- Made it a type error to use <, <=, >, or >= either on operands of different types or on lists. -- Completely rewrote the MOO-code parser, unparser, and interpreter using a new program representation based on vectors of bytecodes instead of abstract syntax trees. This saves space and time and also allows the interpreter to be iterative instead of recursive. This, in turn, allows us to implement various operations that involve saving the entire state of a running task for later resumption. -- Added read() and suspend() functions, using the new capabilities of the interpreter. -- The new interpreter counts a few more ticks than the old one; in particular, every IF, ELSEIF, and FORK now counts a tick. In view of this, and in view of the improved performance of the interpreter, the ticks limit for tasks has been increased from 20,000 to 30,000. -- Added an indexed assignment statement for updating lists stored in variables and properties. -- Fixed bug in numeric comparisons where `-2147483647 > 1' was false but `-2147483647 > 2' was true. -- Fixed database dumping to detect and recover from filesystem errors. For normal periodic checkpoints, the dump attempt is simply abandoned. For panic and shutdown dumps, the server retries the dump from the beginning after waiting 60 seconds (during which time, presumably, the archwizard will fix whatever the problem was). In all cases, explanatory messages are printed in the log. This should help mitigate the problems that arise when there's no more room left on the filesystem on which the server is dumping. -- Enhanced the callers() function to return the verb-location and initial value of `player' for each caller. Version 1.4.5, 20 September 1991 -- Made unparser failures panic the server rather than try to press on. -- Changed the task scheduler in two important ways: 1) Effectively gave each player's input tasks a higher priority than their forked or suspended tasks. This should both improve interactive response and make it possible for users to recover from accidentally forking off many tasks; they can now run commands to kill those tasks and avoid having those commands wait behind all of the forked tasks. 2) The scheduler no longer uses a strictly round-robin algorithm for choosing which player's task to run next. Instead, it keeps track of how many seconds have been consumed by each player's tasks and runs the next task from the queue with the least total `usage'. Fresh queues and those queues that are empty when it becomes their turn to run again are kept apart from non-empty queues. When an empty queue next gets a task to run, its usage is set equal to the least usage of already-non-empty queues. Thus, such queues go to the head of the list but have no advantage over the queues that were never empty. This more complex scheduling scheme is intended to ameliorate the problems with some players having long-running tasks constantly runnable in the background; such players will now get only their fair share of the server. -- Fixed queued_tasks() to list the player being read from as the owner of a reading task. This makes it consistent with kill_task(), which only allows that player to kill such a task. -- Fixed a bug with my use of signal() on some POSIX-compliant machines. One possible consequence of the bug (actually manifest on the IRIS) is that all checkpointing processes after the first one will never be waited for, leaving `defunct' processes lying around until the server itself terminates. -- Fixed a small bug in my use of ANSI C: the `typedef enum Opcode Opcode;' line in opcode.h must come *after* the definition of `enum Opcode'; unlike struct types, enum types may not be forwardly referenced. Thanks to yduJ and the Lucid C compiler for finding this bug. -- Fixed a bug in the line-number-finding code; this bug was identical to one we fixed long ago in the unparser. Since the LNF code is *derived* from the unparsing code, this is unsurprising. What's embarassing is that we forgot to fix the LNF code at the same time... Version 1.5.0, 3 November 1991 -- Added connection_name(player) to allow wizards some level of reasonable control over whom they allow to connect to their servers. -- Changed read() to accept an optional argument, player, specifying the player from whose connection to read. If this argument is given, the check that the current task is an input task is not done. The only good use for this that I'm aware of is in conjunction with the next change... -- The new compilation option OUTBOUND_NETWORK enables a new built-in function called open_network_connection(). The arguments are specific to the network module in use, but the intent is that they somehow specify a remote place to which a network connection can be made. For the default, BSD networking module, there are two arguments, a host-name string and a port number. The function creates a connection to that place, if possible, and tricks the rest of the server into believing that a normal connection just came from that place. Open_network_connection() returns the negative object number that is the pseudo-player object for that connection. One can then use read() and notify() to send and receive lines of ASCII text on that connection. Of course, if the server should ever receive a line of input when there's no task read()ing from the connection, that input will be treated as a command in the usual way. Also, unless the connection `logs in' at some point, the usual five-minute timeout-if-no-input-received connection shutdown will take place. The usual practice is to have the same task that opens the connection go into an infinite loop read()ing from it. As usual, boot_player() can be used to shut down the connection. This function can only be called by wizards, of course. If the OUTBOUND_NETWORK option is not used, then open_network_connection() still exists, but always raises E_PERM. -- Changed the bsd_network module to map tab to space on input, rather than simply dropping tabs on the floor. Version 1.5.1, 12 November 1991 -- Fixed bug in checkpointing whereby the server's command log was flushed even if the checkpoint failed. -- Various minor tweaks to make the server work on an IBM RS/6000. Version 1.5.9, 6 December 1991 -- Gave different ticks/seconds limits to non-input tasks. They get only 15,000 ticks and 5 seconds. -- The task seconds limit is now measured in server CPU seconds (as opposed to wall-clock seconds) on systems that can support it. -- Adding some simple object-hierarchy checking and (in very simple cases) repair code to the DB reading process. If unfixable errors are found, the server refuses to run. -- Fixed minor inconsistency in the log format. -- Fixed a few minor memory leaks. -- Fixed a bug whereby the bottom frame of a suspended task that was read from the database file could be given the wrong size runtime value stack. -- Made the DB-reading code a bit pickier about the correctness of its input. -- Tweaked create() to make it more properly initialize the new object's location and contents. -- Fixed bug in move() that could be tickled if an `accept' verb recycled either the movee or the destination. -- Fixed several bugs in recycle() that could (and did!) corrupt the contents hierarchy in the database. The implementation is now much simpler, too! Version 1.5.14, 3 March 1992 -- Various minor tweaks to make the server work on a Macintosh running A/UX. -- Tweak to work around bug in isgraph() on systems that default to signed characters. -- Fixed bug in boot_player() that didn't allow players to boot themselves. [I know it says under release 1.3.0 above that boot_players() already worked this way, but as far as I can see, that was a lie!] -- Fixed bsd_network startup sequence so that the server does not `listen' on its port until after the DB has been read. Connections during the DB-reading phase will most likely time out. I think this is better than the old situation, in which those connections succeeded and then hung without printing anything until the server finished reading in the DB. -- Changed behavior of server when there aren't enough file descriptors to accept a new connection. The server now arranges to stop just short of this point and to immediately close any new connections after printing an explanatory message. -- Fixed properties(), verbs(), x.contents, and queued_tasks() to run in time linear in the length of their results, rather than quadratic. Version 1.5.20, 24 July 1992 -- Added check for an empty string as the second argument to strsub(); this now raises E_INVARG. -- Conditionalized the keeping of the command log; it is now controlled by the LOG_COMMANDS symbol. -- Removed a number of causes for GCC 2.1 to issue warnings about server code. (Version 1.5.15) -- Fixed an ordering bug whereby the task queue for a newly-connected player was not fully initialized when the :confunc task(s) were run. In particular, the new, connected queue was not marked as belonging to that player, and certain functions like output_delimiters() would always fail (not that that's a very interesting function to call in a :confunc, since it's guaranteed to return {"", ""} then...). (Version 1.5.16) -- Finally fixed the bug whereby it was possible to use chparent() to create a situation in which an object defines a property with the same name as one defined by one of its ancestors. Such an attempt now results in chparent() raising E_INVARG. (Version 1.5.17) -- Finally fixed the bug whereby add_property() allowed the definition of a property in a parent that was already defined in a child. With this change, it should no longer be possible to create a situation in which a parent and child define properties with the same name. (Version 1.5.18) -- Eliminated some minor warnings from the Dell SVR4 and HP/UX 9000 compilers. (Version 1.5.19) -- Added logging of all instances of x.wizard = 1. -- The very first task run on booting the server is now a server task calling #0:server_started() with player == #-1. This task runs even before the server gets the value of $dump_interval in order to schedule the first checkpoint. (Version 1.5.20) Version 1.6.0, 13 August 1992 -- Changed MOO-code parser to generate the old version 1.3 program- representation based on abstract syntax trees. These syntax trees are then translated to vectors of bytecodes. This is much cleaner and easier to understand than the old method; perhaps surprisingly, it also appears to be significantly faster, too. (Version 1.5.21) -- Added support for compilation on MIPS SVR3 machines. -- Added support for a remote `checkpoint-request' signal (SIGUSR2), analogous to the existing `shutdown-request' signal (SIGUSR1). -- Changed unparser to first convert bytecodes to trees and then use output routines from version 1.3 of the server. Same for line numbers. -- Error tracebacks now include line numbers for all frames on stack. (Version 1.5.24) -- Cleaned up interpreter code. -- Added sqrt() and server_log() built in functions (Version 1.5.25) -- All built-in function registration procedures are now located in one list, so it is easier to add new ones. -- Added match(), rmatch(), and substitute() built-in functions to handle regular expression searches. (Version 1.6.0) Version 1.6.1, 14 August 1992 -- Fixed off-by-one bug in GNU regexp searching code. (*sigh*) Version 1.7.0, 23 October 1992 -- Fixed behavior of index(), rindex(), match(), and rmatch() on empty pattern and/or subject strings. (Version 1.6.3) -- Added the prefix `> ' to all log messages generated by the server_log() built-in function, so that they can be distinguished from server-generated messages. -- Added a compile-time option (OUT_OF_BAND_PREFIX) enabling a method of entering commands that bypass both normal command parsing and any pending read()ing task. If OUT_OF_BAND_PREFIX is #define'd (in config.h) as a non-empty string, then any lines of player input that begin with that prefix will be parsed into a list of words and those words passed as arguments in a server task invoking #0:do_out_of_band_command. This is intended for use by fancy MOO clients that need to send reliably-understood messages to the server, such as window-event notifications. -- Added output to the `.program' built-in command, stating that it is obsolete and will be replaced `soon' (i.e., in the next release). -- Rearranged the per-system configuration procedure for compiling the server; now it should be easier for people to understand just what needs to be done for their particular system. The user changes in the Makefile are now confined to specfying details of the compiler; all other kinds of options are handled by editing the `config.h' file, which has been radically reorganized. -- Renamed a few files (notably parse_command.[ch]) to fit within the 14-character limit imposed on certain systems. (Version 1.6.4) -- Implemented reference counting on string and list values, replacing the old, CPU-intensive deep copying method of storage management. -- Added subrange assignment for strings and lists. Added indexed assignment for strings. (Version 1.6.5) -- Fixed bug whereby a read()ing task with no more input to consume was never resumed if the connection being read was subsequently closed, from either side. Now the read() call raises E_INVARG, just as it would if the read() were begun when the connection was already closed and no more input was left to consume. -- Added TYPE_CLEAR value to properties, which cause property lookup on the parent. Added built in functions `clear_property()' and `is_clear_property()' to assist in TYPE_CLEAR property manipulation. -- Added hash-lookup for properties and changed property definition representation to be arrays instead of linked-lists, both to speed up property lookup. (1.6.6) -- Added support for multiple complete networking implementations. The first use of this flexibility is a new SINGLE_USER option, which creates a version of the server that accepts only one connection at a time and uses the server's own standard input and output streams for it. -- Added a new built-in property on objects, the `f' (for `fertile') bit; it replaces the use of the `r' bit to allow children to be made of the object by either create() or chparent(). That is, now those operations check that the `f' bit is set and disregard the setting of the `r' bit. (Version 1.6.7) ******** Before upgrading an existing MOO to use this version of the server, * NOTE * you should make sure that no object in your database already has a ******** property named `f'; the following MOO program, run by a wizard, will perform this check: for i in [0..tonum(max_object())] o = toobj(i); if (ticks_left() < 1000 || seconds_left() < 2) notify(player, tostr("Checking ", o, " ...")); suspend(0); endif if (valid(o) && "f" in properties(o)) notify(player, tostr("*** ", o.name, " (", o, ") has an `f' property!")); endif endfor After eliminating all such properties from the database and restarting with this version of the server, you should, as a wizard, run the following MOO code to initialize the `f' bits of all of the objects: for i in [0..tonum(max_object())] o = toobj(i); if (ticks_left() < 1000 || seconds_left() < 2) notify(player, tostr("Fixing ", o, ".f ...")); suspend(0); endif if (valid(o)) o.f = o.r; endif endfor -- Reorganized the files so that all database modifying procedures were in one of four modules. (Version 1.6.8) -- Reorganized existing network protocol and multiplexing wait implementations into a pluggable modular form. (Version 1.6.9) -- Fixed bug in the interpreter that could pass a garbage program counter to the line-number-finding code, used in printing error tracebacks. The l-n-f code responded semi-robustly by printing a message in the server log and returning a line number of zero. (Version 1.6.10) -- Fixed bug whereby passing negative numbers to random() failed to evoke an E_INVARG error. -- Fixed a bug in MOO-code compilation that led to real nastiness if there were more than 255 literals in a single verb. -- Incorporated several new networking implementations, allowing for use by a single-user (using the standard input and output streams of the server itself), and by multiple users on either System V or BSD-style UNIX systems, either with or without TCP/IP networking. (Version 1.6.11) -- Fixed a bug in built-in function management that caused max() and min() to always raise E_TYPE on some systems. -- Added a new automatic configuration system to the server distribution, so that people do not, in general, need to know much of anything about their local system in order to compile the server. (Version 1.6.12) -- Fixed a misfeature of substitute whereby it was not possible to include a percent-sign in the string that was to survive into the output. Now, `%%' is replaced by `%' in the output. (Version 1.7.0) Version 1.7.2, 3 August 1993 -- Made tabs input as themselves, rather than being mapped to spaces. This is necessary in order to communicate with Gopher servers from within the MOO. -- Reduced the default `seconds-left' limit for task; forground tasks went from 15 to 5 seconds and background tasks from 5 to 3. -- Fixed regular-expression matching to respect task seconds limits, aborting the match. -- Fixed bugs whereby both kill_task() and queued_tasks() would miss reading tasks that had no ready input (i.e., whose associated task queues were empty). -- Fixed memory leak when a read() call fails with E_INVARG; the copied interpreter stack was never being freed. -- Fixed possible free of the null pointer during object recycling. -- Fixed a number of places that generated warnings in newer versions of GCC. -- Fixed inconsistency of the behavior of strcmp() between different server machines; now it always returns one of 1, 0, or -1. -- Added a five-second timeout to outbound connection attempts, aborting them if the timer expires. -- Made the error message to users about network buffers overflowing and lines of output being flushed much more understandable. -- Added log entry for each time the server refuses a connection due to being full. -- Fixed premature-free bug in preposition matching. -- Bowing to popular pressure, removed warning from `.program' command about its (no longer) imminent demise. -- Added more error checks to options.h to make it easier for installers to diagnose problems. -- Fixed a bunch of auto-configuration problems encountered by installers. -- Fixed a bug in substitute() where certain successful match() results were rejected as invalid. -- Removed restrictions that only player objects could be used in place of `x' in the following: object.owner = x add_property(object, pname, value, {x, perms}) set_property_info(object, pname, {x, perms}) add_verb(object, {x, perms, names}, args) set_verb_info(object, vname, {x, perms, names}) set_task_perms(x) That is, non-players may now own objects, verbs, and properties, and running tasks may take on the authority of a non-player. -- Fixed bug where it was possible for a task to run out of seconds during a call to set_verb_code() and have a truncated version of the program installed as the new verb code. Now, either all of it gets installed or none of it. -- The log messages produced by successful open_network_connection() calls are more useful now, containing the hostname and port of the connection. -- Outbound network connections are no longer susceptible to connection timeouts due to a lack of input. -- Fixed a bug whereby verbs with numeric names screwed up DB loading, with the wrong code getting put into some verbs and others getting no code at all. Version 1.7.3, 4 August 1993 -- Outbound connections no longer get a spurious blank line of input as soon as they're created. -- Fixed stupid bug introduced in fixing bug in substitute(). -- Fixed some bugs in the Makefile and the configuration script. Version 1.7.4, 10 August 1993 -- Fixed more configuration problems and warnings from GCC. Version 1.7.5, 12 August 1993 -- Added more log entries during loading, to make it clearer where the time is being spent. -- Fixed a long-standing denial-of-service attack vulnerability, in which a connection could keep the server from ever running any tasks at all by *always* providing ready input. Version 1.7.6, 11 October 1993 -- The built-in command parser now matches the (in)direct object string against the names of objects as well as their aliases. -- The server no longer calls :confunc or :disfunc directly when users connect or disconnect; instead, it calls the following verbs on #0 in the appropriate circumstances: :user_created(USER) When #0:do_login_command() returns USER, a valid player object whose number is greater than the value max_object() returned before the server called #0:do_login_command(). That is, when a new user has been created. :user_connected(USER) When #0:do_login_command() returns USER, a previously-existing valid player object for which no active connection already existed. :user_reconnected(USER) When #0:do_login_command() returns USER, a previously-existing valid player object for which there was already an active connection. :user_disconnected(USER) After the end of a task that called either boot_player(USER) or set_player_flag(USER, 0). :user_client_disconnected(USER) When USER's client unilaterally closes its connection to the server. ******** Before upgrading an existing MOO to use this version of the server, * NOTE * you should install the following verbs on #0 in order to remain ******** compatible with existing code; NOTE WELL that they should be installed with the `d' permissions bit UNSET: #0:"user_created user_connected" user = args[1]; fork (0) user:confunc(); endfork user.location:confunc(user); #0:"user_disconnected user_client_disconnected" user = args[1]; fork (0) user.location:disfunc(user); endfork user:disfunc(); -- Non-wizard users can now use the read() built-in function when they give a connected object they own as an argument. Thus, it is possible, for example, for a non-wizard to prompt itself for input. -- The match() and rmatch() built-in functions now cache some number of the most recently used patterns, somewhat speeding up the matching process. -- The server now calls #0:checkpoint_started() whenever it starts to dump a checkpoint of the database and #0:checkpoint_finished(SUCCESS) whenever the dump finishes, where SUCCESS is true iff and only if the checkpoint was successful. -- Fixed a bug (!) in the built-in command parser whereby in the command ="This is in quotes" and this is not the variable `verb' would correctly be set to "=This is in quotes" but the variable `argstr' would be "is in quotes\" and this is not" instead of the correct "and this is not". (Reported by Shirgall.) -- Fixed some compilation problems with the SYSV/TCP configuration on Solaris machines. -- The value of `argstr' when the server calls either #0:do_login_command() or #0:do_out_of_band_command() is now the raw command line as received from the connection, as opposed to the empty string. Version 1.7.7, 20 December 1993 -- Added some extra configuration tests and other changes to work around bugs in HP/UX. -- Fixed an output-timing bug in calling the #0:checkpoint_started() hook. -- Fixed bug in the SYSV/LOCAL client whereby some output from the server could be lost, never being printed. -- Added the server FIFO file name to an error message in the SYSV/LOCAL configuration. -- Added the first rudimentary support for in-DB command parsing. Each user command is broken up into words, a list of which is passed as the arguments in a call to #0:do_command(), if it exists, with `argstr' initialized to the raw command line typed by the user. If #0:do_command() does not exist, or if that verb-call completes normally (i.e., without suspending or aborting) and returns a false value, then the built-in command parser is invoked as usual to handle the command. Otherwise, it is assumed that the DB code handled the command completely and no further action is taken by the server for that command. Version 1.7.8, 30 September 1994 -- Many serious bugs in the MOO-code compiler (some server-crashing) were fixed by rewriting the code-generator portion from scratch. This should fix the problems some folks have had with panics due to `Unparse failed!'. -- As a side-effect of the support I wrote for debugging the new code generator, there is a new built-in function, disassemble(OBJ, VERB), which returns a (longish) list of strings giving the `assembly-language' listing for the compiled code of the verb VERB on the object OBJ. The permissions checks are the same as for verb_code(OBJ, VERB). I don't make any promises about how that output might change in later releases and I probably won't be writing any detailed documentation of the opcode set, etc. Think of it as a little puzzle gift, from me to you... :-) -- Added a number of new configuration tests, especially to aid compilation on the DEC Alpha running OSF/1 and any machine running NeXT Mach 3.1. -- The checkpoint process now changes how its command line shows up in the output of the `ps' command (on systems where that's possible), to make it clear what process it is. -- Added the first stages of a general new facility for customizing the operation of the server dynamically, from inside the database. At a number of times, the server checks for whether the property $server_options exists and has an object number as its value. If so, then the server looks for a variety of other properties on that $server_options object and, if they exist, uses their values to control how the server operates. There are only a couple of uses of this facility in 1.7.8; many more are expected in the future. -- As the first use of the above facility, the server uses the value of $server_options.fg_ticks and $server_options.fg_seconds (if they exist and are numbers) as the numbers of ticks and seconds allotted to command tasks and server tasks; the values of DEFAULT_FG_TICKS and DEFAULT_FG_SECONDS in options.h are used if the corresponding properties either don't exist or are not numbers. Similarly, the values of $server_options.bg_ticks and $server_options.bg_seconds are used for forked tasks and suspended tasks, falling back on DEFAULT_BG_TICKS and DEFAULT_BG_SECONDS. These properties are looked up anew every time a task begins or resumes execution. -- As the second use of the $server_options facility, on every call to a built-in function `foo()', if the property $server_options.protect_foo exists and is true, and the programmer is not a wizard, then E_PERM is raised. Thus, for example, to make the `chparent()' function wiz-only, simply set $server_options.protect_chparent to 1. -- Fixed a bug in the MOO-code unparser where any use of a string naming a MOO keyword as a property or verb name would be unparsed in such a way as to make the resulting code syntactically incorrect. Thus, for example, the expression x.("while") now unparses into the same code, instead of into x.while which won't parse properly. -- The server now uses a pair of additional (smallish) UNIX processes to perform all network host-name lookups. Under the new system, it can reliably abort stalled lookups without causing problems; in particular, this provides the real fix for the long-standing `VM exhausted' problem on NeXT machines. -- A few small storage leaks were fixed. -- A number of new progress-report log messages were added to the DB loading process. Version 1.7.8p1, 3 October 1994 -- Fixed a file-descriptor leak in the new name-lookup code; if the server has enough name-lookup timeouts, killing the name-lookup worker process over and over, then the name-lookup intermediary will eventually run out of file descriptors and fail, shutting down name-lookup services until the server is rebooted. -- Fixed a minor portability problem in the new code generator; believe it or not, there are still C compiler's out there (on Ultrix, for example) that don't implement initializers for locally-declared structures... -- Fixed a bug in disassemble() whereby it couldn't find !x verbs. Version 1.7.8p2, 6 October 1994 -- Fixed another little bug in disassemble(), whereby it could crash the server if you try to disassemble an unprogrammed verb. -- Added a hard minimum of 100 ticks and 1 second for all tasks, to prevent disaster when somewiz somewhere blows the values of $server_options.fg_ticks and company. -- Fixed a bug whereby the new name-lookup code would fail to recognize even dotted-decimal numeric addresses once the intermediary was presumed dead. -- Fixed a null-pointer free in the disassembler that was crashing servers that were using the GNU malloc() implementation. -- Added Craig Horman's patch to the regex.c code for the Alpha, which avoids occasional random failures of the matching code. I don't know why it took me so long to apply this... Version 1.7.8p3, 19 October 1994 -- Fixed a typo in the "hard minimum" code added in 1.7.8p2; the minimum seconds allocation is now really 1, not 2... -- Fixed a bug in the disassembler wherein it could produce garbage or even possibly crash the server if the C compiler decided to evaluate certain sets of function arguments in other than left-to-right order. -- Fixed a bug in is_clear_property() that could attempt to access an uninitialized variable. (Thanks to Quowong P Liu for pointing out this and others of the bugs fixed in the various 1.7.8 patch releases.) -- Fixed a bug whereby the server was, on some systems, always susceptible to being shut down prematurely by the `hangup' signal. Now, the `restart' script uses `nohup' and the server continues to ignore SIGHUP if it was started ignoring it (i.e., by `nohup'). -- Fixed MOO-level tracebacks to include lines for built-in functions that call verbs, such as eval(), move(), etc. -- Added MOO-level traceback printing for all server panics that happen during the execution of a MOO task. Version 1.7.8p4, 2 November 1994 -- Fixed minor memory leak in disassemble(). -- Fixed bug whereby verb_info() and friends allowed an argument of "" to always match the first verb on an object. -- Fixed bug whereby MOO keywords (e.g., `if', `E_PERM', `return', etc.) were case-sensitive. -- Fixed (I hope...) the 32-bit assumptions that break on the (64-bit) Alpha. Version 1.7.9alpha1, 25 September 1995 -- Fixed a few minor configuration and portability problems. -- Fixed a potentially server-crashing bug in subrange assignments. -- Renamed the internal server function `log()', to avoid name conflicts with the logarithm function in the math library. -- Changed the name-lookup subsystem to be more optimistic about the chances of eventually recovering from an earlier failure to restart the lookup process. Also bullet-proofed it against problems with its read() calls getting interrupted by the checkpoint timer. -- Added a paragraph to `README' explaining how to boost the limit on the number of connections a server can support. -- Added two new built-in functions enabling faster case-sensitive tests: equal(X, Y) returns true iff the values X and Y are completely equal, including the case of any strings they might contain; this is just a case-sensitive version of the `==' expression. is_member(X, L) is a similarly case-sensitive version of `X in L'. -- Made the `.program' built-in command wiz-only if $server_options.protect_set_verb_code exists and is true. -- Added the built-in function `set_connection_option(CONN, OPTION, VALUE)', for controlling various optional behaviors on the connection CONN. The only allowed values for OPTION in this release are as follows: "hold-input" -- if VALUE is true, then input received on CONN will never be treated as a command; instead, it will remain in the queue until retrieved by a call to read(). "client-echo" -- (NP_TCP configurations only) sends the Telnet Protocol WON'T ECHO or WILL ECHO commands (depending on whether VALUE is true or false, respectively). For clients that support the Telnet Protocol, this should toggle whether or not the client echoes locally the characters typed by the user. Note that the server never echoes input characters under any circumstances. -- Fixed stupid bug that let people lose by setting $server_options.fg_ticks and company to negative values. -- Added an optional second argument to the `read()' built-in function. If it is provided and true, then this call to `read()' will not suspend the calling task under any circumstances. If there is input currently available, it will be returned immediately; otherwise, `read()' returns 0. (As before, if no input is available and no more is coming, `read()' raises E_INVARG as an end-of-input indicator.) -- Added flow-control to the server's input-handling: if more than a reasonable amount of unprocessed input accumulates for any connection, the server will temporarily stop trying to read from that connection at all, until the backlog drops down substantially. -- Fixed a longstanding bug in match() that could make it return garbage in certain circumstances. More bugs in match() almost certainly still exist. (Thanks to Judy Anderson for finding this.) -- Fixed a minor memory leak in the case where #0:do_command() exists and returns a list or string value. (Thanks to Ian Macintosh for finding this.) -- Fixed a possible race condition in the TCP networking code, where a timer could go off before we've installed the exception handler. (Thanks to Alex Stewart for finding this.) -- Officially deprecated the USE_GNU_MALLOC option in options.h, since it's not aging very well. -- Completely replaced the regular-expression matching implementation that underlies match() and rmatch(); it used to be the GNU `regex' package and is now the GNU `RX' package shipped with GNU `sed'. This may not eliminate all bugs in match(), but it almost certainly has moved them around a bit. ********** The old GNU regex package had a bug in its handling of certain ** NOTE ** patterns with parentheses in them, and it is reasonably likely ********** that many MOO programmers have, perhaps unconsciously, come to depend upon this buggy behavior. Unfortunately for such programmers, RX does not have this bug, so you will want to fix your regular expressions before upgrading to this release; the fixed patterns will work correctly on both releases. The old bug concerns patterns of the form `%( ... %)*', that is, a starred parenthesized sub-pattern; for example, consider the MOO expression match("foo", "%(o%)*") Using the old regex package, this returns {2, 3, {{2, 3}, {0, -1}, ...}, "foo"} which is *wrong*; the last successful match of the parenthesized sub-pattern covered just the third character, not the second and third ones. Using the new RX package, this expression returns the proper value: {2, 3, {{3, 3}, {0, -1}, ...}, "foo"} To get the effect of the old bug, you need another set of parentheses around the whole starred sub-pattern: match("foo", "%(%(o%)*%)") Under both GNU regex and RX, if M is the result of this expression, we have M[3][1] == {2, 3} You should look carefully at your uses of match() and rmatch() before upgrading to this release, fixing those places where your code depends on the old, buggy behavior. -- Added an optional third argument to the built-in function notify(); if it is provided and true, and if there isn't enough room left in the given user's output buffer to hold the given line, then notify() will return false and the line will not have been queued for output. In all other circumstances, notify() now returns true. If the new optional argument is false or not provided, then the old behavior is invoked, in which some of the already-queued output is discarded to make room for the new line. -- Made it possible to change the maximum verb-call depth from inside the DB. The MAX_VERB_DEPTH constant in options.h was replaced by DEFAULT_MAX_STACK_DEPTH, which can be overridden by $server_options.max_stack_depth. The maximum stack depth for any task is set at the time that task is created and cannot be changed thereafter. This implies that suspended tasks, even after being saved in and restored from the DB, are not affected by later changes to $server_options.max_stack_depth. -- The task scheduler is now guaranteed never to assign a task_id() of zero. -- The built-in functions notify(), connected_players(), connected_seconds(), idle_seconds(), connection_name(), and set_connection_option() now treat connections on which boot_player() has been called as if they did not exist. -- A number of the messages printed to a connection by the server under various circumstances can now be customized or eliminated from within the DB. In each case, a property on $server_options is checked at the time the message would be printed. If the property does not exist, the standard message is printed. If the property exists and its value is not a string, then no message is printed at all. Otherwise, the string is printed in place of the standard message. The following list covers all of the newly customizable messages, showing for each the name of the relevant property on $server_options, the default/standard message, and the circumstances under which the message is printed: timeout_msg "*** Timed-out waiting for login. ***" This in-bound network connection was idle and un-logged-in for at least CONNECT_TIMEOUT seconds (as defined in options.h). recycle_msg "*** Recycled ***" The logged-in user of this connection has been recycled. boot_msg "*** Disconnected ***" The function boot_player() was called on this connection. redirect_from_msg "*** Redirecting connection to new port ***" The logged-in user of this connection has just logged in on some other connection. redirect_to_msg "*** Redirecting old connection to this port ***" The user who just logged in on this connection was already logged in on some other connection. create_msg "*** Created ***" The user object that just logged in on this connection did not exist before #0:do_login_command() was called. connect_msg "*** Connected ***" The user object that just logged in on this connection existed before #0:do_login_command() was called. -- The `for VAR in [EXPR..EXPR]' looping construct can now be used with either numbers or object numbers. That is, the construct `for o in [#0..#100]' is now legal and does the obvious thing. NOTE that in the example `o' will take on each of 101 object numbers in the specified range, regardless of whether or not those object numbers are valid. -- By popular request, added the built-in function `value_bytes(VALUE)', which returns the number of bytes of memory required to store the given value. [I was also asked to provide an `object_bytes(OBJ)' function, to give the total memory required to store the given valid object, but I wanted to think longer about possible interactions with 1.8.0's new modularity wall between the DB implementation and the rest of the server.] -- At long last, there is a DB-settable limit on the number of queued tasks any single user can have at once. If $server_utils.user_task_limit exists and is a non-negative number, then that is the `task limit' for normal users; otherwise, the task limit is infinite. For wizards, the task limit is controlled similarly by $server_utils.wizard_task_limit. Whenever a `fork' statement or `suspend()' call are executed, the server checks whether or not the current verb's owner (really, the current task perms) is already at or above their task limit; if so, E_QUOTA is raised instead of either forking or suspending. Reading tasks are not affected by the task limit. -- The result of `tostr(E_QUOTA)' has been changed to the string "Resource limit exceeded". -- Applied Alex Stewart's pAS4 patch, which modifies the result of the built-in function `connection_name()' on TCP networking configurations to contain the remote port of the connection as well as the host name, in the following format: "99 from FOO.BAR.COM, port 9999" As before, the first number in the result is pretty useless to MOO programmers (it's the server's file descriptor for the connection) but can, believe it or not, occasionally be useful to the maintainer. ******** Before upgrading an existing MOO to use this version of the server, * NOTE * you should modify the verb $string_utils:connection_hostname_bsd as ******** follows: @chmod $string_utils:connection_hostname_bsd -d @program $string_utils:connection_hostname_bsd s = args[1]; return strsub($string_utils:explode(s)[3], ",", "") || ""; . This code should work compatibly with either version of the server. -- Applied the key part of Alex Stewart's pAS7 patch, which fixes a problem with the server occasionally hanging under Sun's Solaris 2.X system. [I just removed the (useless) call to `shutdown()'; the SO_LINGER setting didn't seem necessary or very useful.] -- Added a new item to options.h, UNFORKED_CHECKPOINTS (off by default), that prevents the server from forking a separate process to make checkpoints; instead, the main server process performs the checkpoints itself, halting all user interaction and MOO task execution for the duration. -- Made the server's log output during the initial database load *slightly* easier to understand. -- Removed perhaps the last hard limit in the server; you can now have input lines with more than 500 words on them. (Thanks to Bill Drury for sending the message to MOO-Cows that finally got me to fix this longstanding bug.) -- Disallowed empty verb names and those made up only of spaces. -- Added a new built-in function `toliteral(VALUE)' such that eval("return " + toliteral(VALUE) + ";") == {1, VALUE} for all MOO values. -- The `create()' built-in function will now create children of #-1. -- Added an `emergency wizard mode' to the server's start-up sequence; if you give an initial `-e' option on the command line, then after loading in the database but *before* running #0:server_started(), the server will use its standard input and output streams to allow execution of wizardly `eval' commands and re-programming of verbs. For more details, type `help' from inside the mode. -- The verbs #0:user_disconnected() and #0:user_client_disconnected() are now called for un-logged-in and outbound connections, too, just as they are for logged-in ones. ******** Before upgrading an existing MOO to use this version of the server, * NOTE * you should check your versions of these verbs to ensure that they ******** will work appropriately when passed negative (and therefore invalid) object numbers. In most cases, it is probably sufficient to add the following lines to the top of these verbs: if (args[1] < #0) return; endif This code should work compatibly with either version of the server, since these verbs weren't being called with such object numbers before. -- The server now saves, in the DB file, a list of all users with active connections at the time of the checkpoint, shutdown, or panic that made the file. Upon server start-up, if such a list is present in the DB file, a call is made to #0:user_disconnected() for each formerly active connection. These calls are made *before* the call to #0:server_started(). In this way, there are no longer any discontinuities across a server reboot; from the point of view of code in the DB, the only evidence of a reboot is that, first, a relatively long time has passed since the last task execution, second, all connections to the server (in-bound or out, logged-in or not) have simultaneously been closed, and third, #0:server_started() has been called by the server. -- Fixed RX to do reverse-searching properly, so now rmatch() works again. Also added a abort-check in the searching inner loop for the MOO interpreter having run out of seconds. Version 1.7.9, 18 October 1995 -- Fixed stupid bug in registration of toliteral(); it now shouln't raise E_TYPE on all calls... -- Added new built-in function `queue_info([USER])'. If USER is omitted, returns a list of object numbers naming all users that currently have task queues inside the server; if USER is provided, returns the number of tasks currently queued for that user. It is guaranteed that queue_info(X) will return zero for any X not in the result of queue_info(). In essence, queue_info(X) is a very efficient version of set_task_perms(X); return length(queued_tasks()); In particular, it (a) doesn't have to allocate a large list structure, and (b) tells you when passed no arguments the complete set of users for whom there might actually be any queued tasks. -- Fixed initialization bug in code to track number of queued tasks. -- Liberalized the rules for when a call to read() without arguments will succeed. Wizards won't get E_PERM if the current task is the one that was last spawned by a command from the connection in question. You can assure this in a number of ways: 1) Never suspend, but rather only call read(). This is the one way that used to work. 2) Get lucky, go ahead and call suspend(), and have it just happen that no commands have been read since the last time you called read(), perhaps because the user was waiting for a prompt and no typing ahead. This technique is *not* recommended. 3) Before suspending, call set_connection_option(player, "hold-input", 1) thereby ensuring that no commands will be taken from this player's queue until you call set_connection_option(player, "hold-input", 0) This new third technique, suggested by Alex Stewart, was the impetus for making this change. -- Fixed a memory leak in the server's use of the new matcher. -- Fixed a bug whereby the third element of a successful result of match() or rmatch() contained 29 elements instead of the usual 9. -- Changed the method for determining a given user's queued-task limit, which is checked on every `fork' or `suspend()' from code running with that user's permissions. If the current task perms are valid, and that object has a `queued_task_limit' property, and the value of that property is a non-negative number, then that number is the limit. Otherwise, if $server_options.queued_task_limit exists and its value is a non-negative number, then that's the limit. Otherwise, there is no limit. (Thanks to Gustavo Glusman for his suggestion on how to do this.) -- The following release note, given for the 1.7.9alpha1 release, contained an error; the following version corrects it. ********** The old GNU regex package had a bug in its handling of certain ** NOTE ** patterns with parentheses in them, and it is reasonably likely ********** that many MOO programmers have, perhaps unconsciously, come to depend upon this buggy behavior. Unfortunately for such programmers, RX does not have this bug, so you will want to fix your regular expressions before upgrading to this release; the fixed patterns will work correctly on both releases. The old bug concerns patterns of the form `%( ... %)*', that is, a starred or plussed parenthesized sub-pattern; for example, consider the MOO expression match("foo", "%(o%)+") Using the old regex package, this returns {2, 3, {{2, 3}, {0, -1}, ...}, "foo"} which is *wrong*; the last successful match of the parenthesized sub-pattern covered just the third character, not the second and third ones. Using the new RX package, this expression returns the proper value: {2, 3, {{3, 3}, {0, -1}, ...}, "foo"} To get the effect of the old bug, you need another set of parentheses around the whole starred sub-pattern: match("foo", "%(%(o%)+%)") Under both GNU regex and RX, if M is the result of this expression, we have M[3][1] == {2, 3} You should look carefully at your uses of match() and rmatch() before upgrading to this release, fixing those places where your code depends on the old, buggy behavior. -- ********** There is a severe performance bug in the new matcher, causing it ** NOTE ** to run exponentially slowly in certain cases. Fortunately, these ********** cases are usually easy to work around. If you get an `out of seconds' traceback inside a call to `match()' or `rmatch()' (which will be accompanied by an error message in the log giving the pattern in use), you should probably check first for an instance of this problem. The problem concerns starred or plussed sub-patterns inside a starred or plussed parenthesized pattern. For example, here is a perfectly reasonable pattern matching MOO strings: "\"%([^\"\\]+%|\\.%)*\"" (It matches double quotes around a sequence of either (a) a cluster of characters that don't require escaping, or (b) a single escaped character.) Note, however, that it contains a plussed sub-pattern inside of starred parentheses, precisely the bad case for the new matcher. Fortunately, this pattern can be altered slightly, removing the `+', without changing the meaning: "\"%([^\"\\]%|\\.%)*\"" This pattern does not cause problems for the new matcher. I don't know of any problematical patterns in LambdaCore, but there is at least one in JHCore, in $code_utils:safe_eval, where it uses this pattern: "^%([^\"()=]+%|\"%([^\\\"]*%|\\.%)*\"%)*$" ^ ^ ^ There are three instances of the problem here, indicated by the up-arrows; the first two of them can be removed without changing the meaning: "^%([^\"()=]%|\"%([^\\\"]%|\\.%)*\"%)*$" The third instance, in my testing, does not appear to cause any problems in practice. I am continuing my search for a better regexp implementation, but this one has worked well enough for us in practice that I didn't feel it was worth holding up the release for it. Version 1.7.9p1, 18 October 1995 -- Fixed another place in the MOO's interface to the regexp code where it was assuming 29 registers instead of 9. I think that's all of them, now... :-( Version 1.7.9p2, 28 October 1995 -- Fixed a bug that could crash the server if #0:do_command suspended or aborted. -- Fixed a minor storage leak in the emergency wizard mode. -- Switched over to yet another implementation of regular expressions, this one extracted from the Python-1.3 release. It's much simpler than RX, and has seen a lot more use, so it's much more likely to be stable. It does share some of the same inherent limitations as the old regex package used in the 1.7.8p4 release of the server (which it resembles pretty closely in many ways), but I think I've tweaked the parameters in such a way as to render those limitations invisible for the vast majority of MOO uses. This matcher also doesn't have the exponential performance bug that existed in the one used in 1.7.9p1. -- The match() and rmatch() functions now raise E_QUOTA if they run out of memory during matching. Previously, they simply returned {}, just as if the pattern had not matched.