Home     Screenshots     Download     FAQ     Donate     Links

[Changelog & Developer's Notes]


TO-DO LIST:


- ** Forking of a GUI thread currently suspends the parent.

- Scripts are receiving list of items improperly in SF.

- Some scripts remain on heap after being killed.

- Win32 `wrap' feature (systemu?).

- Support for compressed streams.

- SF `fetchloot' behavior is incongruent with Wizard's.

- Next matching room feature for goto.lic.





2008-06-18:
(v3.57)

- Ruby v1.8.7 has been released, and Lich's embedded Ruby interpreter is now built from this version.





2008-06-13:
(v3.57)

- Implemented the in-script methods ``hide_me'' and ``hide_script'' -- the former hides the calling script, the latter hides any scripts whose names you pass it as arguments.  "Hidden" scripts will be excluded from the standard ``;list'' command, but will still show up if the (also new) ``;listall'' or ``;la'' commands are used.  The user commands ``;hide'' and ``;unhide'' is the user-inputted equivalent to the in-script methods.  See ``;help hide'' for details.




2008-05-18:
(v3.57)

- Added the Guardians of Sunfist sigils to the spell-list.txt file so that they'll be recognized by infomonitor.lic.

- Added recognition of the circle for above Sunfist sigils.





2008-04-02:
(v3.57)

- Added additional checks to ensure proper recognition of the current character's name at login.






2008-02-16:
(v3.57)

- An intermittent issue with StormFront was causing the user's character name to be improperly detected at login; fixed.





2008-02-04:
(v3.56)

- Added some additional sanity checks to the lichxml.c code.





2008-02-04:
(v3.55)

- Added the Map Browser and Console GUI modules.  See the `;help' information for further info and usage details; note that these currently are only functional in UNIX environments, and that you must have libFOX-1.6 (the 1.6 series of the core Fox Toolkit library) installed to make use of the feature.  See http://www.fox-toolkit.org for details about the Fox Toolkit.





2008-02-03:
(v3.54)

- The embedded Ruby interpreter is now built from Ruby v1.8.6.





2008-01-14:
(v3.54)

- When interpreting Wizard/SF scripts, the `%c' and `%s' character-sequences were case sensitive.  This behavior differs from SF/Wizard, and has been changed to be case insensitive.





2007-11-15:
(v3.54)

- Eliminated some unnecessary error messaging when the "debug" flag isn't set.




2007-10-27:
(v3.53)

- Wrote the SFDialog class, which is basically a primitive GUI implementation for use with StormFront.  It operates in a similar fashion to the majority of GUI frameworks: create a master window object, create labels and progress bars, attach any labels or progress bars to the window, and tweak their respective values as desired (positioning, display value, and so on).  The class will handle the translation of all of this information into the proper format for SF to understand.  If you're interested in usage details, see the file "sfdialog.rb" from the source.




2007-10-22:
(v3.53)

- A bug was preventing any room description that included a hyperlink from being properly processed when playing with SF; fixed.





2007-10-16:
(v3.52)

- Previously when Lich was used with StormFront, scripts would receive any game data the server sent to the client (with the XML elements stripped out); scripts will now only receive data the server sends if it shows up in the main window.  This behavior mimics Wizard's functioning, and results in scripts only seeing the data they "should" see.  Note that any script with its "status" flag toggled on will continue to receive all server-sent data without modification.




2007-10-11:
(v3.52)

- Fixed a bug that was causing Wizard & StormFront scripts with leading spaces on any match lines to be improperly executed (e.g., the line "  match (LABEL) (STRING)" would have resulted in a "match" check that wasn't properly recognized).




2007-10-09:
(v3.52)

- Fixed a bug that was preventing Lich from properly detecting the position of your character's body (e.g. whether the character was kneeling, sitting, standing, etc.).





2007-10-08:
(v3.52)

- Fixed a bug that was causing the list of items in a room to be tracked as part of the room description when playing with SF.

- Added proper recognition of room transitions (e.g. moving from room A to room B) when playing with SF, so that the global "$room_count" variable is accurate.




2007-10-04:
(v3.52)

- Fixed a bug in the new SF code that was preventing recognition of any prepared spell that had more than one word (e.g. it handled "Presence" properly, but "Elemental Defense I" wasn't recognized at all).





2007-10-02:
(v3.51)

- Completely rewrote all the basic information tracking for SF and finished implementing the new (proper) XML parsing routines.






2007-09-27:
(v3.51)

- Implemented a number of routines that recognize and store the character information sent to StormFront (via XML).







2007-08-07:
(v3.51)

- Wrote the LichXML class (lichxml.c), which serves as a simplistic XML parser.





2007-06-22:
(v3.50)

- An oversight in the StormFront code was preventing Lich from properly detecting the presence of NPCs after initially entering a room.  Fixed.  Thanks goes to Divid for helping me test.





2007-04-30:
(v3.50)

- Fixed an intermittent bug in the auto-updating of the current character's mind state (waiting-to-be-absorbed exp).






2007-04-18:
(v3.49)

- A bug in the cachedarray.c lib was causing several oddities, primarily influencing initial autodetection during login.  Fixed.

- Updated the script repository and lichnet scripts to store the dotted-quad IP address of the server after it's fetched once, and to clear the stored value if a connection attempt ever fails.

- Improved the reliability of automatic critter detection when playing with SF.





2007-04-06:
(v3.48)

- Fixed a bug preventing the current character's name from being automatically determined.




2007-03-07:
(v3.48)

- Fixed a bug making checkright/checkleft return inaccurate results in StormFront.

- The `cast' method will no longer consume game data silently.




2007-02-18:
(v3.48)

- Fixed a bug preventing certain game lines from being accessible to Lich scripts in StormFront; this was being caused by Lich's XML parsing.

- The `checkprone' method was checking if the player's character is currently kneeling, not if the character is currently fully prone.  Fixed.

- When playing with SF, Lich wasn't properly recognizing when the current character is kneeling/prone/sitting.  Fixed (and made somewhat more efficient).







2007-1-30:
(v3.47)

- Organized some of the data files into a devoted "data" subdir of the program's "home" directory; other miscellaneous cleaning-up of files.

- Implemented a Gem library (gemlib.rb); the `Gem' class is now usable and contains a list with various information about various gems.  Refer to `Gem.list.inspect' for a peek at the details.

- Fixed a bug in the CachedArray (cachedarray.c) lib that was resulting in various empty cache files being created and not being deleted until the next run of the program, when they were detected as stale temp caches and deleted as such.

- Added the "split_as_list" method to strings, which will return an array of elements built from a comma-delimited English-sentence list of items.  The all-too-often appearing sentences such as "You also see an item, some other item, a rock, a treasure beyond your wildest dreams and a shiny bracelet." can be split up into an array (here, it would result in `["an item", "some other item", "a rock", "a treasure beyond your wildest dreams", "a shiny bracelet"] ' as a return value).  You can also use this in a more function-call style with `parse_list("the sentence of comma-delimited items to split up")' instead of ` "an English-listing of items".split_as_list ' if desired.






2007-1-05:
(v3.47)

- Fixed a bug introduced in v3.45 preventing a Wizard script that's being executed with Lich from being able to make use of "LICH { ... } LICH" sections.




2006-12-30:
(v3.47)

- Altered the `matchwait' command to accept regular expression objects in addition to strings.





2006-12-17:
(v3.46)

- Several additions to the `;stats' information.






2006-12-15:
(v3.46)

- 40% speed increase to the `cachedarray.c' lib.





2006-12-02:
(v3.46)

- Added support for decompressing and executing script files that have been compressed (zipped) with Zlib (for example, ".gz" and ".Z" files).

- Fixed a bug in the Wizard script parsing routines that caused the program to crash if an object of a non-file type was passed to the C functions as an argument.





12-02-06:
(v3.46)


- Fixed a bug preventing the time left until you decay from persisting if you logged out while dead.

- Fixed a bug preventing any spell with a duration of 1 minute or less to be tracked as having a 0 duration; spells are tracked properly under such circumstances now.






11-26-06:
(v3.46)


- Added the in-script method `Spells.known', which returns an array of all spell objects known to the current character; note that only Lich-recognized spells (as determined by "spell-list.txt") are candidates for being returned in the resulting array.

- Added the "Opt" class.







11-25-06:
(v3.45)


- Script files weren't being closed until the corresponding script object had been cleaned out of memory; fixed.






11-22-06:
(v3.45)


- Fixed a thread-identification bug preventing scripts from being properly recognized while executing `before_dying' code blocks; this resulted in the vast majority of Lich commands being unusable from before_dying procs.

- Added support for a new list of "rooms to pause in before walking through," for goto.lic (for the Griffin's Keen area, Icemule Trace rooms which cause slipping/falling, and so on).  See "unique_map_movements.txt" to make use of it.

- Squeezed 20-45% more speed out of some of the core libraries with a few tricks: script thread recognition is now implemented with a hash function, changed the libs to cache their most-frequently-used symbols/method-identifiers to avoid constant interning, etc..

- Aliases weren't properly truncating to a single line when echoing back to the user; fixed.







11-17-06:
(v3.45)


- `reget' and `regetall' were only working if you didn't provide a string to search for; fixed (again).

- Added the in-script command `idle?' which returns true if the user hasn't typed anything for 60+ seconds, or false if they have.  You can also provide the amount of seconds it should consider as meaning the user is "idle" (e.g., `idle? 300').

- `selectput' was only returning the matching portion of a game line that caused it to return, not the entire line; fixed.






11-16-06:
(v3.44)


- Added a new method of logging in to the game; it's certainly more reliable, but is only compatible with Simutronics games.  See the DIRECTIONS.txt file for more info.







11-14-06:
(v3.44)


- StormFront seems to be forgiving of label jumps, allowing a script to "goto label:" successfully; Lich will now mirror that behavior when interpreting an SF script.

- Added a pause of 0.01 seconds to scripts that jump to the "labelerror:" label to avoid maxing CPU usage if a script gets caught in an endless loop.







11-13-06:
(v3.44)


- Corrected a few minor inconsistencies in the Wizard/SF script interpretation routines.





11-12-06:
(v3.44)


- If a script cannot be found, Lich will now attempt to start it as a Wizard/SF script.  The end result is that Wizard/SF scripts can now be started without using the `;wiz' command; note though that a matching .lic file will supercede a .cmd file.





11-08-06:
(v3.44)


- The `reget' and `regetall' commands were broken; fixed.

- Ditched the buffer.c extension I wrote ages ago in favor of the new `cachedarray.c' lib; long story short, it's a CPU-intensive portion of the program and the new lib is significantly faster than the old one.







11-06-06:
(v3.44)


- The global variable names of the server socket, client socket, server & client buffers, and last server & client strings have all been changed to "$_CLIENT_", "$_SERVER_", "$_CLIENTSTRING_", "$_SERVERBUFFER_", etc..  To ensure backward compatibility with any scripts that depend on the old names being usable, they remain valid symbols for the corresponding global objects.  What I originally found intuitive 14-16 months ago is anything but; these more self-explanatory names were rather overdue.







11-05-06:
(v3.44)


- Minor change to the display of information when the user is playing a Simutronics game and is using the Wizard; purely aesthetic tweak that helps clean up the text that Lich sends to the game window.







11-01-06:
(v3.44)


- Optimized some of the Lich script file-reading and memory storage routines; scripts now load faster and require slightly less RAM while executing.

- Optimized the Wizard script file-reading routines and increased the efficiency of the actual interpreting of Wizard scripts by several hundred percent (changed the method of recognition to a hash lookup rather than a sequential string comparison).  Loading and execution of a Wizard/SF script are both significantly faster.

- You can now make the first line of a script "# QUIET" or "# HUSH" to suppress the "script active/ended" messages Lich normally echoes to the game window (as opposed to the only way of doing this being to type the word `quiet' after the script name when starting it).

- The Wizard `echo' command will now behave as the Lich `respond' command, as opposed to the Lich `echo' command (which encloses the string in brackets and prepends the script name to it before display).

- Changed the character signifying a script-sent command to the standard Wizard '>' instead of the previously used '^'; you can now also set the character to your liking by assigning to the `$SEND_CHARACTER' variable or typing in the game window ";sendchar #{character}".








10-30-06:
(v3.44)


- A bug was preventing any non-string objects from being converted to a string with their to_s method when the `respond' command was being used; fixed.

- A bug introduced in v3.42 (?) was making the program fail to properly query the Windows registry for local paths to files/applications, etc., resulting in only the "last resort" search information being used.  Fixed.






10-28-06:
(v3.44)


- A bug was causing all alias-lines to be unconditionally downcased, which made certain tasks impossible.  Fixed.

- I've added support for writing and running a script in any programming/scripting language your local system can execute; the catch is that I've yet to figure out how to make it work in Windows, so for the time being this capability is Linux-only.  See the `;help wrap' info for more details.





10-24-06:
(v3.44)


- A bug in the new pathfinding algorithm was causing it to find shorter routes than the prior algorithms, but still not the shortest -- fixed.





10-23-06:
(v3.43)


- Scripts were (yet again) not being given time to execute their "before_dying" code blocks when the user exited the game; fixed.

- Added a new algorithm to the pathfinding extension; the new method is less efficient than the pre-existing one, but is also guaranteed to return the shortest possible path from the origin to the destination (as opposed to the "estimated shortest," which is the prior behavior).  In practice, the reduction in moves required is something like 10% -- it cut 50 or so rooms off the EN<->WL trip (now 640 or so total), and usually yields a path that's 2-10 rooms shorter when running around locally (local meaning trips with 10-100 moves total).

- Changed the behavior of the "goto" script: it now by default uses the new guaranteed-shortest-path algorithm.

- A bug in the LichNet client script was causing any received chat message with a `%' character to raise a cryptic error and sometimes outright crash the script; fixed.

- Lich scripts can now have file extensions of ".lic", ".rb", or ".rbw" -- all three will be properly recognized (note that the script repository, however, will still only accept a ".lic" extension).

- An enormous amount of enhancements to the help information and several other miscellaneous user-interface and/or documentation tweaks (whether it LOOKS enormous from the user's side or not, believe me, it took me several hours to beef up the help info to this point).

- Removed the AUTOGC command; its usefulness ended around v2.0 and its only real function since has been to confuse and confound users.

- When used with StormFront, Lich was mistaking any game line with brackets around it as being a room title that signified the player had moved; e.g., if you were running a movement script and something used a Combat Maneuver on you as you walked by, Lich would mistake it for meaning you'd finished moving to the next room, which would throw off all subsequent MOVE commands and quickly result in the script exiting in error.  Fixed.

- Added a new in-script command, `selectput'.  It is by no means intuitive (unless you happen to be familiar with the behavior of the C function `select()', et al), but it is extremely flexible.  The timeout (which is measured in seconds) and block (see directly below for details) are optional, and the basic behavior, in order, is (1) send the command, (2) watch for any of the given "succeeded" and "failed" messages, (3) return the matching game line if a "succeeded" message is seen, or call the block and pass it the matching "failed" message before starting over again at step 1 if an "action failed" message was seen.  If a timeout value is given and an "action succeeded" message has not been seen before the timeout (in seconds) expires, the command gives up and returns `nil'.  The end result is that if the action succeeds within the time limit, the return value of the command will always evaluate as logically true; if it is logically false (i.e. returns `nil') then the timeout count expired, since those are the only two conditions that by default result in the command returning.  Syntax and usage (note that as is the custom, {curly-brackets} signify a required argument and [brackets] signify an optional one):

	selectput({command to send}, {array of `succeeded' messages}, {array of `failed' messages}, [timeout]) { block to call if failed msg is seen }   # (block is optional)

	selectput("send to the game", ["a success message", "another success msg"], ["a failure message", "another, etc."], 15) { |failure_line|

		echo "This block will be called everytime an `action failed' line is seen"

		echo "Right after this block is called, the game command will be re-sent and the script will go back to waiting"

	}

	goto "mylabel"

	goto :mylabel

	walk { checknpcs "priestess" }

	debug { echo "Hi there, user!  You currently have the `verbose debug output' on for scripts!" }

	debug("Hi there, user!  You currently have the `verbose debug output' on for scripts!")

	fetchloot('cloak')

	fetchloot('backpack')

	buffer = Buffer.new

	(push 50 items onto the buffer, which is identical to an array)

	buffer.push "test"	# Our 51st item

	(buffer is now an array with only 1 object in it, the string "test")

	old = buffer.history		# Constructs a new array (from the first 50 objects, not including the 51st which is in RAM at the moment) and returns it

	wait_until("Waiting until you have enough mana...") { checkmana 14 }

	cast 414

	multifput 'sign of striking', 'sign of smiting', 'sign of swords'

	echo "Hi! You just killed me, so I'll exit now."

	echo variable1		# 'variable1' existed when the 'before_dying' command was used, so this will echo 'I exist' to the user

	echo variable2		# 'variable2' didn't exist until after the 'before_dying' command was used, so the script will cause an error at exit

	(do Wizard stuff)

	(this will be executed as though this block were an actual Lich script)

	(back to Wizard stuff)

	(do lots of Lich stuff)

	person = matchfind("? just arrived", "? just went ")

	name, direction = matchfind("? just went ?")

	container, items = matchfind("In the ? you see ?")

	item_array = items.split(',')

	put "take my #{item_array[0].split.last} from my #{container}"

	loop {

		put "take my #{item_array[0].split.last} from my #{container}"

		item_array.shift

		break if item_array.empty?

	}