freebsd-dev/games/larn/Fixed.Bugs
Jordan K. Hubbard 554eb505f8 Bring in the 4.4 Lite games directory, modulo man page changes and segregation
of the x11 based games.  I'm not going to tag the originals with bsd_44_lite
and do this in two stages since it's just not worth it for this collection,
and I've got directory renames to deal with that way.  Bleah.
Submitted by:	jkh
1994-09-04 04:03:31 +00:00

217 lines
12 KiB
Plaintext

This is a list of the fixes/enhancements made to larn V11.0 in Version 12.0.
(Version numbers consist of 2 parts: ver.subver. When the save file format
changes, ver must be bumped. This is why the next release of Larn is 12.0
and not 11.1. This is used in the savefile routines to check for out-of-date
save files). This list was mainly meant to be a record of what changed,
for my own sanity. It's included for your benefit (Warning: SPOILER!):
0. lprintf() in fileio.c (now called io.c) has been changed to use varargs
so that its variable number of arguments usage is now portable. Pyramids
primarily had this problem.
1. Panic handler was added to signal.c. This routine catches fatal errors
like segmentation faults, bus errors, illegal instructions, etc., and
trys to performs a savegame() before dumping core. This helps prevent
the loss of a good game due to a game malfunction. Also, the name of the
signal received is printed, instead of just its number.
2. The version number of the program is now selectable from the Makefile.
see the symbols VER and SUBVER.
3. When at an altar, pray and donate 3000000000 gp. and ye used to receive
a whopping amount of gold due to a wraparound problem with the signed
ints. This has been fixed by using unsigned longs when asking for money
amounts.
4. It was possible that when compiled with work hours checking, checkpointing
enabled, and having "play-day-play" in the .larnopts file a segmentation
fault would occur at its first attempt to do a checkpoint. This was due
to an improperly declared savefilename array in tok.c. This has been fixed.
5. on level H, casting a missile weapon (mle cld ssp bal lit) off the edge of
the level would mess up the display, as it didn't know when to stop. This
is needless to say, fixed. Absolute bounds are now in effect for missile
type spells, see godirect() in monster.c.
6. The create monster routine will now create monsters in random positions
around the player. Before, the 1st one would always be created to the
upper left.
7. If you vpr or lit at a throne, it would summon a gnome king that you
would have to deal with. However, as each throne has only one king with it,
successive vpr's should not create more gnome kings. Presently, successive
vpr's will create more kings. This has been fixed.
8. The mechanism to manage spheres of annihilation has been reworked to provide
a cleaner design and to eliminate some possible problems.
9. The spell gen (genocide monsters) has been implemented.
10. When dropping a ring of strength and having been weakened to STR=3 the
player might end up with a negative strength. Strength is now stored
in 2 variables, real strength, and strength bonuses. Only real strength
can now be weakened down to a minimum of 3, so unless you have a ring of
strength -3 or less, strengths below 3 should not occur.
11. larn -h will now print out a list of all available command line options.
12. larn -o<optsfile> now lets you specify a .larnopts file on the command
line. This was necessary as part of the solution to number 14 below.
13. The "savefile:" statement has been aded to the .larnopts format to allow
specifying the savefilename (full path) for the savegame operation.
This too was needed as part of # 14 below.
14. A player id facility has been added to larn. The complaint was that
the game used the userid to order the scoreboard, thus only one scoreboard
entry was allowed for each userid. If the compile time symbol UIDSCORE
is defined at compilation time (see Makefile), this will still be true.
However, if this define is omitted, the game will create and manage a
file called ".playerids" where names are taken from the specified
.larnopts file (now a command line option) and assigned a unique playerid.
playerid's will now be used to govern scoreboard entry posting. This
feature makes it easy for one person to have many characters, each
appearing on the scoreboard. Be kind to your fellow players!
The philosophy of one score per player gives more players the opportunity
to bask in glory for all to see!
15. It is no longer required that the player be WIZID to create the scoreboard
or to examine the logfile. Anyone with the correct wizard's password can
now use these command line options (password is only needed to create/clear
the scoreboard). If you want to prevent players from zeroing the
scoreboard, change the wizard's password. (in config.c) By the way, wizards
may be alot of fun, but they are prevented from being placed on any
scoreboard. (for clarification)
16. Monsters now have intelligence, that is some of them. This determines if
the monster moves using the previously stupid movement method, or by using
the new IMM (intelligent monster movement) algorithm. With IMM, monsters
will move around corners, avoid pits, traps, etc. With increasing levels
of difficulty, more monsters will be using IMM. Beware of IMM when
aggravated! Those little beasties can really find you!
17. Added the scroll of life protection.
18. Larn now consults the file ".holiday" to check for holidays if the TIMECHECK
option (no playing during working hours) is enabled. Before, larn knew
nothing about holidays. It should now let people play if it is a holiday.
The format for a .holiday entry is: "mmm dd yyyy comments . . .".
19. In nap() and napms() it is possible that with nap(0) or napms(0) there
would be an infinite loop and the game would hang. The case of nap(0)
is now looked for.
20. The granularity of gold piles has been increased. iarg[] has been changed
from char's to short's, so instead of 255 x 10^n granularity we now have
32767 x 10^n granularity. This also means more than 255000 gp can be
dropped in one place. Not realistic, but it prevents a worthless
annoyance. Who said games were supposed to be realistic?
21. Termcap capability has been added to larn. If the symbol VT100 is defined
in the makefile, the game will be compiled to use only VT100 compatible
terminals (Much more efficient). If the symbol VT100 is omitted, the game
will be compiled to use the termcap entry for whatever terminal you are
using. This involves an extra layer of output interpretation, as every
byte sent to the terminal must be inspected for control tokens.
Only 3 termcap entries need be found for the game to be functional:
CM (cursor movement), CE (clear to end of line), and CL (clear screen).
For a better display, the following are optional: AL (insert line), DL
(delete line), SO (Standout begin), SE (Standout end), and CD (clear to end
of screen). The .larn.help file was left as is, with VT100 escape
sequences in it. If the termcap version of larn reads it, it is translated
for the desired terminal type. The .mail60* files have been removed, and
their text is now included in bill.c so it can be used with any terminal.
Note: If compiled for termcap, and using a VT100, the display will act
a little different. This is because the VT100 does not have insert line/
delete line codes, and the scrolling region must be simulated with vertical
wraparound instead of scrolling. Thanks goes to Michiel Huisjes for the
original termcap patch.
22. When playing as wizard, if you go down stairs on 10 or V3, or up stairs
on H, 1, or V1, etc. you would be placed in a phantom zone where the display
was really weird ([-1] subscripting), and would eventually lead to a
segmentation fault. Stairs and volcano shafts now check for the level
they are being used on.
23. In response to some sites having only unsigned chars (flame the
manufacturer), the chars that were used to store positive and negative
numbers have been changed to shorts. This includes diroffx[], diroffy[],
iarg[][][], ivenarg[], and some others. I believe the changes are correct,
but I have none of these machines to try it out on. (Volunteers?)
24. The function fullhit(n) in monster.c was supposed to return the damage
done by n full hits on a monster. It only returned the damage for ONE hit,
thus severely limiting the usefulness of the web and sle spells.
25. Someone said that they were getting segmentation faults when they were
reading scrolls as the wizard. I couldn't find the problem, which may
have had something to do with the signed char problem mentioned above.
However, I've added a check in read_scroll() and quaff_potion() to trap
any scroll or potion types that are not in the game.
26. "vt125" has been added to the acceptable terminal list
(checked only if compiled with -DVT100).
27. In savegame() and restoregame(), there was a 6 hardwired into the i/o
statements which assumed the size of struct cel was 6. On some machines
this caused the rightmost part of each level to not be saved in a savefile.
These 6's have been replaced with sizeof(struct cel), and should now be
portable.
28. The option "no-beep" has been added to the .larnopts file. When specified,
beeping is inhibited at the terminal.
29. When becoming wizard, no longer to you wear the ring of protection, and
null scrolls and potions are no longer created.
30. Many spelling errors have been fixed, both in player messages, and in the
code itself. A thanks goes to Mars Gralia who sent me a detailed list of
the mistakes.
31. When a player wins a game, if getlogin() fails, a segmentation fault will
result, because the NULL returned from getlogin() is used as a pointer.
This call has been replaced (now using loginname already determined).
Also, the mail creation upon winning has been rewritten, mainly to allow
termcapping of the text.
32. The Larn Revenue Service will now always appear on level H. Before, it
was only created if the player had outstanding taxes. In that multiple
save files per player are now more possible, this was seen as incorrect.
33. Input buffer flushing is now in effect. If the input char queue exceeds
5 bytes, the excess is discarded. Also, if the player hits or gets hit
all input bytes are flushed (within 1). This relieves the situation
where many moves have been typed ahead of the display and the player keeps
getting hit while the queue of moves is processed.
34. When a savefile has been altered, a warning message is displayed to the
effect that you've cheated, and you will not be placed on the normal
scoreboard. If you then save the game, and start 'er up again, memory
of the cheating was lost. This has been fixed, by letting the scoreboard
routines consult the cheating flag. Also, the I node number of the
savefile is written into the savefile, so cp'ing, etc., will avail the
cheater not. If high security is needed, the game should be run suid.
This suid mode has not been made the default because most installations
do not want to install it that way.
35. The sources have been run through lint, and most of lint's complaints have
been taken care of. An attempt was made to adjust the code for 16 bit int
machines. Many casts to long have been put in. I don't know if it will
run on a 16 bitter, but it should be closer to that end.
36. When larn starts up, if it can't find the scoreboard, it will now make a
blank one instead of complaining that there is no scoreboard. It is not
necessary to do "larn -c" to initially create the scoreboard.
37. When listing out the logfile (larn -l), the error message "error reading
from input file" has been fixed. Also, the date & time of a player's
demise is now included in the logfile.
38. When casting web or sle into a mirror, the game will no longer bash the
player. Instead, the player will either fall asleep or get stuck in his
web.
39. Items like cookies, books, chests, swords of slashing, and Bessmann's
flailing hammer can now be sold at the trading post.