⟦e651920b0⟧

TextFile

                ==[ All the bugs that fit, we print ]==
Version 1.1 Old Empire (Empire Classic!) - circa 1978, decompiled from 
                        optimized 11/70 object 1982-3 (honest!).

Bugs should be reported to ihnp4!druxn!tsf  -  Tom Fisher
Credit for the game belongs to Peter Langston.  I take credit for
any bugs remaining in this version.  If you are using a version
that was modified (in addition to the parametric goodies in empglb.c)
since distribution, try to locate the bug and fix it before reporting it.

The only things considered bugs are those that appear to change the
original intent of the code.  Core dumps, meaningless or inaccurate
messages, and obvious cheating opportunities are considered UN-
intentional, and therefore, bugs.  Enhancements, (such as adding a
coastwatch command) are not listed here.

Items containing "Fix:" are fixed in this release.  The others will be 
fixed later.


1) Production sometimes goes negative.  Results in negative civilians,
    gold bars, etc.  Very elusive.  Suspect max127() or update().

2) Signals  SIGHUP, SIGINT, SIGQUIT not handled at all.  This stuff was
    more or less ignored during the decompilation process.  Sorry.

3) Country report heading "time" should be BTU ?

4) When country is joined and first two-sector allocation is diagonal,
    the welcome message still claims you have 0,0 and 0,1.

5) Mobility goes negative in sector where crash landing occurs.

6) Nav'ing ships at 2,2 causes your ships at 34,2 to move also ?
   This is in a 64,64 world.  This is caused by [xy]wrap() routines
   mishandling the range check.  Needs new algorithm.

        Fix: [xy]wrap() are OK. The range check is in nxtshp.c .  Fix
             the range check so that ships half way around the world aren't
             considered to be within range.

7) World of size other than 64,64 never tested after decompilation.
   At least one problem is that in a 128x128 world, empcre sets the
   gold to 127 for almost every sector.
   There are probably hardcoded 2**6 or 64 references somewhere.  Sorry.

8) Telegrams get screwed up.  It appears to be placing data from initial-
   data section in the telegram buffer.  Suspect this happens when more
   than 511 characters are in the message or when there is concurrent
   activity in the telegram file.

9) You get a core dump if you enter numbers that are greater than +127
   or -128 when doing a spy command.  E.g.  spy -156:-14,-25:-23 dumps
   memory.  This is because sargs doesn't check for reasonable values
   on return from atoip().  Suspect that a lot of other commands will
   also behave this way.

10) The radar command doesn't indicate the source of the radar when it
    is a land-based radar.  It should print out something similar to the
    ship-based radar:
        radar at 3, 3
        Efficiency 100%  ...etc....

        Fix: In rada.c insert a printf right before the call to radmap 
             for land-based radars.  The printf prints out the line:
                radar at x, y

11) The opening messages are unbuffered - too slow.

        Fix: Remove setbuf from and insert fflush(stdout) in empire.c
             Add fflush(stdout) calls to prmptrd.c, cleared.c, empdis.c,
             and chec.c .
             And also remove setbuf from empcre.c and EMPFIX/main.c .

12) Firing torpedos from a sub used to set the mobility to -10 on the
    first one, then decrement the mobility by 10 for each additional one.
    When the limit of -120 was reached, it would stay there.  This appears
    to have broken, because now the mobility stays at 0.

        Fix: A previous 'fix' broke this one.  The check to limit the
             mobility to -120 to keep it from going positive, used a
             macro that used characters as arguments.  Now that the
             mobility is a short, something has to change.  Replace one
             statement in torp.c to decrement the mobility by ten until
             -120 is reached.

13) You can tend from ships other than a tender.

        Fix: In tend.c add a check to make sure that the "tender" is
             really a tender.  If it isn't, print
                "You can't tend from a xxxx!"
             where xxxx is the ship type of the "tender".  Also, get
             rid of all the goto statements in this file, since it is
             a simple routine and it has to be tested anyway.  Also
             correct the spelling in the "Seas to rough ..." message.

14) When a shell fired from a ship hits another ship, the percent
    damage prinout in  the telegram gets messed up.  It prints something
    like ...did 27 223460amage to ...

        Fix: This got broken when all the pr(fmt()) and prfmt() calls
             got changed to printf() calls in order to make this more
             portable.
             There was a printf(buf) call in rea.c (the read command).
             If the telegram contained a %, then printf considered that
             to be a formatting directive.  Change printf(buf) to
             printf("%s", buf).
             Also - in prmtprd.c, chkok.c, and map.c there were similar
             constructs.   Add the %s in each of these cases too.

15) I can't believe it!  You can use someone else's submarine to torpedo
    another ship!  What a terrific cheating opportunity !

        Fix: Add a check in torp.c, similar to the ones in other places
             where getshno() is used, to make sure that the sub is owned
             by the attacker.

16) The ok_ret.c file is no longer needed in EMPSUBS.  This apparently
    caused a problem at HP because the loader looked for the superet
    function (referenced inside of ok_ret.c) which has been removed.

        Fix: The ok_ret.c file should have been removed when the removal
             of superet() was done.  So remove it.

17) One recipient of Version 1.0 would like all the ^G characters
    changed to \007 so that his printer prints out the listings without
    going crazy.  Also, he would prefer all the printfs to fit within
    80 columns so that the listings can be printed portrait mode.

18) loo reports only 50% efficiency even if the sectors are yours.

        Fix: In look.c change the printf argument for printing out the
             the sector efficiency.  Change it from an unconditional call
             to round() to only call round() when the sector is owned by
             someone else.

19) You have a sub in sector x,y.  You accidentally torpedo a ship in
    that sector that was already sunk.  If there are sunken destroyers
    in that sector with shells still aboard, your sub will get hit with
    depth charges from the Deity.  Also, the news will report that you
    torpedoed a Deity ship, and that the Deity fired on you in self defense.

        Fix: In torp.c, add a check that insures that a victim ship
             is NOT owned by the Deity.  Deity-owned ships numbers are
             treated the same way as a non-existent ship number.

20) You can use civilians more than once - (the "rolling civilians" bug).
    If you place 1 military in a sector, then three days later force
    a delivery of civilians into that sector, you will get three days
    worth of production out of the civilians.  This is great for getting
    as many civilians as you want in the beginning of the game.

21) If you try to specify something that isn't a commodity,
        e.g.   cen -5:5,0:6 ?xxxx<25
    it prints a message saying that 'xxxx' isn't a commodity, then proceeds
    with the command. It should return a syntax error.

        Fix: In nxtsct.c (in the snxtsct() function), the return from the
             encode function wasn't being checked.  Add a check so that if
             the encode function returns -1, snxtsct will return -1.

22) Your country accrues BTUs even if 0,0 isn't a capital, and even
    if some other country owns the sector in the 0,0 position.

        Fix: In empdis.c, make sure that BTUs accrue and tech and research
             levels decay only if 0,0 is a capital AND your country owns 
             the sector.
             While we're at it, get rid of the goto statements in init_na(),
             use nat.nat_stat instead of the uninitialized nstat, and
             when a sect.sct_lstup == 0, add a putsect(0,0) after setting
             sect.sct_lstup = curup (the putsect was missing).
             The goto removal was verified with identical objects before
             and after the removal.

23) The "shell" command only accepts one command (line).  It returns to
    empire after first command (line) is executed.

        Fix: In empglb.c, change shllrg1 from "-t" to "-i".

24) In empdis.c, dispatch() is called with one argument, but the
    definition of that function does not include any arguments.

        Fix: In empdis.c, remove the argument from the call.

25) In empdis.c, there is an lseek call with a second argument that is
    not a long.

        Fix: In empdis.c, make the 0 (the second argument) a 0L.

26) The weather command produces a map with only the top and right border.

        Fix: In weat.c, add the left and bottom border.  Also fix the 
             weather manual page.  Also, while we're at it, remove the
             goto's from weat.c.  Goto removal verified with identical
             objects.

27) Trying to designate a sector that's not yours costs you a BTU, and
    there is no error message.

        No Change:  It works this way so that you can do:
                     des #1 ?des=- g
             which will usually contain many unowned sectors or sectors that 
             can't be designated (., ^), without getting an error message for 
             each undesignatable sector.  I guess it's assumed that a 
             no-BTU-cost census is done before trying to designate something.

28) Redirection and piping don't work.

        Fix: Oops! This got broken in the conversion from pr(), prfmt(),
             and pr(fmt()) to printf().
             In empdis.c, when redirection (>, and >>) or piping is
             indicated, switch file descriptors (1 and proto) so that
             proto becomes the new file descriptor 1 and the old fd1 is
             saved in the global variable, savfd1.  Flush stdout right before
             switching back to the original fd1 (savfd1).  In empdef.h, add 
             the extern declaration for savfd1, and in empglb.c, add the
             definition of savfd1.
             Also in empdis.c, change write(1,...) to printf call.
             This now works, but it is a little different from the original
             design.  The output goes to either your terminal OR the output
             file instead of to both as in the original design.  If you
             want it to go to both places, use:
                map #1 | tee emp.out
             (assuming you have a tee command).
             Also, when redirecting output into a file, the file is
             created with mode 622.  Also... if empire runs with the
             setuid bit on, e.g. setuid to games, the uid, games, has to have
             permission to access the complete path to the output file.
             There can't be any intervening directories with mode 700,
             for example.

29) The "shell" command sometimes doesn't work.  It prints "Sorry, Empire
    problems" after exiting the shell and then exits empire.  Sometimes
    it prints:
        : invalid identifier
    and returns to empire without allowing any shell commands to be
    entered.

        Fix: The fixes are all in empdis.c .
             The "Sorry ..." problem is because storing stuff into call:argv
             overwrites call:cnumbuf.  This causes the main to be called
             with a crazy country number.  In call(), declare the automatic
             variable argv[4] (instead of argv[2]), set argv[0] back to
             emprog[thisprog] (just for grins), and call main with argc=4
             instead of argc=2.
             The "invalid identifier" problem is a lot trickier.  This
             happens when empire is called with a null second argument:
                empire "cname" ""
             Executing the shell command after entering empire thus produces
             the error message from the shell (Bourne or Korn- ha, ha).
             This is because the environ pointer is getting changed.  It
             ends up pointing to somewhere near where it is supposed to, but
             it's pointing to a null string.  When the shell is entered, it
             tries to read the environment and chokes on the null string
             thinking it's supposed to be some kind of identifier.
             I don't know how to fix this, but you can get around it by
             invoking empire as follows for countries with null rep names:
                empire "cname" "" anything
             This makes the last argument non-null, and everything seems to
             work out hunkydory.

30) In empdis.c, main(argc, argv) is called recursively after "shell",
    "fixup", etc.  There's a better way to do this.  This way, the stack
    grows on every shell command until a switch is made to a new module.

31) In empglb.c, the value of a tech center is zero.  The value of a 
    research lab is 10, airfield 10 etc.

        Fix: In empglb.c (in the dchr structure), set the tech center 
             value to the same as the research lab.

32) 4.2 BSD doesn't have a fcntl.h (apparently).

        Fix: Remove dependency from info.c, powe.c, rea.c, getele.c,
             wu.c, empdis.c, empcre.c, empire.c, EMPFIX: {loans.c,
             main.c, sectors.c, ships.c, treats.c}.  Put constants,
             O_RDONLY, O_RDWR, and O_WRONLY in empdef.h.

33) def # produces a core dump on some systems.

        Fix: In defe.c, before checking to see if argp[2][0] == '%',
             make sure that argp[2] points to something legal.
             
34) Checkpoints, and other things aren't cleared when confiscating a
    sector to repay a loan.

        Fix: In coll.c, after changing the ownership of the sector,
             clear checkpoint and defense pointers, and also update
             the timestamp of the sector (matches what happens when
             defeating a sector with "attack" command).

35) The makefiles aren't very good.  They don't work on 4.2 BSD.

        Fix: Read the make(1) manual page and have another try at it.
             Replace :mk and all the Makefiles with new versions. Now
             to make the whole thing, type:    make
             or:  make emp1
                  make empfix
                  make empire
                  make empsub.a
                  make empglb.o
                etc.  to make individual parts of the system.
           They still don't work very well on 4.2BSD, especially in EMPSUBS,
           COMMANDS, and EMPFIX, where libraries are involved.

36) When you do a census #0 ?des=u&civ=1270 
    not only does it fail to show you the sectors desired, but it shows
    you urban areas where civ==0.  A cen #0 ?des=w&shell=1270 works
    the same way.

        Fix: In nxtsct.c, in the encode and decode functions, a bit
             mask of 0200 is used.  This is fine if numbers aren't
             greater than 127!  Increase the mask to 04000 to allow
             numbers up to 2047.  Also, in encode, return(atoi(cp)&03777)
             to make sure that the 04000 bit is off.

37) Yet another problem with buffered output.  There are times where
    little delays are supposed to occur between bursts of output.
    During the world creation, bursts of three periods ('.') are supposed
    to indicate progress.  Likewise during a battle (attack, assault, board),
    the melee is supposed to proceed in small bursts.  When getting
    flakked, the POOF!'s are supposed to show up in little bursts, not
    all in one shot.

        Fix:  Insert a fflush(stdout) just before any sleep() call in
              empcre.c, atta.c, assa.c, boar.c, torp.c, fly.c.

38) 4.2BSD rand() function produces 32-bit results.  Code expects 16-bit
    results.

        Fix:  Make sure that everywhere rand is used, it's used something
              like this:  if( chnce < rand() % n )
              where n <= 32768.  This affects files in EMPSUBS, COMMANDS,
              EMPCRE, and EMPIRE.  Also remove argument from rand calls in
              diss.c, and empire.c.

39) In 4.2BSD, time.h is in /usr/include/sys rather than /usr/include.

        Fix: In empdis.c, insert ifdef to grab the header from the right
             place.  Add #define BSD in empdef.h (commented out for SysV).

40) 4.2BSD doesn't have a termio.h.  Sread.c won't compile.

        Fix: Insert a new sread routine for 4.2 that uses sgtty.h.  So
             there are really two sread routines in there, separated by the
             #ifdef BSD.

41) The spy info returned from an urban area doesn't get the number of
    civilians right, and reports 10 times the number of shells present.

        Fix: Actually, it didn't get warehouses right either.  In spy.c
             correct the offsets into ichr to get the packaging factor
             right.  Offsets should be; c=12, m=13, s=14, g=15, o=17, p=16.
             Bug introduced when porting to Amdahl UTS.

42) If you type in too many characters when sending a telegram, you
    can get into a situation where the error message, "Try that last
    line again..", gets typed out several times.

        Fix: In getele.c, if too many characters are typed in,
             make sure that the input is flushed up to a newline before
             reading the retry-line.

43) Using redirection creates a file owned by "games" if the empire
    command is setuid to games (e.g. map #1 >xxx    xxx is owned by games).

        Fix: In empdis.c, there should have been a resetui() before
             creating the file.  Remove the statement that tries to create
             the file assuming it doesn't exist, allowing execution to
             fall through to a resetui(), unconditional unlink, and
             re-creation.

44) If your umask disallows write permission for other (the normal case),
    redirected output files won't be writable by a setuid empire.

        Fix: In empdis.c, insert a couple umask(0) calls before the creat
             calls.  This allows setting mode 0622.

45) Trying to append to a file that doesn't exist results in error messages
    and no appending.

        Fix: In empdis.c add check (access(cp, 0)) to see if output file
             exists.  If it doesn't exist, create it.  Otherwise open it
             and seek to the end.

46) Contracting production units in a research center results in the
    sector being owned by either the deity or country #2.  Sort of
    like a surprise grant.

        Fix: In cont.c, check to see if ip->i_del is zero.  If
             so, there is no delivery item associated with the
             production units (e.g. medical discoveries or technical
             breakthroughs).  In this case, don't modify the delivery
             field in the sector record (i.e. setting it to 0 or 2).

47) On some 4.2BSD systems, having curup declared as an int and the
    sector timestamp declared as a short causes strange updating
    behavior.

        Fix: In empdef.h and empglb.c declare curup as a short.

48) If the world is created, countries added, and those countries
    log in and occupy new sectors all within the first half hour,
    the sector timestamps of several sectors could have a value of
    zero.  This results in those sectors never being updated.

        Fix: In empcre.c, on the line where up_offset is calculated for
             the deity, subtract one.  This ensures that curup in empdis.c
             will have a minimum value of 1.

49) It appears that if you contract production units from a  d, i, t, r,
    *, #, or h you don't really get paid for it.  You get the commodity
    instead.

        Fix: In update.c, in the switch statement where each type of
             sector gets updated (right before the S_GMINE case), there
             should be a check to see if the sector is contracted.  If
             so, cashout should be adjusted and the production units
             cleared. (copy the three lines preceding X6030 sans the
             goto's).
DataMuseum.dk

DKUUG/EUUG Conference tapes

⟦e651920b0⟧ TextFile

Derivation

TextFile
