C/C++ Users Group Library 1996 July

home *** CD-ROM | disk | FTP | other *** search

/ C/C++ Users Group Library 1996 July / C-C++ Users Group Library July 1996.iso / vol_200 / 240_01 / backgmmn.doc < prev next >

Wrap

Text File | 1987-07-29 | 13.5 KB | 317 lines

/* ** backgmmn.doc - history file Program : PC-Gammon IV with User-Defined ** COLOR ** Computer : 100% IBM PC compatibles Language : Microsoft C 4.00 : Author : David C. Oshel : 1219 Harding Avenue : Ames, Iowa 50010 : August 10, 1986 HISTORY ------- v7.0 6/13/87 - Ported from MSC 4.0 to Turbo C 1.0. Generally an easy port, but Turbo C has a number of undocumented macros, such as min() and max() in stdlib.h, that were briefly a puzzle. Also, MS C "signal" replaced by Turbo C "ctrlbrk", and various other small edits such as "alloc.h" instead of "malloc.h", "farfree" for "_ffree", etc. Turbo C introduced a number of functions or macros which collided with names I had already defined -- esp. sleep(), peek(), min(), max(). Compile the CIAO.C module using UNSIGNED CHARS, as there is a difference in implementation there somewhere. This may have introduced a bug into Typeafile, which no longer wordwraps properly. Borland seems to have found ways to prefer idiosyncrasy even in ANSI standard C. The program runs noticeably faster -- sounds are higher pitched, play is excessively rapid, etc. -- under Turbo C. v6.0 9/08/86 - Now linking to CIAO.LIB, with fast video RAM output v5.3 9/01/86 - Minor change, just added typeafile for the user's manual. Typeafile is a new addition to the IO.LIB. v5.2 8/21/86 - Got rid of clairol.obj, in favor of an improved io.lib. Includes ambiguity test in clairol(), adds a mode argument to the vid_init() routine, adds a vid_exit() routine. The library is smaller, and allows a smaller obj file where only sound or keyin is needed. The clairol() function now does a wputs("^0"); on exit to ensure the new normal mode is set. PC-Gammon now allows a command line parameter. Sets video mode 2 instead of video mode 3, if present. Usage: backgmmn b&w Actually, any argument will do, but document "b&w" for users. v5.1 8/18/86 - minor add: "^0" to mycount(), yrcount() format strings minor edits in backgmmn.h to remove obsolete function decls Note that wprintf("^%d", cnt); is DANGEROUS, depending on how the cnt argument gets formatted. The correct usage in this case is wprintf("^0%d", cnt);, with the zero! This ensure that the ^ escape finds a legitimate command char in the anticipated position. Note that the current main() and command() fns in bg0.c were split apart in version 5.0: vid_init() was in the wrong place to correctly save the startup screen for cga! The public domain version is ok, has the change. Note also that setting video mode 3 absolutely will clear the cga startup screen and send cursor home, so there's no point saving it, but we do it anyway. It's a good screen to come home to. TO DO: ruminate on improvements in playing strategy. Torve seems unexpectedly strong, lately. Have I overlooked some strength not noted there before? Need to pin that last bug in the Arrange command, if I can figure out what it is! Is Torve reacting to full moon? Are small but noticeable changes in strategy the result of using a $600 compiler for a change? Have I finally gone mad? Version 5.0 August 17, 1986, the LAST public domain version of PC-Gammon! a) Reorganized the code into four sensible modules, with a common header file, for use with MAKE. The source files are backgmmn.h, bg0.c, bg1.c, bg2.c, bg3.c. Supporting modules are clairol.obj, and its .h file. b) Revised code to use the new wprintf() function. (<stdarg.h> in Microsoft C 4.0 is magic!) Simplified the savescreen(), restorescreen() calls to do automatic heap allocation/free. c) Added the CLAIROL module to change screen attributes. (Combined module from CURSES.C, SOUND.C., plus clairol() routine.) d) In other words, PC-Gammon now has full color support for the cga and color monitor. WITHOUT sacrificing special support for the monochrome adapter. Version 4.4 August 12, 1986 a) Revised BGTOKENS.C to default to my preferred stones. b) Added H(ues stub to main command line, if not monochrome. Version 4.3 August 10, 1986 a) Minor changes to CURSES.OBJ file. Variables for screen attributes can support color selection in a H(ues command. Appended the old GETMS file to CURSES. Some changes in argument order for windowing commands (in the direction of easier to remember and more consistent). Version 4.2 August 3, 1986 a) V.4.1 FAILED its test on the cga card! Modified CURSES.OBJ to correctly set 80x25 text 16/8 mode on PCs with cga cards. Moved call to title() into setup(), so the change is noted before anything is output to screen. Tested results on NCR PC Model 3 with cga, and on Epson Equity I. Passed with flying colors (for B&W, that is)! Version 4.1 August 2, 1986 a) Minor revision to CURSES.OBJ allows wink() to handle '\b'. Discovered that Microsoft C 3.00 interprets '\008' as '\0', not '\b', so changed the backgammon source code accordingly. Version 4.0 August 2, 1986 (PC clone version Gammon IV.40) a) Added sound and ultra-fast screen io. Makes it a game for 100% clones only, now. Totally ignores ANSI.SYS, if present. b) Altered keyin() and getms() to accept function parameters for the wait-for-input and the do-on-ESC routines. For separate compilation later (maybe). c) Revised references to '\n' in user input routines because keyin() does not return newline when it encounters ENTER. In effect, the "All ok?" question now accepts ENTER and does not force users to type 'Y'. d) Fixed a bug in user's dice routine. e) Known BUG: Computer does not come in from the bar on the first roll after Arrange-ing the board. Version 3.1: July 29, 1986 (MS-DOS version Gammon IV.31) a) Changed the keyin() routine to correctly handle Fn key inputs. Extended ASCII is returned with high bit set from now on. Version 3: July 27, 1986 (Original MS-DOS version Gammon IV.30) a) Numerous minor changes to the code reflect differences in C compilers and their standard libraries. In particular, Microsoft C implements some functions as macros -- locating all the side effects of this approach can be troublesome when porting. Otherwise, this is just a "quick and dirty" port; old code which implemented for the Lear-Siegler ADM-3A terminal's lack of a clear-to-end-of-line command has not been optimized (or cleaned up, depending on your point of view). b) The graphics toggle (CP/M Kaypro) is a hilite toggle (MS-DOS ANSI). c) User bailout is ESC. Control-C is trapped by MS-DOS and halts the program. d) The "kill typeahead" feature (in BG0.C) is commented out to improve play under the abominably slow conditions of IBM-clone i/o. Version 2: March 28, 1986 (CP/M) a) Fixed the bug in which the doubling cube might revert to its default value if the computer took back a move, and if the cube had been doubled one or more times during the opening roll-off but not yet offered. b) Fixed an odd bug in the Arrange command which occasionally caused stones to switch allegiance to the opponent's color, or else to mysteriously multiply to more than 15 on a side. c) An improved error handler now hints that BAR and HOME are words used in the game; also, slows down the incorrect entry routine so that it is less mysterious than before. d) Added a Graphic toggle to main command line. Same as Control-K, as in previous versions; turns on/off the Kaypro '84 video graphics; uses the alternate token set defined in the CRT module below. e) Changed the break character back to Control-C; there is a proximity problem in keyboarding the escape key on a Kaypro keyboard. (Various CP/M versions may have used either ESC or Control-C, depending.) f) There is no longer a sanction against cheating; if you bail out during play and re-Arrange the stones, your opponent does not change. g) As before, the computer will not always take (or find) a forced move. The rule is that the player must take both dice if possible, or the larger if both are possible singly but not together. The playing algorithm attempts to find the legally required move, but will sometimes fail. It is sometimes advantageous to avoid a forced move but the computer cannot make that distinction in this version; the program allows the same lenience for the human player. The cube is doubled when doubles are thrown during the opening rolloff to decide who goes first. However, if the cube reaches 8, further pairs on the dice during rolloff have no effect. Gammon IV knows all the other OFFICIAL rules of the game; U.S. Navy and/or Saloon rules don't count. h) Improved coding in some routines, especially putstone(), which was far more baroque than necessary. There is also some occasional re-coding, where I could not prevent myself. The game strategy has not been changed, mostly because I don't fully understand it anymore, even though the documentation there is better than usual. It was written months ago in a white heat of inspiration; but I feel reluctant to submit myself to that strain again, without the prospect of remuneration. INTRODUCTION ------------ Gammon IV is a full-screen backgammon game. I wrote it for CP/M some time ago, and have finally gotten around to porting it to MS-DOS. Gammon IV has always been designed to run nicely on the lowest common denominator of available computer terminals. The CP/M version has a number of sophisticated i/o features which are nevertheless compatible with the Lear-Siegler ADM-3A (extremely) dumb terminal. This version assumes 100% IBM-compatible io, for screen and sound. Screen control is via ROM BIOS calls, sound control uses port io. DISCUSSION ABOUT THE PORT FROM CP/M TO MS-DOS (Gammon IV.30) ------------------------------------------------------------ This is a straightforward "port" from CP/M to MS-DOS. GAMMON IV was originally written for the Kaypro 10. It has now been translated for the Epson Equity I, or any IBM-PC clone with an ANSI terminal (below). The original language was Toolworks C/80, version 3.1 (CP/M), the current language is Microsoft C, version 3.00 (MS-DOS). Program modules have been altered to account for the PC-clone environment. In addition, certain in-line assembly kludges which were required in the CP/M version have now been replaced by pure C code in the MS-DOS version; i.e., there are some minor changes involving setjmp() and longjmp(), which now control program logic in a more portable fashion. DISCUSSION OF THE PROGRAM _________________________ Gammon IV is probably the best FREE backgammon game ever written.... I'm not modest! I wrote it because I was disgusted with the price and dullness of all other public domain programs which allegedly play backgammon. This program has THREE DIFFERENT PLAYING STYLES, any one of which can consistently beat a novice player, and occasionally beat a good player. In all three levels, the computer's strategy can even seem brilliant; there is nothing routine about it. This version incorporates a few minor changes and bug fixes which make it different from previous editions which have appeared on various bulletin boards and club offerings around the country. This is a public domain program. Feel free to distribute or improve it. Credit to the original author (me) will be appreciated, but is not strictly required since the copyright owners are abnormally mild-mannered (and extremely distant; see Acknowledgments, below.) Acknowledgments: The opening dialogue, in which you chose your computer opponent, is freely adapted from Alexei Panshin's long out-of-print, and now classic, "Star Well" travelogues. Scholars of computer history will recognize many allusions to this same opus in various passages of Mike Goetz's 550-point Adventure (especially in the Ice Caverns!). The Xochitl Sodality, which owns the copyright on Gammon IV, is a philanthropic society first publicized by Panshin. Persons who wish to obtain information on site licensing for Gammon IV should contact the Monist Association imaginary properties secretary at the following address: Monist Association I.P.S. c/o Xochitl Sodality Semichastny House Delbalso, Nashuite Empire */ /* eof: backgmmn.doc */