The Arcade Companion 1

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

/ The Arcade Companion 1 / The Arcade Companion 1.iso / mahjong / fixes.doc < prev next >

Wrap

Text File | 1991-10-19 | 17.1 KB | 386 lines

SER NUM PROBLEM & FIX ---------------------------------------------------------------------- 1029 - 9/30/89 SYMPTOM: A couple people have reported that The Draftsman wouldn't recognize their VGA card, although MJVGA ran perfectly. PROBLEM: A real good question! The source code for the VGA test is identical in both programs! SOLUTION: I re-compiled The Draftsman using the same memory model as MJVGA. I also added an option which lets you attempt to run The Draftsman even if it doesn't see a VGA card. The Draftsman has always worked fine on my system & on the systems that I test it on. Therefore, I don't know if this fix will work or not. If you should experience this problem, please let me know. 1055 - The "official" release of Version 2.1 SYMPTOM Someone with a VGA card reported that neither MJVGA nor The Draftsman would recognize his card. PROBLEM I dunno.... Strange VGA card maybe??? SOLUTION He tried the above-modified Draftsman and said it worked fine after selecting "try anyway." MJVGA now includes this option. SYMPTOM If you select What's Left, then exit the What's Left screen by clicking the LEFT mouse button, the screen re-draws and either A) you return to the What's Left screen or B) you select some other menu item or tile. PROBLEM You're SUPPOSED to leave the What's Left screen by clicking the right mouse button, but in reality any button will get you out. Using the left button, however, was putting a "left button click" message on the stack. When the MJVGA game board returned, it saw this message & acted on it -- possibly taking you right back to What's Left. SOLUTION This was a simple one -- I just cleard the button stack after exiting the What's Left screen. SYMPTOM The computer locks up after the game board is displayed. PROBLEM This has only been reported once. The fellow corrected the problem himself by running MJVGA with the "D" command line parameter to disable the monochrome display routines. It seems that his computer didn't let MJVGA write to the "dead space" where a monochrome card should have been. It usually doesn't matter -- if there's no monochrome card, the data just "falls into the bit bucket" and causes no harm. But he's proven that in certain systems it DOES matter! SOLUTION Since the vast majority of people don't have dual monitor systems, I've changed the default from using the second monitor to NOT using the second monitor. Now, if you wish to use two monitors, YOU MUST START MJVGA WITH THE "D" COMMAND LINE PARAMETER!!!! This may inconvenience a few people, but will increase default compatibility to everyone. 1105 - 2/10/90 (Problem not fixed -- serial # indicates revision of this document.) SYMPTOM Someone reported that the abbreviated HELP command didn't work properly. The regular HELP menu appeared instead of just the number of moves left. PROBLEM MJVGA first checks the left mouse button to determine the command. It then expects to see BOTH the left AND right buttons pressed if you want the shortened HELP screen. If you release the left button before the computer can register that both are pressed, you'll get the standard HELP menu. SOLUTION I haven't fixed this one yet. It will be fixed in the next release, though. Until then, if you experience this, just hold down both buttons for a few extra moments to be sure that MJVGA sees them. You shouldn't experience this unless you either have a slow machine or are a fast clicker. I had to load down my 25 MHz machine with four background processes before it ran slow enough to let me see this problem. 1150 - 3/6/90 SYMPTOM A couple people reported that they couldn't change "brush colors" in The Draftsman. Clicking on the color squares did nothing. PROBLEM This problem seems unique to Logitech's new Series 9 serial mouse. To date, all reports involved that mouse. Moreover, the Series 9 Bus mouse works perfectly! Only the serial mouse seems to cause this problem. I traced the exact cause to a timing problem. Briefly, I was asking the mouse which buttons were pressed before it knew. Consequently, it was telling me that NO buttons were pressed! My guess is that the bus mouse's hardware is faster than the serial mouse's for obvious reasons. The bus mouse was therefore able to register the buttons faster than the serial mouse. SOLUTION The solution was easy once I found the cause. I just execute a several-millisecond time delay before asking the mouse which buttons are pressed. This seems to give the rodent ample time to organize it's thoughts, even on my fast machine. 1234 - 6/1/90 The "official" release of MJVGA version 2.2 SYMPTOM See the write-up for S/N 1105. PROBLEM See the write-up for S/N 1105. SOLUTION I've changed the mouse-check routine so that you'll get the abbreviated help whether or not the left button is still down. I have a suspicion, though that this problem may have been caused (at least in part) by the Logitech Series 9 serial mouse problem described above. If you should still experience this, PLEASE be sure to tell me whether you have that particular rodent or not. 1355 - 7/21/90 * NOTICE * This release is the first to be compiled with Borland's new Turbo C++ compiler. The compiler is supposed to be compatible with all software written with Turbo C 2.0 which is what I've been using. But my efforts to even get MJVGA22 to compile and link under C++ have led me to believe otherwise! I did get it to work, though, so the differences aren't insurmountable. I've already noticed one performance difference in MJVGA compiled on the two compilers. The sound from the C++ compiled file seems "choppy" compared to the same thing (almost!) compiled under C 2.0. I may be wrong, but I think C++ does some form of "multitasking" to get & dispatch messages to the objects. This "stealing" of CPU time would account for the choppiness. If anyone has a better explanation, (or a way to overcome it), please let me know. I've finally added the long-promised fix to let the "L" tile set loader access more than 24 tile sets. Actually, I HAD to because I now have more than 24 of them! It's still not done in graphics mode (although I AM planning to), but at least it's better than it was. I've also added a built-in registration form printer. Pressing CTRL R at any time during the game will print a registration form to printer port LPT1. See manual for complete details. SYMPTOM Only the first thirteen tile sets can be accessed via the "L" command with a Microsoft mouse, although a Logitech mouse can access the maximum of 24. PROBLEM When the screen goes into text mode, the Logitech mouse remains in the 640x480 mode while the Microsoft mouse switches to 640x200. (I don't know what others do.) To approximate the screen line, I was >>4'ing the Y coordinate from the mouse. Unfortunately, since the Microsoft mouse was only returning a maximum value of 199, the largest line number possible was 13! SOLUTION This was simple enough! I just changed the shift from four places to three. Problem solved! 1404 - 9/1/90 SYMPTOM If you remove a tile pair, select a tile which was just freed, back up a move, then select a match to the tile just selected, they'll both go away -- even the one that's no longer free!! PROBLEM A selected tile is verified only before highlighting it, not before actually removing it. So, the tile was verified as being OK and later removed without further verification. SOLUTION I actually had code in place to solve this problem, but just forgot to call it! Whenever a menu item is selectd, I SHOULD de-select any tile that's been selected. I forgot to include the call when BACK-UP is selected. All I had to do was insert that call and the problem was solved. Many thanks to Erik Brooks for discovering this strange problem! SYMPTOM The 'T' key is used to bring up the list of tile sets available. PROBLEM This in itself isn't a problem -- but the manual says you should use the 'L' key to load tile sets!!! OOOPS! I GOOFED!!! SOLUTION I could have changed the manual, but I decided to change the program to accept 'L' instead of 'T'. Why? Because the 'L' is used on the command line to load a tile set and I thought I should use the same letter in both places to try to standardize things a bit. 1487 - 12/1/90 The official release of MJVGA version 3.0 (a month earlier than I expected, believe it or not!) * NOTICE * There have been a lot of subtle changes to MJVGA for this version, so I probably caused more bugs than I fixed. If you find anything strange, please let me know! SYMPTOM If you somehow work the dragon from left to right & have only the two left-center tiles remaining, the inner of the two is still not consideed "free". PROBLEM I never thought anyone could get to that tile from the other end, so I never bothered to declare it free if the two tiles that it's blocking are removed. But, of course, someone has proven me wrong! Actually, TWO PEOPLE have already gone through the game that way & have found the problem! SOLUTION This one wasn't too bad. I just added logic to free the afflicted tile when the two to its left are removed. 1579 - 12/25/90 SYMPTOM EDITHOF doesn't seem to work -- the name field contains garbage. PROBLEM Here's the first problem caused by all my changes. It seems I changed the HOF file, fixed EDITHOF to accept it, then changed the HOF file specifications again -- without fixing EDITHOF again! So, EDITHOF was looking in the wrong place for the name -- and all it found was junk. SOLUTION All I had to do was get EDITHOF back "in sync" with MJVGA's HOF file. It was a rather simple fix. SYMPTOM When the fancy new windows close, they don't restore the screen properly. PROBLEM I have no idea. I've never seen this problem on any test machine. Furthermore, it has only been reported by one person so far. She said that her machine uses the Trident VGA chipset, so that may have something to do with it -- but I'm not sure. If you experience this problem, please let me know EXACTLY what hardware & software (stuff loaded from your CONFIG & AUTOEXEC files) you're using. SOLUTION The problem hasn't been solved yet. It's reported here only for documentation purposes. If you happen to run into it, PLEASE LET ME KNOW because I've never seen it & can't correct it until I have some idea of what's happening here! 1637 - 1/20/91 ADDITIONS Starting with this serial number, MJVGA30 will support the Adlib sound card. See the What's New or .DOC file for complete details. I also added auto-sensing to the No Moves Left message. Again, see the manual or What's New file for complete details. Since these additions are rather trivial, I didn't think they were worth a new version number -- nor were they worth sitting on until the next release. So, I just added them to the current release. 1704 - 2/21/91 SYMPTOM When you cancel a message with the RIGHT mouse button (although you're supposed to use the LEFT), the PEEK/ROTATE/REMOVE menu may pop up. PROBLEM I never bothered clearing the right-button queue, since the left button is supposed to be used to cancel messages. SOLUTION I just clear the right-button queue along with the left-button queue whenever a message is cancelled. Now you can use either button to cancel a message. SYMPTOM The QUIT verify box sometimes comes up & then quickly disappears PROBLEM Only one fellow reported this problem -- and I think his mouse button was "bouncing" (producing a double click) for some reason. SOLUTION Just to be safe, I inserted a slight time delay and a de-queueing call after the QUIT select and before the verify box. This should cure the problem -- if indeed it IS a problem. I couldn't duplicate it, so I won't say whether it's fixed or not. 1775 - 3/12/91 SYMPTOM If you work the dragon in from the left to the two rightmost tiles excluding the the two "offset" ones, the tiles can't be selected -- you're told that the tiles aren't gree. PROBLEM To see if these tiles were free, the code just looked at the two offset tiles. If they were gone, these tiles were considered to be "free", otherwise they were considered to be blocked. I never bothered to check if they could be moved to the left. SOLUTION I wish they were all this easy. I just added a check to the left. One interesting note -- awhile back (see fixes for #1487) someone reported that they couldn't remove the inner "offset" tile to the left if the two "standard" tiles were removed. That problem was valid and was fixed -- but they never should have gotten that far! This problem should have surfaced first. They should NOT have been able to remove the two rightmost "standard" tiles to the left to free up that inner "offset" tile. I have no idea how they found that problem without running into this one! 1853 - 5/1/91 SYMPTOM When you remove the last two tiles, you see the No Moves Left message before going to the congratulations screen. PROBLEM This isn't exactly a bug, because when there aren't any more tiles left, you really DON'T have any more moves. However, it IS rather ridiculous to display at this point. SOLUTION They don't get much easier. I just don't check for moves left if there are no tiles left. 2021 - 8/3/91 SYMPOM The registration form print command wasn't appearing in the "K" menu selector. PROBLEM I forgot to add 1 to the number of items in the menu last time I added something. SOLUTION Add 1 to the number of items in the menu. I think this may also solve a hang-on-exit problem. Although I wasn't providing space for that one extra menu item, I was writing stuff to it. Needless to say, this could have been overwriting something important. 2198 - 11/1/91 The official release of MJVGA 3.1 * NOTICE * This release has been compiled with the new Borland C++ 2.0 compiler. This shouldn't affect anything, but one never knows. There's not much new stuff in this release. I just figured that, between the new compiler & all the stuff that was added to v3.0 without a version number change, one's about due. ADDITIONS A pause feature has been added. See the manual or What's New file for complete details. Note: The listed serial number is the first which contains a fix. All releases with later (ie larger) serial numbers will also contain the fix unless otherwise stated.