Last Crusade (Amiga version): Corrupt IQ points counter!

[fingolfin]

If you provide details about the precise script bugs, we might be able to modify ScummVM to workaround those script bugs, without requiring any patching of game data files. We did that for various other script bugs in the past.

[JohnnyWalker2001]

Wow, great job guys!

Let me know the bugs you've found and I might actually add them to my guide... one day. (I'm not great at updating it, but it is pretty comprehensive.

http://indyguide.mixnmojo.com

[icanntspell]

I used your guide to map the puzzles, it's the best walkthrough available. It would be great if you find the time to update it. You already seem to have had a few suggestions for improvements.

In fact, bluntly using your guide, updating it and putting it somewhere is one of the thoughts that crosses my mind. However if you update it, that's even better. Let me know if I can help.

[icanntspell]

Sorry for raking up this old thread, but I found the discussion again when cleaning up my mail. I remember I had the intention to make a nice patch for the bugs but (insert favorite excuse here)...

I still have all the details about this so an average programmer should be able to create a patch for it. I gladly forward the details to anyone interested in making scummvm handle the scriptbugs. (Assuming this isn't fixed yet)

[ssdsa]

I still have all the details about this so an average programmer should be able to create a patch for it. I gladly forward the details to anyone interested in making scummvm handle the scriptbugs. (Assuming this isn't fixed yet)

Why don't you just post your findings right here?

[icanntspell]

Ok, here come my findings. I compared the amiga version to the PC/Steam version. The thing to look out for is

startScript(125,[30,15]);

This script 125 is the one responsible for giving the IQ points. First parameter is the puzzle number (here 30), 2nd one is the amount if IQ points you get.

[icanntspell]

I quote "adalsgaard" on this comment about this bug.

About the points error leaving Castle Brunwald: It seems to "reversed"! When you get caught, free yourself and escape, you get 25 IQ points even though you're not supposed to. However if you escape WITHOUT getting caught, you get 0 IQ points (supposed to get 25 IQ points). Weird stuff...

Once I managed to dump the scripts using the build-in dump-resource option in ScummVM it all became obvious.

First the PC version :

Code: Select all

root@gateway:/home/ftp/incoming/dumps# ../scummvm/tools/trunk/descumm -u -3 Steam/roomobj-25-363.dmp
Events:
  43 - 0A17
  21 - 0000
  66 - 7269
  65 - 6C70
  61 - 6563
[0000] (C9) faceActor(VAR_RESULT,??Local??[3840]);
[0005] (43) Var[363] = getActorX(15616);
[000A] (06) VAR_NUM_ACTOR = getActorElevation(38);
[000E] (02) startMusic(93);
[0010] (00) stopObjectCode();
[0011] (43) ??Var??[2583] = getActorX(33);
[0016] (00) stopObjectCode();
[0017] (66) ??Local??[617 + ??Local??[101]] = getClosestObjActor(24940);
[001E] (63) Var[101] = getActorFacing(15);
[0022] (6C) Var[257] = getActorWidth(161);
[0026] (00) stopObjectCode();
[0027] (A8) if (Var[92 Bit 15]) {
[002C] (40)   cutscene([]);
[002E] (0F)   if (getState(367) == 0) {
[0034] (0A)     startScript(125,[36,25]);
[003D] (**)   }

and this is the Amiga version :

Code: Select all

root@gateway:/home/ftp/incoming/dumps# ../scummvm/tools/trunk/descumm -u -3 Amiga/roomobj-25-363.dmp
Events:
   A - 001F
[001F] (0F) if (getState(364) == 1) {
[0025] (A8)   if (Var[92 Bit 15]) {
[002A] (40)     cutscene([]);
[002C] (0F)     if (getState(367) == 1) {
[0032] (0A)       startScript(125,[36,25]);
[003B] (**)     }

Apart from the leading rubbish in the PC version (haven't fot a clue on why this happens) there is a tiny difference

PC : [002E] (0F) if (getState(367) == 0) {
Amiga : [002C] (0F) if (getState(367) == 1) {

The Amiga version tests for the wrong value which explains the "reversed" behaviour. To fix this bug I edited 25.LFL and changed the "1" into a "0". Loaded my savegame, went through the fireplace and got the expected 25 points.

[icanntspell]

For the 2nd bug (not getting the points for KO of the security officer a.k.a. puzzle 29) :

Amiga :

Code: Select all

[0038] (48) if (Var[115] == 2) {
[003F] (40)   cutscene([]);
[0041] (0A)   startScript(125,[30,15]);
[004A] (1A)   VAR_EGO = 8;
[004F] (36)   walkActorToObject(8,465);
[0053] (3B)   ??Var??[3848] = getActorScale(209);
[0057] (01)   putActor(0,9,281);
[005D] (D1)   animateCostume(VAR_EGO,??Local??[3072]);
[0062] (80)   breakHere();
[0063] (80)   breakHere();
[0064] (5D)   setClass(8,[150]);
[006B] (1E)   walkActorTo(8,78,80);
[0071] (3B)   ??Var??[3336 + 8] = getActorScale(26);
[0077] (BC)   stopSound(Var[512]);
[007A] (00)   stopObjectCode();
[007B] (1A)   VAR_EGO = 1;
[0080] (C0)   endCutscene();
[0081] (18) } else {
[0084] (13)   ActorOps(8,[Costume(33),Palette(6,14)]);
[008C] (48)   if (Var[180] == 403) {
[0093] (13)     ActorOps(1,[Costume(30)]);
[0098] (48)   } else if (Var[180] == 406) {
[00A2] (13)     ActorOps(1,[Costume(35)]);
[00A7] (48)   } else if (Var[180] == 405) {
[00B1] (13)     ActorOps(1,[Costume(36),Palette(6,0)]);
[00B9] (18)     /* goto 00BC; */
[00BC] (**)   }
[00BC] (0A)   startScript(62,[4]);
[00C2] (1A)   Var[222] = 70;
[00C7] (1A)   Var[223] = 258;
[00CC] (0A)   startScript(35,[8,0,1,1]);
[00DB] (80)   breakHere();
[00DC] (68)   VAR_RESULT = isScriptRunning(35);
[00E0] (28)   unless (!VAR_RESULT) goto 00DB;
[00E5] (0A)   startScript(52,[]);
[00E8] (0A)   startScript(120,[8,11]);
[00F1] (1A)   Var[188] = 1;
[00F6] (0A)   startScript(125,[30,15]);
[00FF] (**) }

Steam

Code: Select all

[0038] (48) if (Var[115] == 2) {
[003F] (40)   cutscene([]);
[0041] (0A)   startScript(125,[30,15]);
[004A] (1A)   VAR_EGO = 8;
[004F] (36)   walkActorToObject(8,465);
[0053] (3B)   ??Var??[3848] = getActorScale(209);
[0057] (01)   putActor(0,9,281);
[005D] (D1)   animateCostume(VAR_EGO,??Local??[3072]);
[0062] (80)   breakHere();
[0063] (80)   breakHere();
[0064] (5D)   setClass(8,[150]);
[006B] (1E)   walkActorTo(8,78,80);
[0071] (3B)   ??Var??[3336 + 8] = getActorScale(26);
[0077] (BC)   stopSound(Var[512]);
[007A] (00)   stopObjectCode();
[007B] (1A)   VAR_EGO = 1;
[0080] (C0)   endCutscene();
[0081] (18) } else {
[0084] (13)   ActorOps(8,[Costume(33),Palette(6,14)]);
[008C] (48)   if (Var[180] == 403) {
[0093] (13)     ActorOps(1,[Costume(30)]);
[0098] (48)   } else if (Var[180] == 406) {
[00A2] (13)     ActorOps(1,[Costume(35)]);
[00A7] (48)   } else if (Var[180] == 405) {
[00B1] (13)     ActorOps(1,[Costume(36),Palette(6,0)]);
[00B9] (18)     /* goto 00BC; */
[00BC] (**)   }
[00BC] (0A)   startScript(62,[4]);
[00C2] (1A)   Var[222] = 70;
[00C7] (1A)   Var[223] = 258;
[00CC] (0A)   startScript(35,[8,0,1,1]);
[00DB] (80)   breakHere();
[00DC] (68)   VAR_RESULT = isScriptRunning(35);
[00E0] (28)   unless (!VAR_RESULT) goto 00DB;
[00E5] (0A)   startScript(52,[]);
[00E8] (0A)   startScript(120,[8,11]);
[00F1] (1A)   Var[188] = 1;
[00F6] (0A)   startScript(125,[29,10]);
[00FF] (**) }

In plain english ; in the Amiga version you get the 15 points for puzzle 30 if you give the book or KO the guy. The PC version correctly gives 10 points for puzzle 29 for KO and 15 for puzzle 30 when giving the book. I changed the file 31.LFL to give the 10 points for puzzle 29 in case of a KO as the PC version does and tested it.

[fingolfin]

For the newer 256 color (VGA) PC versions, they changed the disk format and some opcodes, so instead of -u you have to pass -n in order to get sensible output out of descumm for indy3.

All in all, the descumm options are a bit messy, and they don't cover all possibilities, so for some scripts in certain games, the current descumm is not able to 100% correctly decompile them.

[fingolfin]

Oh, and in all versions of indy3 I have, for the first bug you report, the script has
getState(364) == 1

This includes my Indy3 FM-towns, and both an english and a german VGA version of the game, as well as an english EGA version. So I guess the PC version you got from steam received fixes from LucasArts ?

It shouldn't be too hard to override both script bugs from within ScummVM, too. Very nice catches, both!

EDIT: Oops, my mistake. Of course the relevant line is the one with
if (getState(367) == 1)
resp.
if (getState(367) == 0)

I get "1" in the EGA version, and 0 in the VGA + TOWNS versions. So, it is exactly as you described, and there is no weird steam-only fix

[icanntspell]

Thanks. It would be great if it would make it into ScummVM at some time. If it ever does, please drop me message. I don't have the time to track all patches that make it into ScummVM anymore

[fingolfin]

If you (or somebody else) wants to ensure these are not forgotten, then please file a bug report on our website, with the information you included here.

[icanntspell]

OK, posted a bug report for it.

https://sourceforge.net/tracker/?func=d ... tid=418820

Thanks for your time.

[fingolfin]

I posted a patch on your bug tracker item which should cure the bugs. But I currently don't have the time to properly test it, so it's not (yet?) in our code repository.

[icanntspell]

I updated and tested your patch and it is working as expected. It would be nice if it was merged at some point.

In case somebody is interested, I uploaded my other ScummVm patches at my website. I didn't enter them in the patchtracker since they are not really into the spirit of ScummVM.