COMPILE: Getting really frustrated with crash

[MrBaggins]

I complied the stuff and had it working... kind of... yesterday. It was crashing at end of turn.

I did a fresh install of the code and game, and now it crashes at the point of starting a new game, just as the map is started to be displayed. I made a version with debug info, and the damn debugger crashes... :/

Sound works up until it crashes. The regular off the CD version works ok.

I'm running XP.

I just put the new miles.h and miles.lib in the appropriate directories, fixed a couple of warnings in debugcallstack.cpp (unused local var defs - moved inside #ifdndefs)

What OS are the people who got the miles.h and miles32.dll to work, running?

Anyone else having similar problems?

[Martin Gühmann]

I had a similar problem, I did some changes and every build after these changes crashed when I wanted to start a new game. Finally I deleted all the precompiled files and compiled the whole project again and it did work. I gues it was some sort of a linker problem.

-Martin

[Locutus]

I have the same problem as well, but can't solve it either, Martin's solution doesn't work for me. Solver has the same problem as well, we've been discussing it over ICQ but can't figure it out. Very frustrating indeed

[DDowell]

Me too! It is hopeless! And it is occuring in all builds (althoug in the debug build it at least says where the error occurs... it is those bloody assert-lines that just keep on failing).

Frustrating is the word alright.

[Martin Gühmann]

It seems to be a different problem then. OK we should now answer the OS question. My is Win 98. And maybe also the compiler question: My is VC++ 6.0 the so called authors edition, with some shareware like features. but better then nothing.

-Martin

[MrBaggins]

I have Visual Studio 6 Enterprise Edition. I installed and patched to SP5.

DirectX 9.0b (debug) and DirectMedia 6.0

Windows XP Pro. SP1. (not checked build)

[Martin Gühmann]

I should have DirectX 8.0 or 8.1 installed. I wasn't able to compile the game with DirectX 7.0a. But as far as I know Mr Ogre used DirectX 9.X.

-Martin

[MrBaggins]

Yes, but I think he did something different than that mss.h stub. He did stubs in other files instead, and removed mss.h and mss32.lib.

[MrBaggins]

Ok... I am in the process of compiling completely fresh. When I'm done, I'll see if I can try it on a system other than XP... although its a bit of a pain in the ass, and its a big problem to not have this working on XP.

XP has gotten to be a bit of a standard nowdays.

[J Bytheway]

Mine seems to be working perfectly, including the sound which I've just compiled in.

I'll try to list everything that might differ from my setup to yours:

Windows XP Home, latest SP (I think it's SP 1).

VC++ 6.0 SP5 (I think it's enterprise, but it might be professional).

DX 9.0b, Summer 2003 release

I copied the whole DXMedia folder into the libs directory like Mr Ogre, even though he said it was probably unnecessary. I think this might be crucial.

I am always using the final build (The debug build takes more than 5 minutes to start, and I can't wait that long).

I added #include <multimon.h> to display.h. I did that because it was necessary on an earlier build even though I didn't need it on the later ones.

[Locutus]

Well, I have pretty much the same setup as John, except that I don't think I have any kind of SP for VC++ (I have Enterprise) I too copied the library files (which was necessary because without that it wouldn't link them).

I didn't include mulitmon.h, but adding it doesn't seem to make a difference (I'm now trying it again on a clean build, I'll let you know if that does work -- but don't count on it).

The strange thing is, the debug build doesn't take nearly as long to start for me as for John, I don't really notice any significant differences with the other builds.

[MrBaggins]

OK. Done EXACTLY as John has said (from a fresh start), and bang.... Crash. (at the point of the map being displayed when starting a new game.)

Now, this means that there is somehow some difference in the compiled file.

To verify that we (Locutus, me, whoever else) need to try out that exe that John has.

If that works, then we need to find out what John did that we didn't.

I believe that the uninstaller won't delete modified files... so here is my idea:

*Copy the directory to another location.

*Uninstall the CTP Source, using the unwise exe.

*Delete the compiled exe's, the maps and the contents of the final and debug directories.

*Zip/Rar whats left up. That should leave just whichever source files are different from those, as produced that work, plus whichever libs work.

Locutus and myself should be able to figure out what the problem is by testing/comparing those.

Thoughts?

[J Bytheway]

Originally posted by Locutus
I too copied the library files (which was necessary because without that it wouldn't link them).

The point is that I copied all the CXmedia files, not just the library files.

I didn't include mulitmon.h, but adding it doesn't seem to make a difference (I'm now trying it again on a clean build, I'll let you know if that does work -- but don't count on it).

I also doubt this will make a difference, but such doubts have been proved wrong in the past.

The strange thing is, the debug build doesn't take nearly as long to start for me as for John, I don't really notice any significant differences with the other builds.

That is indeed strange. What processor are you using? Mine is a 1.6GHz Intel P4. I would say that the slowness is due to my shortage of RAM causing disk-swapping, but during the loading the hard disk is not being accessed, so it can only really be due to the processor, or possibly I have slow memory...

[J Bytheway]

Originally posted by MrBaggins
To verify that we (Locutus, me, whoever else) need to try out that exe that John has.

I can't email it, and I can't fit it in the Apolyton upload area anywhere. I also need to slap "This material is not made or supported by Activision" onto the startup screen somewhere if I'm going to distribute it .

If that works, then we need to find out what John did that we didn't.

I believe that the uninstaller won't delete modified files... so here is my idea:

*Copy the directory to another location.

*Uninstall the CTP Source, using the unwise exe.

*Delete the compiled exe's, the maps and the contents of the final and debug directories.

*Zip/Rar whats left up. That should leave just whichever source files are different from those, as produced that work, plus whichever libs work.

Locutus and myself should be able to figure out what the problem is by testing/comparing those.

Thoughts?

It mightbe easier to use diff - but I'm not sure exactly how diff handles multiple directories, if at all. Fortunately I have cygwin installed so I can go find the man pages...

[J Bytheway]

Originally posted by J Bytheway
I can't email it, and I can't fit it in the Apolyton upload area anywhere.

Actually I can if I zip it up. Silly me .