What are your .conf settings?

Mmm, I do not think the problem is as I have configured the dosbox, since with the same settings in the latest version (0.74) the game runs without problems (besides that I did several tests with various configurations and nothing, always the same error)

I've changed the title a bit to reflect the regression status a bit.
Thanks for the report.

Ok, thanks Qbix.

You could fix this problem?

Here I leave the game if anyone is interested in fixing this problem or do some testing (is a shareware/playable demo)

I haven't been able to reproduce the problem. Tried default settings with latest SVN, also tried machine=ega and machine=vgaonly, none show the problem described.

What platform did the problem appear on? Maybe it is an endian issue... Also, where did the SVN build come from? Is it clean SVN? Is it self-compiled or obtained from somewhere?

I tried with all possible settings and nothing, still the problem (but as I said before, with dosbox 0.74, with any configuration, the game works perfectly)
The SVN was downloaded from emucr.com, i also tried with the latest build of ykhwong.x-y.net, and with that version of dosbox the game works without any problem.

I do not know then what is the problem...

The emucr builds and the recent builds of 0.74 with integrated munt show the problem. I don't know if there are any changes in the munt builds that would have an impact on graphics, but it suggests that the problem is a compile or build issue. I could be mistaken, but I think the emucr and munt builds are compiled with VC. My builds of SVN source with GCC do not show the problem.

EDIT: MB6 shows the problem, but ykhwong's builds do not. This looks more and more like a VC vs. GCC build issue.

Mmm, then, please, you could give me your version of dosbox, so I test it (with this game) and tell you if that's the problem?

Thanks.

Ripsaw can reproduce it with vc builds thus that part of your report is done 😉

Strange how such vc vs gcc bugs come up from time to time. We used to run in those with exult since our snapshots are always done with vc but most of our rleases with vc 😉

Well... Then, if that is the problem, where I can get the latest *good* svn, but not of emucr?, there is another site similar to that?

compile yourself using GCC or wait for the issue to be identified and fixed.

Or use 0.74 release for this game or does it have any other problem fixed by SVN?

As I do not know how to compile I stay for now with version 0.74 of dosbox (at least to play this game)

well it could be a bug on our side. (array too small, uninited variable)
Did somebody already check if it happens with 0.74 compiled in VC as well ?
I could provide a SVN build compiled with the build system that compiled 0.74.

The munt builds are based off 0.74 source and appear to be compiled with VC, so that's one example that points to it being related to which compiler is used.

MB6 has the issue, so I guess hal will be in a good position to track it down.

I've tried a number of other chained mode EGA side-scrollers from Id and Apogee, and they don't have the issue. Only the Jason Storm trilogy, both older and newer versions, have the issue. So, perhaps a combination of memory handling (fastmem) and the particular scrolling method used.

Qbix, I'm interested with you svn compiled in the same way that the version 0.74, if you can, could you upload it? (So, I use one only dosbox for all)

And thanks to all for the cooperation trying to see what is the problem.

Ripsaw already verified that the bug doesn't happen with gcc.

Yes, that I know, is for that why I'm asking the last svn compiled with GCC (and no, I do not know how to compile), if someone is so kind, could upload it to me?
If not, no matter, I will continue as usual (using various versions of dosbox)

0.74 works fine for the very most stuff anyways.

It occurs here with msvc 2008 as well, debug builds exhibit the problem only with /O2 which makes things a bit tricky.