This almost falls into Issue #3, but it's a bit worse than just having a bunch of x86 asm everywhere. There is a lot of pointer-y stuff being thrown about, and on 64-bit this gets ugly fast due to everything being written with only 32-bit in mind.
It's hard to determine if the errors will be genuinely problematic or if it's just bad precision at the compilation stage. Just fix them as you see them, but get worried if the errors hover around areas that directly deal with MadTracker-related data. That will probably mean redesigning the file format to work on any architecture.
This almost falls into Issue #3, but it's a bit worse than just having a bunch of x86 asm everywhere. There is a lot of pointer-y stuff being thrown about, and on 64-bit this gets ugly fast due to everything being written with only 32-bit in mind.
It's hard to determine if the errors will be genuinely problematic or if it's just bad precision at the compilation stage. Just fix them as you see them, but get worried if the errors hover around areas that directly deal with MadTracker-related data. That will probably mean redesigning the file format to work on any architecture.