Vincent Rivière wrote:
Maybe it's time we build a freemint, xaaes, aranym, etc test suite for regression testing so this stuff (hopefully) doesn't happen. Just an idea, maybe it's already there.Paul Wratt wrote:.. it appears to be mintara.prg, if possibly only on certain hardware, which does not explain why 1.16 is fineYou use Alan's build. It's OK for me in text mode. As I'm unfamiliar with XaAES stuff, I never tried graphics mode.Alan's builds usually use the "enhancements" branchs. That can make some difference.And if I'm not wrong, Alan uses GCC 4.x. There is maybe a compiler issue. Some existing hidden bugs appears when changing the compiler, or when compiling at high optimization levels. The biggest source of problems can be the "strict aliasing rule", or assembly routines that forgot to backup some registers.The first step should be that someone is able to reproduce your problem. If you zip your ARAnyM setup and make it available for us, we will tell you quickly if we encounter the bug or not.
Thanks, Mark