Re: MiNT 1.11 problems

[Dirk Stadler <dirk@lstm.uni-erlangen.de>]

----- Begin Included Message -----

If changing the compiler used to compile MiNT fixes the problem, it
isn't likely that the library has anything to do with it.  Unless of
course the libs are also recompiled with the same compiler.  No one
has mentioned anything about having done that or not, though.
Currently, the kernel only uses 3 or 4 library functions...things like
memcpy, bcopy, and so on...and these functions are all quite stable.

Has anyone that has been having problems with GCC-2.5.8 and MultiAES
tried compiling a kernel with optimization disabled?

I'd love to be of some help in tracking this down, but I don't have
MultiAES.  My MiNT 1.11 kernel seems to be running fine (compiled with
gcc-2.5.8 -DDEBUG_INFO -DMULTITOS -DONLY_030 with full optimization)
on my TT with the normal AES.  But then again, I haven't used any GEM
software in so long that I've almost forgotten what the desktop looks
like. 

Cheers,
entropy

----- End Included Message -----

Ok, let's wind up the problem again !

Any version (from 1.08 through 1.11) of MiNT compiled with GCC 2.58 does not work 
together with the MULTIAES 4.1. It does not matter if optimization is disabled or 
not. The libs are compiled with GCC 2.58.

Switching back to GCC 2.3.3 (Libs also compiled with GCC 2.3.3) improves the problems
a lot, but did not solve them at all ! Programs like 1STGUIDE, ST-GUIDE and BOXKITE
still have problems mainly with the window gadgets.

The developer version of MiNT 1.08 (august 93), which comes with MULTITOS 4.1, works
without having these problems mentioned before. While the same MiNT 1.08 self-compiled 
with GCC 2.3.3 does not work correctly.

I think this is neither a problem of the libs nor of MiNT itself ! 

Cheers,
Dirk