> TH> Report: > TH> I just tried the dos/win version of the os/2 port under Win95, and > TH> it works ok (the only problem I had is probably related to virtual > TH> memory and/or RSX: sorting 15000 entries gave me "Kernel32 has > TH> performed an illegal instruction" after a lot of disk activity. > TH> Then again, this is more or less expected behaviour for Win95). > > Really? 15000 entries are not so much. > > In your experience, is it a general problem with Win95 and memory > extenders that happens in other circumstances as well? Yes. Especially RSX is very vulnerable, but it also happens with other programs (not only extenders, also "real" win32 applications like the cygnus gcc. strangely enough, all are commandline/console apps). The Win95 box crashes about once every two hours if I try to do 'real' work. The only thing that never crashes is tex, and that's probably because tex starts with malloc(8Mb) or something similar instead of fetching memory as needed. Final point I can think of: memory from malloc() is _not_ zeroed by Win95. Greetings, Taco