I've meanwhile discovered a problem with my GHC compile for PowerPC. While it manages to compile simple programs and the FFI seems to work (for my simple test program), many programs compiled by it abort with a message like: scavenge_stack: weird activation record found on stack: xx where xx is some number (not always the same). The problem happens for example with green-card 2.03, and in the genprimopcode program (when I try to compile the compiler using itself). However, both green-card and genprimopcode manage to print out usage information if I feed them with incorrect arguments. Can anyone give me some hints on how to debug this? Wolfgang -- GMX - Die Kommunikationsplattform im Internet. http://www.gmx.net