OK, so long as the users are 100% trust worthy (not just not malicious, but
skilled enough not the write buggy code that crashes everything). I'd still
suggest using a safe interpreter, and use some of the features to allow "safe"
versions of the potientally "dangerious" procedures. Your users will get a
fully functional interpreter, but with safeguards so they can't "crash"
things, even by mistake. Read the man page carefully.