From: ts Date: 2002-01-15T18:48:41+09:00 Subject: Re: possible problem with PLruby (Re: Possible bug with struct.c) >>>>> "n" == nobu nokada writes: n> I may misunderstand, but is this only with C extension? It's n> exactly impossible to break $SAFE without C, however, the n> conflict can occur with Ruby script? Well, this is at least the case for all my extensions because when I call rb_class_define() I *always* expect to have a new class and I don't want to retrieve an old class, possibly created by another extension. All my extensions, actually, first check if the constant is defined * if it's not defined it continue * else it stop with an error message (rb_eNameError) >> If someone use plruby with ruby >= 1.6.6, it's best to load the new >> version (0.2.5) because it can exist a security problem. Sorry. n> 0.2.5 may fail with autoload? Can you give me an example ? Actually autoload begin with rb_secure(4); and because I've $SAFE >= 4 this must not be a problem. Guy Decoux