From: ts <decoux@...>
Date: 2003-08-06T21:51:18+09:00
Subject: Re: Ruby 1.8.0 embedded

>>>>> "S" == Stefan Staudt <stefan.staudt@isw.uni-stuttgart.de> writes:

S> If the ruby script is o.k. I have no problems, but if the ruby script has 
S> an error (state not equal to 0) I get problems.

 You are lucky

svg% cat test.c
#include "ruby.h"

static void
tt(char *path)
{
    int state = 0;

    ruby_init();
    ruby_init_loadpath();
    rb_load_protect(rb_str_new2(path), Qfalse, &state);
    if (state != 0) {
        VALUE err = rb_inspect(rb_gv_get("$!"));
        fprintf(stderr, "error %s\n", RSTRING(err)->ptr);
    }
    ruby_cleanup(state);
}

int main()
{
    tt("b.rb");
    tt("b.rb");
}
svg% 

svg% cat b.rb
#!/usr/bin/ruby
p Object
svg% 

svg% ./test
Object
error #<SystemStackError: stack level too deep>
Segmentation fault
svg% 

S> Is there a bug in ruby_cleanup or shoul I use other functions?

 ruby_init() and ruby_cleanup() don't do what you think.

S> Next question: is it possible to stop the ruby interpreter when 
S> interpreting a script? How can this be done?

  use 2 threads

S> Another question: is it possible to "step" through a ruby script when ruby 
S> is embedded?

 Well, you can define a trace function.


Guy Decoux