From: Kent Sibilev <ksruby@...>
Date: 2006-07-30T01:44:30+09:00
Subject: Re: Trace API bug

Anyone?

On 7/21/06, Kent Sibilev <ksruby@gmail.com> wrote:
> Does this patch make any sense?
>
> Index: eval.c
> ===================================================================
> RCS file: /src/ruby/eval.c,v
> retrieving revision 1.616.2.182
> diff -u -r1.616.2.182 eval.c
> --- eval.c      20 Jul 2006 07:04:14 -0000      1.616.2.182
> +++ eval.c      21 Jul 2006 05:02:24 -0000
> @@ -5837,8 +5837,14 @@
>         break;
>
>        case NODE_BMETHOD:
> -       ruby_frame->flags |= FRAME_DMETH;
> -       result = proc_invoke(body->nd_cval, rb_ary_new4(argc, argv),
> recv, klass);
> +        {
> +            struct BLOCK *data;
> +            ruby_frame->flags |= FRAME_DMETH;
> +            Data_Get_Struct(body->nd_cval, struct BLOCK, data);
> +            EXEC_EVENT_HOOK(RUBY_EVENT_CALL, data->body, recv, id, klass);
> +            result = proc_invoke(body->nd_cval, rb_ary_new4(argc,
> argv), recv, klass);
> +            EXEC_EVENT_HOOK(RUBY_EVENT_RETURN, body, recv, id, klass);
> +        }
>         break;
>
>        case NODE_SCOPE:
>
>
>
> On 7/20/06, Kent Sibilev <ksruby@gmail.com> wrote:
> > Hi Everyone,
> >
> > I've noticed a small bug while playing with my ruby-debug extension. I
> > think the current Ruby interpreter misses a 'call' event when calling
> > a method defined with Module#define_method. Below is the small script
> > that proves it:
> >
> > $ cat -n t.rb
> >      1  class A
> >      2    def m1
> >      3      "m1 called\n"
> >      4    end
> >      5    define_method("m2") do
> >      6      "m2 called\n"
> >      7    end
> >      8  end
> >      9
> >     10  a = A.new
> >     11  set_trace_func proc {|event, file, line, *args| puts "#{event}:#{line}"}
> >     12
> >     13  a.m1
> >     14  a.m2
> >     15
> > $ ruby t.rb
> > line:13
> > call:2
> > line:3
> > return:4
> > line:14
> > line:6
> >
> > Notice how the invocation of A#m2 doesn't emit any 'call' or 'c-call' events.
> >
> > --
> > Kent
> > ---
> > http://www.datanoise.com
> >
> >
>
>
> --
> Kent
> ---
> http://www.datanoise.com
>
>


-- 
Kent
---
http://www.datanoise.com