[ruby-talk:00643] Re: Backtrace of SIGSEGV

On 12 Aug, Yukihiro Matsumoto wrote:
> Hi,
> 
> In message "[ruby-talk:00637] Backtrace of SIGSEGV"
>     on 99/08/12, Clemens Hintze <[e-mail]> writes:
> 
> |I have examined a backtrace for my problem of the interpreter crash! It
> |seems to me, that something with the GC went wrong.
> 
> Thank you.  Here's the patch.

Thank you! But unfortunately the patch doesn't seem to fit in any
version of Ruby I have :-/

I have tried to find that piece of code in Ruby 1.3.6, 1.3.7 and the
latest tarball I have checkout 2-3 days before.

Could you please point me to the code. In line 5382 of file `eval.c'
there begins the body of function `rb_f_binding'. In the whole function
there is not a variable `frame' used! :-(

An improvement proposal! You should perhaps insert the keyword `$Id$' or
at least `$Revision into the sources. Because I can see the revision
`1.1.3.2.32' from your patch below! If we had the keywords above, I
could check if I have the right file, or not.

What do you think?

Sorry for inconvenience,
\cle

> 
> --- eval.c	1999/08/11 07:23:55	1.1.1.3.2.32
> +++ eval.c	1999/08/12 03:53:39
> @@ -5382,4 +5385,5 @@
>  	    frame->argv = argv;
>  	}
> +	frame->tmp = 0;		/* should not preserve tmp */
>  	if (!frame->prev) break;
>  	tmp = ALLOC(struct FRAME);