Re: Module.new(&block) in Ruby 1.9

Sylvain Joyeux wrote:
> And for the 'def in module_eval' issue ? I saw the thread, but frankly
> the description of the fix do not speak to me. Since you already have
> fixed it, maybe you could provide us with the patch ?

 I can't because I was lazy on this patch, rather than waiting a
 modified version of rb_iseq_t, in my private version I duplicate it
 even if it's a big structure

vgs% fgrep rb_iseq_dup *.c
class.c:		    iseqval = rb_iseq_dup(node->nd_rval);
class.c:		    obj = rb_iseq_dup(proc->block.iseq->self);
eval.c:	    iseqval = rb_iseq_dup(block.iseq->self);
iseq.c:rb_iseq_dup(VALUE orig)
vm_insnhelper.c:	iseqval = rb_iseq_dup(miseq->self);
vgs%

    VALUE
    rb_iseq_dup(VALUE orig)
    {
        VALUE volatile obj;
        rb_iseq_t *iseq, *oiseq;

        obj = iseq_alloc(CLASS_OF(orig));
        GetISeqPtr(obj, iseq);
        GetISeqPtr(orig, oiseq);
        MEMCPY(iseq, oiseq, rb_iseq_t, 1);
        iseq->self = obj;
        iseq->orig = oiseq->self;
        if (oiseq == oiseq->local_iseq) {
            iseq->local_iseq = iseq;
        }
        return obj;
    }

 Normally this function must be changed in rb_iseq_partial_copy(),
 (or something like this) when the modification given in
 [ruby-core:14728] will be done.

 It was more important, for me, to have the possibility to test my
 modifications rather than have a "clean" version.



Guy Decoux