I'm coding an external that can is like a [receive] object, but it can expand its "$0" argument according to the parent patch if you have a first float argument for the depth level.

I can do it if I escape it with a backslash, as in [receiver 1 \$0-x], but I wish I could just do it without escaping. I can do that if I query for the raw argument in "x->x_obj.te_binbuf".

The problem is I can't do that in the "new" function, since Pd just crashes. On the other hand, I do need to bind the symbol right there so the object is properly initialized.

I tried using a clock with '0' delay in the new function and then initialize it like this:


static t_symbol *receiver_get_rcv(t_receiver *x){

    int idx = 1 + x->x_depth_arg;

    t_binbuf *bb = x->x_obj.te_binbuf;

    if(binbuf_getnatom(bb) > idx){

        char buf[128];

        atom_string(binbuf_getvec(bb) + idx, buf, 128);

        return(gensym(buf));

    }

    else

        return(&s_);

}


static void receiver_bindlater(t_receiver *x){

    x->x_raw = receiver_get_rcv(x);

    x->x_sym = canvas_realizedollar(x->x_cv, x->x_raw);

    if(x->x_sym != &s_){

        pd_bind(&x->x_obj.ob_pd, x->x_sym_1);

        x->x_bound = 1;

    }

    clock_free(x->x_bindclock);

}


It works, but if I have a patch that is initialized with a [loadbang] that sends a value to this receiver, I won't get it unless I also use [delay 0] in the loadbang.

So what is my hack now, I am using the "init bang", something like

static void receiver_loadbang(t_receiver *x, t_floatarg action){

    if(action == LB_INIT)

        receiver_bind(x);

}


It works, as it is initialized before a loadbang in the sender, so it's "cool".

Nonetheless, I wanted to come here to ask if I am being smart, or if there's a better or smarter way to do this. Or, if this is not smart at all and there's no safe way to do this.

thanks