From: David Alan Black <dblack@...>
Date: 2001-07-02T22:48:59+09:00
Subject: [ruby-talk:17145] Re: RCR: Exception methods (was: Re:  Re: keeping an Exception object alive)

Hello --

On Mon, 2 Jul 2001, ts wrote:

> >>>>> "D" == David Alan Black <dblack@candle.superlink.net> writes:
>
> D> A followup question... how useful/used is Exception#exception?  Maybe
> D> I've just been staring at it too long :-) but I'm wondering what it
> D> does that couldn't be done already.  I notice it isn't used at all in
> D> the distribution (plus whatever site_ruby stuff I've got):
>
>  You have not made the good grep :-)
>
> pigeon% grep \"exception *.c
> error.c:    rb_define_singleton_method(rb_eException, "exception", rb_class_new_instance, -1);
> error.c:    rb_define_method(rb_eException, "exception", exc_exception, -1);
> eval.c: exception = rb_intern("exception");
> eval.c:     rb_raise(rb_eTypeError, "exception class/object expected");
> eval.c:     rb_raise(rb_eTypeError, "exception object expected");
> pigeon%
>
> pigeon% less eval.c
> [...]
>         exception = rb_intern("exception");
>         if (!rb_respond_to(argv[0], exception)) {
>             rb_raise(rb_eTypeError, "exception class/object expected");
>         }
>         mesg = rb_funcall(argv[0], exception, n, argv[1]);
>         break;

True, my grep is a bit selective :-)  But still, if the only reason
#exception exists is to allow both:

     raise ExceptionClass

and

     raise anException

then I wonder whether it should perhaps be hidden -- that is, whether
the test for valid argument to #raise should be internal to eval.c,
and not resurface as a public instance method of Exception.


David

-- 
David Alan Black
home: dblack@candle.superlink.net
work: blackdav@shu.edu
Web:  http://pirate.shu.edu/~blackdav