From: Arlen Cuss <celtic@...>
Date: 2008-02-22T08:20:07+09:00
Subject: Re: Kernel#putc vs. $stdout#putc

------=_Part_6306_28992168.1203636013515
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

Hi,

On Fri, Feb 22, 2008 at 5:52 AM, Gordon Thiesfeld <gthiesfeld@gmail.com>
wrote:

> OK, here's why I'm confused.  In the code below, why isn't the output
> of both of the putc methods 42?


Here's the relevant code from io.c:

static VALUE
rb_io_putc(VALUE io, VALUE ch)
{
    char c = NUM2CHR(ch);

    rb_io_write(io, rb_str_new(&c, 1));
    return ch;
}

static VALUE
rb_f_putc(VALUE recv, VALUE ch)
{
    return rb_io_putc(rb_stdout, ch);
}


As you can see, rb_f_putc (Kernel.putc) is hard-wired to rb_io_putc
(IO#putc) - but it's directly hard-wired (I believe), not literally as
"$stdout.putc" but rather going straight to the function as defined here.
This means that the two functions are 'identical' in nature --
Kernel.putcisn't $stdout.putc, but rather will always call
rb_io_write(rb_stdout,
rb_str_new(&ch, 1)), which is $stdout.write with a string of that one
character.
Hence, you actually would have to redefine both ($stdout.putc, by your
ChunkyIO, *and* Kernel.putc) -- or redefine $stdout.write (in which case
both functions will be affected) -- in order to get the effect you want.


Cheers,
Arlen.

------=_Part_6306_28992168.1203636013515--