From: Carl Youngblood <carl.youngblood@...>
Date: 2004-10-24T00:17:51+09:00
Subject: Re: Ruby/DL: bignum->long

I think the call is failing because of some code in bignum.c:

lines 765-776:

long
rb_big2long(x)
    VALUE x;
{
    unsigned long num = big2ulong(x, "long");

    if ((long)num < 0 && (RBIGNUM(x)->sign || (long)num != LONG_MIN)) {
	rb_raise(rb_eRangeError, "bignum too big to convert into `long'");
    }
    if (!RBIGNUM(x)->sign) return -(long)num;
    return num;
}

It checks to see if the number goes negative when trying to convert it
to a long, and if so, raises an exception.  In this case you are using
a hex value to set an integer that you intend to be negative, but Ruby
treats this as a positive integer.  Perhaps you could get around it by
putting the number in as a negative:

#!/usr/bin/env ruby

require 'dl'

DL::dlopen("libc.dylib") do |libc|
	isdigit  = libc['isdigit', 'CH']
	p 0x40000000.class
	print(isdigit.call(0x40000000), "\n")
	p -2147483648.class
	print(isdigit.call(-2147483648), "\n")
end

Here is the output:

Bignum
01073741824
Bignum
0-2147483648

On Sat, 23 Oct 2004 13:23:22 +0900, Jamis Buck <jgb3@email.byu.edu> wrote:
> Using Ruby/DL, it seems as though a 32-bit integer with the MSB set
> cannot be converted to a long integer:
> 
>    require 'dl'
> 
>    DL::dlopen("libc.so.5") do |libc|
>      isdigit  = libc['isdigit', 'CH']
>      p 0x40000000.class
>      print(isdigit.call(0x40000000), "\n")
>      p 0x80000000.class
>      print(isdigit.call(0x80000000), "\n")
>    end
> 
> The above will print the first one successfully, even though 0x40000000
> is a Bignum. The second one results in an error:
> 
>    in `call': bignum too big to convert into `long' (RangeError)
> 
> Is this a bug, or a feature? Is there a workaround? Any advice would be
>    appreciated.
> 
> Thanks,
> 
> Jamis
> 
> --
> Jamis Buck
> jgb3@email.byu.edu
> http://www.jamisbuck.org/jamis