From: nobu.nokada@...
Date: 2003-05-29T19:02:06+09:00
Subject: Re: RCR: unpack/pack Bignum

Hi,

At Thu, 29 May 2003 14:52:59 +0900,
Robert Feldt wrote:
> > I guess Bignum (un)packing would be used for external data
> > exchange, so it'd be better to provide each singned/unsigned
> > and big/little endian conversion, like as 'l', 'L' and so on.
> > But 'w' is used already.
> >
> Hm, I've envisioned Bignum (un)packing for cryptographic applications
> where nonnegative num <-> MSB-first string is all there is, but maybe
> you're right. But there are very many combinations (signed/unsigned,
> big/little endian for full string and for groups of 2/4 bytes etc). Which
> ones should be supported?

Sorry, there's no combination of signedness and endianness,
'N', 'n', 'V', 'v' are always unsigned.

Rather, what about size specifier extends generic integer
conversion?

  [10].pack("i(8)") => "\0\0\0\0\0\0\0\012" # big endian
                    => "\012\0\0\0\0\0\0\0" # little endian

> What is 'w' used for? I had missed that one.

BER (Basic Encoding Rules) compression.

At Thu, 29 May 2003 18:01:41 +0900,
Robert Feldt wrote:
> Maybe I'm mixing different things here, maybe better to have this a a
> method on Numeric? Numeric#to_raw_string and Numeric.from_raw_string(str)?

It would be nice, but perhaps needs better name. 

-- 
Nobu Nakada