Partial body length encoding for Compressed packets

Hasnain Mujtaba hmujtaba at
Tue Feb 8 19:53:15 CET 2005

Hi David,

There's food for thought in your comments. Sanity is very important to
me. So, thanks. 

On a techincal level, I was only curious about this mixing and matching
of partial body length headers and indeterminate encoding. I guess I
will never know for sure why it is that way with 2.x, and perhaps 5.x. 


-----Original Message-----
From: gnupg-users-bounces at
[mailto:gnupg-users-bounces at] On Behalf Of David Shaw
Sent: Tuesday, February 08, 2005 11:35 AM
To: gnupg-users at
Subject: Re: Partial body length encoding for Compressed packets

On Mon, Feb 07, 2005 at 05:09:49PM -0500, Hasnain Mujtaba wrote:
> Hi David,
> I generated a PGP 2 style packet using GPG's --pgp2 option and tore it
> apart to look inside its structure. I found that both the encrypted
> literal data packets are broken into RFC 2440 style partial body
> chunks. But, as you explained, the compressed packet was indeterminate
> length encoded. 
> I find this behavior perplexing. They made PGP 2 RFC2440 compliant
> for encrypted and literal packets. But why not for compressed packets
> well?

PGP 2 isn't RFC-2440 compliant.  PGP 2 dates from quite a few years
before 2440 was even written.

> Is this behavior found in PGP 5.x. PGP 6.x, and PGP 7.x as well, i.e
> 2440 chunking for literal & encrypted packets, but indeterminate for
> compressed packets? Or will PGP 5.x and above, understand chunking for
> all three packets?
> Not meaning to beat on a dead horse, but this forum is my only hope of
> staying sane in a world of interoperability minefields.

It won't work ;)

Sane PGP interoperability requires knowing when to give up.  For
example, there are details between PGP 5 and 7 where you can support
one or the other, but not both.

To a certain extent, supporting bugs from old versions that have been
replaced many times over is actually harmful.  I know that some users
have settled on one version of PGP and will continue to use that
version until the sun goes nova, but given the choice between
supporting that tiny minority of people, and the huge majority of
people who are using something actually RFC compliant, I know where
I'm going to spend my energy.


Gnupg-users mailing list
Gnupg-users at

More information about the Gnupg-users mailing list