Mailing List Archive

gpg <-> pgp5
I've had trouble trying to get pgp5 and gpg to interact. Windows pgp5
doesn't accept keys from pgp. I tried both exporting them as binary and
text, and I tried using the options -z0, --no-commentm, --algo.*, etc. pgp5
for Linux recongnizes the key, but treat them as Sign only (in the case of
DSA), or encrypt only (elgamal). What am I doing wrong?

I'm trying to push free-software in my job, and they are now looking for
source code of pgp-compatible software. They have already agreed to the
posibility of using free-software. Is gpg ready for a poduction environment?
We would need to port it to AS/400...!

Bye!
Re: gpg <-> pgp5 [ In reply to ]
Nicolás Lichtmaier <nick@feedback.net.ar> writes:

> I've had trouble trying to get pgp5 and gpg to interact. Windows pgp5

Sorry, I have no pgp5 for Windows, so I can't talk about it.

> for Linux recongnizes the key, but treat them as Sign only (in the case of
> DSA), or encrypt only (elgamal). What am I doing wrong?

pgp does not support the OpenPGP algorithm identifier 20 for ElGamal
keys which are valid for signtures and encryption. I have changed the
default agorithms to 17,16 as pgp 5 does.

> I'm trying to push free-software in my job, and they are now looking for

A Good Thing(tm)

> source code of pgp-compatible software. They have already agreed to the
> posibility of using free-software. Is gpg ready for a poduction environment?

No: 0.3.2 ;-) The next release will be better, but we need more
testers.

> We would need to port it to AS/400...!

Is it POSIX machine? I know that the /36 is a ugly machine from
a programmers point of view - Years ago I tried to port an OS/2, MVS,
DOS program to the /36 and it was not very successful.


Werner
Re: gpg <-> pgp5 [ In reply to ]
> > I've had trouble trying to get pgp5 and gpg to interact. Windows pgp5
> Sorry, I have no pgp5 for Windows, so I can't talk about it.
> > for Linux recongnizes the key, but treat them as Sign only (in the case of
> > DSA), or encrypt only (elgamal). What am I doing wrong?
> pgp does not support the OpenPGP algorithm identifier 20 for ElGamal
> keys which are valid for signtures and encryption. I have changed the
> default agorithms to 17,16 as pgp 5 does.

Oh.. and pgp5 is recognizing only the DSA part of the "(1) DSA and ElGamal
(default)" key... I guess...

> > I'm trying to push free-software in my job, and they are now looking for
> A Good Thing(tm)
> > source code of pgp-compatible software. They have already agreed to the
> > posibility of using free-software. Is gpg ready for a poduction environment?
> No: 0.3.2 ;-) The next release will be better, but we need more
> testers.

Hum..

> > We would need to port it to AS/400...!
> Is it POSIX machine? I know that the /36 is a ugly machine from
> a programmers point of view - Years ago I tried to port an OS/2, MVS,
> DOS program to the /36 and it was not very successful.

It has a POSIX API and a C compiler. But it probably won't be trivial.
Specially when gpg seems to be more Linux centric than other free software
utilities. It will surely become more portable in the future.

Oh.. BTW, I tried to compile it in Solaris. It failed trying to lock the
memory because it was getting it with malloc (not with mmap, I think there's
not MAP_ANONYMOUS there) and it wasn't page aligned. The problem is fixed
changing malloc for valloc. I don't have the sources here to send a patch.
Re: gpg <-> pgp5 [ In reply to ]
Nicolás Lichtmaier <nick@feedback.net.ar> writes:

> It has a POSIX API and a C compiler. But it probably won't be trivial.

NT has a POSIX API too, but it is useless as you can ONLY do what is
required by POSIX. OpenVMS is better.

> Specially when gpg seems to be more Linux centric than other free software

The only Linux/*BSD part is the random device; every other thing should
work fine also I can test on HPUX only every month or so.

> memory because it was getting it with malloc (not with mmap, I think there's
> not MAP_ANONYMOUS there) and it wasn't page aligned. The problem is fixed

I think I have to use MAP_ANON here.


Werner