Re: applying AES-GCM to secure shell: proposed "tweak"

[nisse%lysator.liu.se@localhost (Niels Möller)]

"denis bider \(Bitvise\)" <ietf-ssh2%denisbider.com@localhost> writes:

> I'm kinda surprised by the attitude "there's only one zlib".
>
> We use the Crypto++ implementation, which is independent of zlib, and
> is freely available.

In order to figure out what complexity really is needed, I had to
write my own implementation. I'm not going to claim that it's less
buggy than zlib (it's the result of a few evenings work and not
thouroghly tested), but it's certainly a lot smaller than zlib and
(imo) well designed and fairly readable. See
http://www.lysator.liu.se/~nisse/archive/inflate-0.2.tar.gz. For now,
it's GPL:ed.

The interesting files, huffman.c and inflate.c, are in total 674
lines, or 490 excluding the comments (2015 and 1467 for the
corresponding files in zlib, inflate.c, inffast.c and inftrees.c).
Performance is a factor two slower than the ordinary zcat program.

Best regards,
/Niels