deprecate MARS due to patent encumbrance?

Zooko Wilcox-O'Hearn

unread,

Jun 5, 2009, 9:32:14 AM6/5/09

to Crypto++ Users

Folks:

I'm trying to get pycryptopp and Tahoe-LAFS into Fedora, and one tiny
little detail that I have to work-around is that MARS is patented.
Fedora already includes a version of Crypto++ 5.6.0 with mars.cpp
stripped out, so the only reason I even have to be aware of this
detail is that I included a copy of Crypto++ v5.6.0 inside the
pycryptopp source distribution, and now I suppose I need to remove
mars.cpp from that embedded copy.

Anyway, I was thinking that it might be nice to let Crypto++ be more
easily/more widely re-used by removing mars.cpp from the Crypto++
upstream. I've never heard of anyone using MARS cipher, personally.
Have you?

Regards,

Zooko

Jeffrey Walton

unread,

Jun 5, 2009, 9:54:49 AM6/5/09

to Zooko Wilcox-O'Hearn, Crypto++ Users

Hi Zooko,

Rather than strip it out, how about a #define: INCLUDE_INCUMBERED or
similar. This way, Crypto++ does not loose features, and folks such as
yourself can choose to gut the encumbered algorithms.

Jeff

Jeffrey Walton

unread,

Jun 5, 2009, 9:55:32 AM6/5/09

to Zooko Wilcox-O'Hearn, Crypto++ Users

Sp: INCLUDE_ENCUMBERED

Zooko O'Whielacronx

unread,

Jun 5, 2009, 11:17:57 AM6/5/09

to Crypto++ Users

As I understand it, the Fedora folks are asking me to give them a
source distribution (i.e. a tarball of source code files) which don't
include a MARS implementation at all. It is apparently not good
enough that we don't actually build it or use it.

Regards,

Zooko

Florian Weimer

unread,

Jun 6, 2009, 5:23:43 AM6/6/09

to Zooko Wilcox-O'Hearn, Crypto++ Users

* Zooko Wilcox-O'Hearn:

> I'm trying to get pycryptopp and Tahoe-LAFS into Fedora, and one tiny
> little detail that I have to work-around is that MARS is patented.
> Fedora already includes a version of Crypto++ 5.6.0 with mars.cpp
> stripped out, so the only reason I even have to be aware of this
> detail is that I included a copy of Crypto++ v5.6.0 inside the
> pycryptopp source distribution, and now I suppose I need to remove
> mars.cpp from that embedded copy.
>

Huh? The IETF has got a statement from IBM on file, stating that "IBM
has publically announced that as of January 18, 2000 we are making
MARS available on a royalty free basis woldwide regardless of the AES
outcome." Or stem the patent problems from a non-IBM patent?

Zooko Wilcox-O'Hearn

unread,

Jun 7, 2009, 10:36:09 AM6/7/09

to Florian Weimer, Crypto++ Users

On Jun 6, 2009, at 3:23 AM, Florian Weimer wrote:

> Huh? The IETF has got a statement from IBM on file, stating that "IBM
> has publically announced that as of January 18, 2000 we are making
> MARS available on a royalty free basis woldwide regardless of the AES
> outcome." Or stem the patent problems from a non-IBM patent?

Thanks for the information. I passed it on to the Fedora folks who
had requested that I remove mars.cpp from the pycryptopp source
distribution.

Of course, I already *have* removed it, so the issue is moot as far
as pycryptopp is concerned, but it might make things simpler for them
if they can stop rm'ing mars.cpp from each new release of Crypto++.