[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: status of bzip

On Fri 12 Dec 1997, Paul Slootman wrote:
> On Thu 11 Dec 1997, Guy Maor wrote:
> > Andy Guy <awpguy@acs.ucalgary.ca> writes:
> > > It is different becuase the lzw patent holders (HP?) have given a
> > > general license for non-profit use of the patent.
> > 
> > That's true.  It's Unisys that holds the patent, btw.
> > 
> > The patent on bzip is moot anyway, as bzip2 does not have any patents
> > on it.  It should go to main.  I think we should discourage the use of

I've received word from the author. Apparently there doesn't seem to be
a concrete patent on the bzip arithmetic coding (not lzw). However,
different arithmetic coding compression schemes _are_ patented, and to
prevent possible trouble, the author decided to remove bzip.

> > bzip by not having it in the archive.  Someone mentioned that there
> > was a bunzip which could go to non-free?
> I started this thread, and mentioned that there is a bunzip-only source
> available. I suspect that that will also have to go into non-us, if the
> original bzip also had to go to non-us due to (silly) US patents.

The author (<Julian_Seward@muraroa.demon.co.uk>) thinks that the bunzip
should be freely distributable; he doesn't know of any restrictions.
So, perhaps the bunzip code could be included in the bzip2 package;
alternatively, offer it as an "extra" package.

BTW, he seems to have a new version of bzip2 almost ready, with improved

Here's the significant part of his message for completenes:

Date: Fri, 12 Dec 1997 21:41:31 +0000                                           
From: Julian Seward <Julian_Seward@muraroa.demon.co.uk>                         
To: Paul Slootman <paul@wau.mis.ah.nl>                                          
Subject: Re: question about the patent problem with bzip                        

Oh, what a tedious business (patents).  The situation, as I see it, is:

-- Arithmetic coding, used in bzip but not bzip2, is widely patented
   in its many guises.  The people who wrote the paper from which bzip's
   arithcoding code is derived (Moffat et al) are academics, and when
   I asked, they all said that they knew of no patent on their work.

   Nevertheless, considering the general paranoia of the GNU people,
   I decided to remove the technique from bzip2, since I knew it would
   never get widely used so long as a (possibly perceived) patent
   risk was associated with it.  Also, arithcoding is troublesome from
   a technical point of view -- it is pretty slow -- so I was happy to
   wave bye-bye to it on engineering grounds as well.

-- I don't perceive the bunzip code on my page as much of a risk.
   Not many people got bzip (far fewer than who have bzip2).  And
   surely a decompression-only version is pretty harmless?

So: I don't know of any specific patent which bzip infringes.  I think
it would be ok to distribute bunzip, along with the full bzip2 package.

As in all patent-related matters, #include <standard_disclaimer.h>.
If you want assurances of anything, please see a lawyer.  This message
does not constitute an assurance of anything.  (end-of-paranoia)
If you intend to use bzip2 as part of the infrastructure of Debian,
rather than merely as a distributed package, I think I would like you
to have a more recent version which I have more-or-less completed.
bzip2 contains some nasty performance anomalies -- the worst-case-time
vs average-case-time to compress can exceed 100:1 in some rare
and I have fixed that.  Also, the decompression code in bzip2-0.1pl2 is
slightly naive, and can easily be made to decompress about 15% faster.


Paul Slootman
Can you get your operating system fixed when you need it?
Linux - the supportable operating system. http://www.debian.org/support.html
home: paul@wurtel.demon.nl | work: paul@murphy.nl
http://www.wurtel.demon.nl | Murphy Software, Enschede, the Netherlands

TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-request@lists.debian.org . 
Trouble?  e-mail to templin@bucknell.edu .

Reply to: