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

Re: nomarch-1.2 - Problems with RLE patent 4,586,027

Hi There!

Checked out the other 2 patents on RLE;

As a result of this nomarch can be included in main, and the other
packages (SDL, OpenGL and Mesa) where RLE are used can stay where they

After sitting down and carefully looking at them for most of the day,
the summary is as follows:

US pat 4,586,027 does not apply to nomarch or RLE as below.

US pat.4,872,009 has preanalysis and dictionary techniques in  it, as
well as being bidirectional.  Basic RLE and algorithm in nomarch is not
like this.

US pat. 4,988,998 is about RLE being used and mixed in with LZW...  not
applicable to nomarch or ordinary RLE...

The compression FAQ referenced below is wrong about ordinary RLE being
patented. in the above patents as they are either quite specific in
their application where RLE is combined with the patented methods as a
minor part of the method and not isolated as a stand alone item or claim
(4,586,027 4,988,998), or they are about compression methods that are
significantly different to ordinary RLE by the use of pre-analysis and
data dictionaries (4,872,009).  

Where ordinary RLE is in a patent it appears that it is used as a common
well known technique and by itself would not be held as patentable.

I am not a lawyer, but I am trained as a Mechanical Engineer so I do
have some skill in reading these sort of documents.  If there is any one
here with a knowledge of patents could they please check my findings,
and get back to me if they can find fault with my reasoning.

I hope that this has cleared this up.

Best regards,

Matthew Grant

On Fri, 2002-05-17 at 16:06, Matthew Grant wrote:
    HOLD!   Don't do ANYTHING, Don't worry! 
    Currently investigating nomarch and patent 4,586,027 applies to hardware
    RLE compression system for tape drives, even though the claims section
    does not mention this, most of the claims are specific to a logic
    assembly to implement algorithm where uncompressed sections have
    preceding and trailing length counts, and compressed sections are datum
    followed by count (basic RLE).  There is also a threshold repeat value
    built in to. From looking at it, the patent stands because of the other
    features combined with basic RLE, and does not cover basic RLE by
    Conlusion: US Patent 4,586,027 does not apply to nomarch, which RLE's by
    identifying a repeat count of preceding byte with an escape code of
    0x90, and the escape code in the input is represented by specially
    encoding a repeat count of zero.
    Going to look at US patents 4,872,009, and 4,988,998 (RLE followed by
    On Fri, 2002-05-17 at 13:19, Daniel Burrows wrote:
        On Fri, May 17, 2002 at 11:58:28AM +1200, Matthew Grant <grantma@debian.org> was heard to say:
        >    Could you please withdraw the nomarch package from the main upload queue. 
        >    This thing belongs in non-free.
          If it violates a patent, shouldn't it be removed entirely?
        >    Sorry about this. I have tried to be careful. Who would have thought that
        >    such an obvious compression technique (Run Length Encoding - count number
        >    of bytes in a row of same value, store count and byte value) is covered by
        >    a software patent?
          Hm, don't SDL and OpenGL...errr, I mean, Mesa...both use RLE compression
        to accelerate some blits?  Should we remove them too?
          I have to say that this patent is pretty despicable, even for a software
        patent.  For people who didn't bother following the URL:
        "The first one covers run length encoding in its most primitive form: a
         length byte followed by the repeated byte."
          Daniel, who thinks he "invented" that one sometime in junior high
        /-------------------- Daniel Burrows <dburrows@debian.org> -------------------\
        |                              The Turtle Moves!                              |
        \------- Listener-supported public radio -- NPR -- http://www.npr.org --------/
    Matthew Grant	     /\	 ^/\^	grantma@anathoth.gen.nz      /~~~~\
    A Linux Network Guy /~~\^/~~\_/~~~~~\_______/~~~~~~~~~~\____/******\
Matthew Grant	     /\	 ^/\^	grantma@anathoth.gen.nz      /~~~~\
A Linux Network Guy /~~\^/~~\_/~~~~~\_______/~~~~~~~~~~\____/******\

Attachment: signature.asc
Description: This is a digitally signed message part

Reply to: