UK professor claims a patent on DBMS implementation techniques I have published on Tunes LLL ml.
Tue Aug 19 13:51:02 2003
I have received this e-mail, my replay follow immediately:
From: "Mr John Wilson" <firstname.lastname@example.org>
Sent: Tuesday, August 19, 2003 2:03 PM
Subject: Compressed database
> USING COMPRESSED BITVECTORS TO IMPLEMENT RELATIONS (D R A F T)
> -- Small steps toward a True Relational DBMSs?
> We saw your paper, we implemented most of this in a system
> patented in 1996. See Computer Journal, 41:5 : 283-296, 1998
> John N. Wilson
> Computer and Information Sciences, University of Strathclyde, Glasgow G1
> 1XH, Scotland
> Phone: +44 (0)141 548 3584 (direct) Fax: +44 (0)141 552 5330 (dept)
I have done a little search on US patent database. Please, can you
confirm that the number of your patent is 6,169,990?
I am a little bit surprised, it is not clear to me where is the novelty
in the techniques your patent claims. The implementation techniques I
have published in that e-mail are old, some very old ("prior art").
I have not claimed an "invention", my was more a synthesis of available
techniques applied to RDBMS implementation and even in this respect some
are not new at all.
I will inform the people of the project I am affiliated (my e-mail was
published on one of the mailing list of such project) and I will quote
your e-mail verbatim because nothing in it seems confidential.
P.S.: Excuse me for my poor English.
He refers to my e-mail:
"Persistence: using compressed bitvectors to implement relations."
and to his paper:
W. Paul Cockshott, Douglas R. McGregor, John Wilson: "High-Performance
Operations Using a Compressed Database Architecture", Computer Journal
Volume 41, Number 5, pp. 283-296, 1998
To read (likely) his patent go to:
"US Patent Full-Text Database Number Search"
then query for patent number 6,169,990.
Reading this seems clear that the principal technique patented is the
pair of domain dictionary and "tokenization". But this is not new: to
support this claim suffice to read  in bibliographic references in my
cited e-mail ("The MacAIMS Data Management System" at MIT, *1971*!!).
Now, even if you, members of the Tunes project, judge not interesting
the technique exposed in my e-mail, I invites you to don't underestimate
this incident: if the project finally overcomes the phase of vaporware
we could encounter other times with the problem of software patents on
publicly available techniques.
Do you have suggestions on how to proceed in this case?