[PeerToPatent Australia] Invitation

Pedro Gimeno gmpdiscuss at personal.formauri.es
Sat Jan 2 07:45:57 CET 2010

Richard B. Kreckel wrote:

> If I understand correctly, this is indeed one of the most apparent 
> "inventions" ever, right? If the FPU's registers are not wide enough, 
> you have to represent the numbers somehow differently. Well, and if you 
> care about speed, you gotta use machine words. The rest is maths. So, 
> the grand idea they came up with is only marginally more sophisticated 

I've taken a look at the patent; it's already filed in the U.S. with 
number 7,523,150.

The patent covers base conversion. Actually it's a generalization of 
hexadecimal to base-N. It can affect GMP only in the cases where N is a 
power of 2 where the exponent divides the target word size, e.g. 
hexadecimal (base 2^4) in 32 or 64 bit systems, because 4 divides 32 and 
64. It wouldn't affect conversion to/from octal, for example, because 3 
doesn't divide 32 or 64.

The concept behind the idea being patented is to store the number in 
base max(N^i : N^i<=word size), where the base-N^i digits are called 
'segments'. For example, for storing base-10 numbers the number is 
stored in 'segments' valued 0 to 9999 in a 16-bit processor, because 
10000 is the maximum power of ten that fits in 16 bits.

It's a trivial variation of my own 'petradecimal' idea which I used in 
2005 to extend a counter beyond the limits of a long using very little 
extra code:


(search for appearances of 'Petra')

The 'innovation' here with respect to my code is that it uses a variable 
number of base-N^i digits, instead of the two which I thought were 
enough for my purpose. I actually didn't care to make them uniform 
because that way I could store bigger numbers.

The fact that projects like Peer To Patent exist reveals in itself 
what's broken with the software patent minefield. Not even the patent 
office knows whether a patent has prior art, how can a regular 
programmer know when he's hitting one while he does his creative job? 
How is it possible that such trivial things that anyone would rediscover 
when the need raises are being considered and even granted? That's the 
protection of a monopoly, not of an innovation.

And no, I won't help P2P with enhancing software patents' quality. 
Software patents must be eliminated, not enhanced.

More information about the gmp-discuss mailing list