From: Beni Cherniavsky (cben_at_nonexisting.hamakor.org.il)
Date: Fri 21 Feb 2003 - 15:29:10 IST
On 2003-02-20, Arik Baratz wrote:
> From: Beni Cherniavsky [mailto:cben_at_techunix.technion.ac.il]
> > I meant code using patented algorithms. They are viral in the sense
> > that even if you derive another program from it, that algorithm is
> > still patented (unless you get rid of it). So bit 7 is probably
> > appropriate here but in a bad sense :-<<. So for GPL, the virality
> > bit increases the value of the freedom meter while for patented
> > algorithms, it decreases it.
>
> That's a point I have not considered. As the inventor of the freemeter I
> can safely take the reserved bit 5 and designate it as 'uses patented
> algorithms'. This breaks the idea that a 1 bit is more free than a 0,
> but what the heck. Doing the reverse (1 being does not use any known
> patented algorithms) will break backward compatability :-)
>
[This is becoming off-topic]. No problem, you can claim the previous
versions left these bits in the High-Z state :-Z. Or document that nobody
should measure reserved bits until their meaning will be defined, so that
they can remain in a superposion of 0 with 1 without collapsing. For less
physical approach, you could attempt to switch to Conway's `Surreal
numbers`_. I think that if computers were based on them, many format
compatibility problems would be easy to solve. Too bad there are no
surreal memory chips :-).
. _Sureal Numbers: A most elegant concept. Google around or start at:
http://www-math.cudenver.edu/~jstarret/surreal.html
The last link contains images which explain them very well.
Of course the normal way to handle this is to attach a meter version to
the value. The most elegant way to represent versions here would be a
mask of meaningful bits. Now let's weave the mask with the bits (once one
reads about Intercal's operators, he can never stop trying to use them).
So every two bits represent one of "undefined", "1" or "0". Choosing
Representing them as 10/01, 11 and 00 respectively you keep the property
of good order (in other words: when defining previously reserved bits you
can also go lower than the reserved value). This also reminds the
symetric ternary representation of numbers (digits are -1,0,1 and their
wieghts are powers of 3)...
> > That can be explained by treating the number as unsigned normally but
> > as signed bytes in the case of patents. That was the intent of my
> > joke anyway...
>
> Every joke can be looked seriously on, thus making the joker indicate
> his joke as a joke, causing embarresment for the people invloved, to the
> enjoyment of the crowd.
>
> Usign the sign won't do, because there is no way to make the distinction
> between signed and unsigned numbers.
>
Of course there isn't, that was my whole point :-).
--- P.S. Could someone enlighten me on the sofware patents situation in Israel? Is there such a beast, has anybody ever been sued here on software patnet infrigement, etc... I tried to search the net but didn't find anything clear. -- Beni Cherniavsky <cben_at_tx.technion.ac.il> Do not feed the Bugzillas. ================================================================= To unsubscribe, send mail to linux-il-request_at_linux.org.il with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail linux-il-request_at_linux.org.il
This archive was generated by hypermail 2.1.7 : Mon 06 Oct 2003 - 23:44:22 IST