|
Authored by: Anonymous on Tuesday, May 08 2012 @ 09:30 PM EDT |
Isn't this programming 401?
Or at least covered in Knuth's book from 198#.
The book he almost stopped writing to creat TeX.[ Reply to This | Parent | # ]
|
|
Authored by: PolR on Tuesday, May 08 2012 @ 09:43 PM EDT |
Code comparison is not useful because the patent is not on the code written by
Oracle. This is not like copyrights which covers the code. A patent covers that
the software does when it executes. Any software that does what is described in
a patent claim when it runs will infringe no matter of how similar or dissimilar
to Oracle code. The comparison must be between Google's code and the words of
the claims.
What Google must do is find one thing which is recited in the claim that their
code doesn't do. Then the claim is not infringed. Given that typical patent
language describe algorithms in a fuzzy manner, quite a lot of different
algorithms may infringe.
IANAL but this is how I understand infringement.[ Reply to This | Parent | # ]
|
- Yabut - Authored by: Tkilgore on Tuesday, May 08 2012 @ 09:56 PM EDT
- Several tests - Authored by: PolR on Tuesday, May 08 2012 @ 10:25 PM EDT
- Yabut - Authored by: greed on Tuesday, May 08 2012 @ 11:23 PM EDT
|
Authored by: mschmitz on Tuesday, May 08 2012 @ 09:53 PM EDT |
Doesn't matter that it's a well known algorithm - they were awarded a patent on
it, that's all it takes. Patent subject matter isn't supposed to be novel,
non-obvious and innovative, now is it? At least not as far as the US patent
office is concerned.
Resolving symbolic references into pointers (i.e. addresses) - that must've been
obvious to anyone in IT since the dawn of the internet. I believe we used to
call it 'late binding' as far as it applied to shared libraries (Linux' a.out to
ELF transition, ca, 1995?) and simply 'linking' or 'binding' as far as it
applied to object code.
A similar but simpler process called 'relocation' does not involve
symbol-to-address conversions and probably can be left out of the discussion
here. Now what is the incredibly clever, non-obvious and novel approach Oracle
take when it comes to dynamic loading/late binding?
Appalled ...
-- mschmitz[ Reply to This | Parent | # ]
|
|
Authored by: Anonymous on Tuesday, May 08 2012 @ 09:58 PM EDT |
What this patent covers is not "a method and apparatus". ITS AN IDEA, AND A
SIMPLE AND OBVIOUS ONE AT THAT.
All software patents are garbage. A
very small fraction of them (a very small fraction!) cover some ingenious idea,
and basically prevent anyone from knowingly using it (other than big companies,
who can afford to pay extor... er, license fees). The rest just cover simple
mundane ideas and that is even worse.
For those who never knew this.. In the
field of data compression, software patents helped retard progress for almost 20
years. A simple mathematical trick called Arithmetic coding was
discovered in the 60's, but didn't get much attention until 1975. In 1987,
Witten et al. published an implementation in the Communications of the ACM. But it had
already been patented.
There are dozens of patents on variations of
arithmetic coding, and all throughout the 90's it was considered too dangerous
to use unless you were willing to pay a large company for a patent license. IBM
had a dozen or so patents about arithmetic coding, including a couple of
important tricks for making it efficient to actually use on computers of that
time. AT&T also had a bunch of patents on it.
Oh, but wait: there's a
"different" technique called Range
coding, which was published in a 1979 paper, and so by the late 90's was
believed to be in the public domain (because the only patents covering it dated
from 1977-78, and had expired). I remember working for a company that
implemented Range coding in a mobile device. The engineer implementing it
cheerfully explained that he couldn't use arithmetic coding because it was
heavily encumbered by patents, but the lawyers had assured him that Range coding
was not patented anymore, and they didn't care that the two techniques were
mathematically the same thing. That's right: Range coding and arithmetic coding
are the same thing! Two algorithms that are mathematically equivalent, yet one
of them is in the public domain, and the other is encumbered by dozens of
patents! I think that was my first clue that the patent system was an unholy
mess, and in the 17 years since, things have gotten worse, not better.
[I'll also point out here that our employer's policy was that we were never
to look at any patents, because it would potentially expose them to triple
liability if the company ever lost an infringement suit and it turned out that
one of us had ever seen the patent in question. I think nothing was wasted
there, because the few I did ever look at were all written in
lawyer-gobbledygook and only contained ideas, never actual implementations that
might have been useful for something. Software patents do absolutely nothing to
promote the Progress of Science and useful Arts, and indeed seem to have the
opposite effect.]
He
re's a list of some of the U.S. patents on compression
algorithms.
Nearly all of these "inventions" can be implemented
completely in software, because they are just mathematical algorithms. The
history of data compression algorithms is littered with patents. Some of them
were just ignored, others effectively killed off the compression algorithm that
they covered (no one wanted to use it anymore, after they found out that it was
patented). Even the simplest of compression algorithms, like LZ77 and
Run-Length Encoding, have been patented in the U.S. Somebody also had a patent
on using those two algorithms in combination.
These patents, and ALL
software patents, simply prevent other people from using certain good ideas
(even ones they thought up by themselves) without paying a toll. Mathematics
has no physical substance. It isn't invented, it is discovered. By allowing
software patents, we allow these companies to have a 17-year monopoly on an
idea. Which is harmful to everyone in the end. [ Reply to This | Parent | # ]
|
|
Authored by: Anonymous on Wednesday, May 09 2012 @ 01:55 AM EDT |
You are applying logic here, but in a court, it is quite possible to find for
both a proof and a counterexample. For one thing, logic is fuzzy rather than
binary. For another, one question is resolved only loosely dependent from the
next. And the whole point of employing a jury rather than a judge is to arrive
at answers through an empiric, common sense oriented process rather than
mathematical logic and a single mind.[ Reply to This | Parent | # ]
|
|
Authored by: Anonymous on Wednesday, May 09 2012 @ 01:21 PM EDT |
Anyone remember a company 'Management Science America' (MSA)? They
had a software product called 'Information Expert' (IE). IE had a language that
a user could write, compile and run, as needed. As IBM hardware was their
main target, it was written in assembler and some COBOL.
With the increased popularity of other hardware vendors, there was a
'portable' version of IE. The portable IE was written in a language common
across the other vendors. The IE language, and other components of IE,
functioned identical to the IBM specific version. The portable version would
also run on IBM hardware running MVS or MVS/XA. The language was
compiled into a byte code. The byte code was input into an interpreter/VM.
The byte code contained references that at runtime were 'dereferenced' into
actual addresses. This was in the early 80's. As far as I know, there is no
documentation for it though. [ Reply to This | Parent | # ]
|
|
|
|
|