|
|
|
| Authored by: BJ on Tuesday, May 22 2012 @ 07:21 PM EDT |
... make yourselves out as the ones I disagree with.
Fine, but don't try to reserve some kind of objective
view for yourself.
Off to bed.
Sillyness!
bjd
[ Reply to This | Parent | # ]
|
| |
| Authored by: eric76 on Tuesday, May 22 2012 @ 07:25 PM EDT |
But instead of that argument, Google overreached (I think a little
bit) by saying that if the instruction doesn't literally contain a symbol then
it can't be called an instruction that contains a symbolic reference. The jury
has seen through that and are not ready to accept that extreme view, even
without the technical understanding to know what is wrong with that. I give them
much credit for that.
That's hardly overreaching. If there is
no symbol that refers to a piece of data, then there is no symbolic
reference.
If there was a symbolic reference, just what would be the need for
the symbolic reference and to what symbol would it refer? Sure, you could create
instructions with random symbolic references, but there would be no point to
doing so unless your wish is to maximize inefficiency. [ Reply to This | Parent | # ]
|
| |
| Authored by: Anonymous on Tuesday, May 22 2012 @ 07:44 PM EDT |
Because Dalvik bytecode instructions do not contain
symbolic references,
Google does not infringe any of the
asserted claims of the ’104
patent.
There is no evidence that either Resolve.c or the dexopt
tool
operate on "instructions containing one or more
symbolic references", as
required by asserted Claims 11, 27,
29, 39, 40, 1 and 41 of the ‘104
patent.
The Court has construed the term “symbolic reference” to
mean...
"a reference that identifies data by a name other than
the
numeric memory location of the data, and that is
resolved dynamically rather
than statically". [ Reply to This | Parent | # ]
|
| |
| Authored by: Gringo_ on Tuesday, May 22 2012 @ 07:57 PM EDT |
Google overreached (I think a little bit) by
saying that if the
instruction doesn't literally contain a
symbol then it can't be called an
instruction that contains
a symbolic reference.
Look here,
Mr. Bug Stomper, if you can't take the patent
literally, then how are you
supposed to interpret it - with
your imagination? Why do you suppose lawyers go
to school
and do articaling all those years - so they can make vague
statements? They are supposed to use precise language in a
patent, and they
are train how to do so. Believe me, if the
patent requires symbolic references
in the instruction, then
it requires that. There is no ambiguity. Undoubtedly
the
patent was worded like that to avoid collision with prior
art that was
worded differently. If you go reinterpreting
that any way you like, you'll just
end up colliding with
what the claims were trying to avoid saying. [ Reply to This | Parent | # ]
|
| |
| Authored by: PolR on Tuesday, May 22 2012 @ 08:07 PM EDT |
A few things here.
If the notion of dynamic vs static is fatal to Oracle case, why does the jury
still asks questions? One fatal point is enough to decide the claim is not
infringed. I suspect the jury has bought Oracle's version, or at least they are
divided on this point and they look if they can get an agreement on something
else.
Next I don't think Google can argue the claim is poorly written and reinterpret
it. May be there is a legal procedure to challenge a patent on the grounds the
invention being claimed is not the one which has been invented, or may be there
isn't, I don't know. But it is clear this is not the defense Google has chosen.
Patent lawyers deliberately write broad claims to increase the protection given
to their client. This procedure is legal. An argument that the words should have
been different can't be used because the patent has been reviewed and granted as
written. The law is that the words of the claim control.
Then given the way the claim is written, Google defense is not overreaching.
They have chosen a reasonable interpretation of the words of the claim and argue
that this interpretation isn't infringed. If their interpretation is accepted,
they win. It is their prerogative to argue the claim means what they think it
means. In patent law, the words of the claim control.
Google's problem is that there is more than one possible construction for the
words of the claim regarding symbolic references. The jury is struggling to find
out which one they should use and they don't receive much help. This is not a
situation where there are two overreaching parties with a truth in the middle.
This is a situation where the words of the claim are not precise enough and
there is a range of reasonable interpretations, some favoring Oracle and some
favoring Google, and no clear place where we can say the truth is here and
nowhere else.
We may have a preference on which claim construction is correct and several
people on Groklaw have strong opinions about that. We may have strong technical
reasons to back these opinions. This doesn't make the other claim constructions
wrong. On the issue of symbolic references, there is ambiguity in the language
of the claim.
I agree that the aspect of dynamic execution is clear and should decide
infringement on the '104 patent. But it appears the jury doesn't see it that
way.
[ Reply to This | Parent | # ]
|
| |
| Authored by: Anonymous on Tuesday, May 22 2012 @ 08:57 PM EDT |
| I have objected, at length, to this argument from you several times over
the
past two weeks. I think this is the fourth time.
The patent says
"instruction containing a symbolic reference".
It doesn't say
"instruction that uses a symbolic reference" or "instruction
that references
something". There is only one way to interpret patent
claims, and that is
strictly and literally.
The patent claim means exactly what it says,
nothing else. It
explicitly says the symbolic reference must be contained
in the
instruction, and Google demonstrated clearly in court that that
isn't the
case for Dalvik.
(And I demonstrated clearly in comments on
Groklaw that it's not true for
a Java VM either, but that's a different story.) [ Reply to This | Parent | # ]
|
| |
| Authored by: Anonymous on Tuesday, May 22 2012 @ 10:23 PM EDT |
If, as part of executing the instruction, a string (or representation of a
string) has to be matched up with another string (or representation thereof)
then the instruction contains a symbolic reference. No amount of indirection
into various tables of strings, or optimisations of that search, remove the
symbolic nature.
That is, if the string has to be looked up within some class table, even if all
the strings are kept in separate tables by unique numbers and only the sorted
unique numbers associated with a given class are what is really matched, the
process is still that of resolving a symbolic reference.
[ Reply to This | Parent | # ]
|
| |
| Authored by: Anonymous on Tuesday, May 22 2012 @ 11:08 PM EDT |
> One honest answer to questions about what does the claim
> mean when it says "an instruction containing a symbolic
> reference" would be "a Java byte code instruction whose
> target is a reference to something that has not yet been
> through the symbol resolution process and so still
> ultimately refers to a symbol that has to be resolved
> before running the instruction".
If the USPTO had taken the words to not mean what they actually say, ie "an
instruction _containing_ a symbolic
reference" and had allowed it to mean "an instruction containing an
index or address" (which may point to symbol or another address, etc) then
they would have (or should have) thrown it out instantly because that is what
_all_ other digital computers do since they first were built.
[ Reply to This | Parent | # ]
|
| |
| Authored by: Anonymous on Wednesday, May 23 2012 @ 09:00 AM EDT |
But instead of that argument, Google overreached (I think a little
bit) by
saying that if the instruction doesn't literally contain a symbol then
it can't
be called an instruction that contains a symbolic
reference.
They say that about the Dalvik bytecode and they are
correct. By the time dexopt finishes, what had been a symbolic reference in the
Java bytecode has been translated/resolved and stored as a numerical offset in
the Dalvik bytecode so that when the Dalvik bytecode is 'run' by the DVM, the
DVM doesn't have to resolve anything and therefore '104 is not infringed.
JWC[ Reply to This | Parent | # ]
|
|
|
|
|
