|
Authored by: bugstomper on Sunday, May 13 2012 @ 07:35 PM EDT |
The definition you quoted is the Court's claimed construction as specified in
the Order in filing #137. That Order goes into detail about Judge Alsup's
reasoning in coming up with that specific wording, including why he did not
accept Google's suggested phrase about "string or character-based".
Start on page 20.
Nothing is mentioned about indirect reference, either to say that a symbolic
reference can be indirect or to say that it can't. Judge Alsup would not have
known to bring up the question. Gosling may not have thought to mention it.
Perhaps Gosling thought that an acceptable simplification for the purpose of
illustration was to say LOAD "y" when of course an actual instruction
would have an operand that is a memory address of a string "y" that is
then used as a key for a lookup in a symbol table.
The other reference I used in researching this is the report from the
re-examiner which I found on the US PTO Public PAIR site
(http://portal.uspto.gov/external/portal/pair) searching for control number
90011490 and selecting the 47 page document dated in February.
Looking over that document again just now, I see that I made a mistake in how I
understood it. I thought that the discussion of how the re-examiner accepted the
prior art from the Gries text on compilers and rejected the prior art from the
Gabriel book on Lisp makes it clear that the examiner took the literal view of
"instructions containing one or more symbolic references", i.e., it
would not count if an instruction has an indirect reference to a symbol. But
that is not what he says.
Gabriel is not accepted as prior art because the symbolic reference resolution
is done after the intermediate form of the code is compiled to native code. Then
it does symbolic reference resolution, then uses "link smashing" to
change the reference address in a link table to be a numeric reference. That is
not prior art for the patent because the link smashing is not done during
interpretation of intermediate code.
Gries, is accepted s prior art only for the claims that say that the symbolic
reference is changed to a numeric reference. For the claims in the case, 11, 27,
29, 39, 40, and 41, only 27 and 29 are not invalidated by Gries. Those two
claims say that the instructions with symbolic reference are replaced with
instructions that contain numeric references. Gries modifies the targets of the
instructions without changing the instructions themselves.
The rejection of claims 11, 39, 40, and 41 as being anticipated by Gries makes
it clear that the re-examiner did accept that an instruction containing a
symbolic reference could have an indirect reference to a symbol. That would
allow the Java VM to be considered as practicing the claims, but still leaves
Dalvik not practicing them on the basis that Dalvik instructions indirectly
reference numerical references, not symbolic ones.
[ Reply to This | Parent | # ]
|
|
|
|
|