Authored by: Anonymous on Sunday, May 13 2012 @ 04:34 PM EDT |
It's not "anything other than", it's "a name other than".
This is an important distinction, as Android uses an
offset/index rather than a name.[ Reply to This | Parent | # ]
|
|
Authored by: bugstomper on Sunday, May 13 2012 @ 04:53 PM EDT |
The patent itself does not make it clear that a "symbolic reference"
in an instruction could be anything other than a name. The example in figure 8
shows LOAD "y" being changed to an instruction with a numeric
reference LOAD 2 where '2' is the memory address of some data.
Most likely Gosling intended the LOAD "y" figure to stand for any
instruction whose data operand points to a data structure that contains a string
"y" that is resolved by looking up the string "y" in a
separate symbol table. By that interpretation, which is completely reasonable,
indirection in not an important characteristic. What counts is whether the data
being referred to is a name that has to be resolved by looking it up in a
separate symbol table. The purpose of the patented operation is to delay symbol
resolution until it is needed at runtime then cache the results for subsequent
uses.
On the other hand, the claims do not make it clear that indirect reference is
being claimed, and there is prior art in Lisp when you do include indirect
reference.
[ Reply to This | Parent | # ]
|
- Nonsense - Authored by: Anonymous on Monday, May 14 2012 @ 10:05 AM EDT
- Nonsense ? - Authored by: Anonymous on Monday, May 14 2012 @ 01:01 PM EDT
|
Authored by: jbb on Sunday, May 13 2012 @ 05:46 PM EDT |
The only definition I could find in the article was:
symbolic
reference: The term “symbolic reference” means “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.”
That seems
reasonable to me. Can you point me to the definition you are referring
to?
--- Our job is to remind ourselves that there are more
contexts
than the one we’re in now — the one that we think is reality.
-- Alan Kay [ Reply to This | Parent | # ]
|
|
Authored by: Ian Al on Monday, May 14 2012 @ 02:33 AM EDT |
Just a reminder that memory address indirection is not a software introduction.
It was built into the instruction set of early eight bit microprocessors (not
the 8080 or Z80 AFAIK).
If there is any suggestion that the patent covers indirect indexing in any way,
then this is completely wrong.
---
Regards
Ian Al
Software Patents: It's the disclosed functions in the patent, stupid![ Reply to This | Parent | # ]
|
|
Authored by: Anonymous on Monday, May 14 2012 @ 10:17 AM EDT |
I believe the lawyers mangled the invention being patented back in 1992 while
they were applying for the original patent that eventually got reissued as the
'104 patent.
Not only does it not cover the Dalvik VM, I don't think it even
covers a Java VM! The patent is explicit that the symbolic references must be
contained IN the instructions. No Java VM actually has that feature.
My argument for this claim is here.
...
By the way, if I said "Java
bytecodes" by itself anywhere, I probably meant "Java bytecode instructions".
That's how you should read the patent testimony e.g. from Andrew McFadden, too.
In a comment on the previous story, jbb pointed out that in conventional usage,
"Java bytecode" is sometimes taken to mean "everything in a compiled Java .class
file". But this conventional usage is wrong. When someone like McFadden says
"look at these bytecodes", he's talking about the instructions only (not other
things such as constant pool, or symbol tables, or anything in a data segment,
etc.) VM programmers usually use the word "bytecode" to mean only the bytes of
the instruction(s). [ Reply to This | Parent | # ]
|
|