The dynamic requirement in '104 can also be interpreted
like
this:
- You can use the list without writing down the
actually busses, and just looking at them every time, like
an Alzheimer's
patient does.
- You will do this if you have not had a chance
to take notes, or failed to do so.
- The fact that you will
sometime do it, makes
the list without the extra notes dynamic
references
that don't point directly to the right place.
- Oracle got
the court to define "synbolic" to
mean just that. Google disagreed,
but Oracle won that one.
- Oracle's "silly benchmark" of running Dalvik
without
dexopt demonstrated that Dalvik will sometimes
process field
index references dynamically.
- The ability to have an instruction that
can be patched
up with a final offset or looked up dynamically at each
execution, and to have a tool that does that unnecessary but
practical
patching up is the essence of '104
That is Oracle's theory in a
nutshell. Not "patching up"
references (old tech), not "patching up in a
virtual machine
instruction set" (old tech), not "having a virtual machine
with instructions that need symbol table lookups for each
execution" (old
tech), but "having a virtual machine that
can do lookups on the fly,
and then patching up the
program not to do so anyway".
So the '104
fight is all about what the words in the '104
patent letter mean exactly and if
that meaning includes the
way dexopt does things or not.
So both sides
are playing word games about where the
property line is between them, does that
"old tree" on the
map refer to the oak tree over there or the redwood tree
over there. If it is the oak tree, then the property line
is a few inches
further west and Google did not step on
Oracle property. If it is the redwood
tree then the
property line is a few inches further east, and Google's
left
foot is on Oracle's soil, hence trespassing.
[ Reply to This | Parent | # ]
|