Re: Is Apple's patent valid?

> From: Glenn Randers-Pehrson <glennrp@home.com>

> Chris Lilley wrote:
> >Thats not true.  The SVG patents page gives eleven companies who were
> >members of the SVG working group in good standing at the time we tested
> >out the PPF, towards the end of the SVG 1.0 timeframe. Of those, seven
> >gave RF licensing. Two gave RAND but identified no patents. Two gave
> >RAND and identified patents. Of those, one made a statement that the
> >patent was not essential and one (Apple) made no statement.
> >
> >Can you explain how this allows "most" to "renage" on RF licensing to
> >the world at large?
> 
> Actually my statement applies to "most" of the license statements that
> I am allowed to read.  Not having paid $15,000 for a 3-year associate
> membership (or even being allowed to, as an individual), I can only read
> those few that are included in the IPR document that's linked from the
> SVG spec.

Glenn, I am still finding difficulty with your definition of "most".
There are eleven companies listed on that page and there are eleven
license statements given in full on that same page. Scroll down a bit,
please. You have 100% of the license statements available.

Sorry if the links are confusing. They are there because they link to
the original emails sent in by those companies, so they (or another W3C
Member) can verify that what is listed on the public page is indeed
exactly what they said.

> Of these, Adobe and Canon both condition their offer of RF licensing to
> anyone on RF licensing by all other W3C members. 

I agree that the "all* is problematic and will seek clarification from
those two companies (I was already seeking clarification from Adobe, I
will ask Canon as well.

>  CSIRI conditions
> theirs on a reciprocal license from the implementor.

Yes. Not, in this case *all implementors* but a reciprocal, case by case
basis.

That is a good thing. It means that the cost of slinging around patent
license lawsuits goes up, which should discourage them. CSIRO says (as
indeed the PPF says) that the RF license is rendered null and void if 

a) someone sues them for patent infringement for SVG, or 
b) if someone only grants them a RAND license.

Yes, there is a cost and an increased vulnerability associated with
offereing a RAND license. If a company offers RAND, none of the RF
licensese from other companies apply. There might be a bunch of license
fees that they suddenly need to pay.

> Bitflash doesn't mention any extra conditions. 

They do. They mention the Patent Policy Framework. Which has the
reciprocal license condition.


> "Most" is two out of four (I had missed
> the short Bitflash statement before and thought it was two out of three).

Ok, if you are counting four, then that would seem to be the four that
offered RAND (in which case you would count Quark and not count CSIRO).

> >I make that distinction because, if Apple were to sue any of the
> >companies who gave a RF license, Apple would be breaking the RF license
> >contract with that company and those companies would be able to
> >counter-sue Apple for any patents that they might have. But Apple has
> >not done that so i don't see how anyone is enabled to renage on RF
> >licensing.
> 
> I don't see how a suit by Apple would be required. 

It seemed to be the only condition on which a previously RF license
cound cease to apply.

> It is a plain fact
> that Apple did not offer the RF license, which seems to be all that is
> needed by Adobe and Canon to retract their RF licensing.

I am agreeing with you that the Apple and Canon licensing seems to be
conditional on all members of the working group (Adobe) or all W3C
Members (Canon) abnd that this is a problem.

> >> Disclaimer: I am not a lawyer, and I have only read the abstract of
> >> Apple's patent
> 
> I am still not a lawyer, but I have now read the full text of Apple's
> patent.  Although their claims seem to be claiming alpha composition
> in general, 

Yes, claim 1 seems to be general and the other claims more specific, but
then the rest of it reads as if they are only talking about the specific
cases.

> they do mention prior art in alpha composition, 

Yes, which leads me to believe that claim one is conditional on the
other claims. But I am not a lawyer, either.

> and somewhere
> in the middle of the document it gives an example of using separate alpha
> masks for each color channel.
> 
> I think NetPBM's "pnmarith" could be considered prior art; at any rate
> it appears to be able to do such compositions in combination with pnmcomp,
> and channel-by-channel composition might be one of its most useful
> applications.

Cool. I assume that this is the Dec 1991 reease of NetPBM, so the code
dates from at the latest, 1 December 1991 which is five months before
the patent application on 8 May 1992. Since NetPBM was extremely widely
announced and deployed, and has freely available source, it is a good
prior art claim.

Could you post a command line that takes A.ppm, does channel by channel
composition using channels from B.ppm and composites the result over
C.ppm? Where A, B and C are any rgb ppm images.

> We should look for prior art in the prepress field in addition to the
> computer graphics field.

Yes.

-- 
Chris

Received on Thursday, 4 October 2001 10:31:43 UTC