In message <pan.2005.07.29.10.51.32.978746@pc313.imc.cas.cz>, Pavel Cahyna writ
es:
>On Fri, 29 Jul 2005 12:14:20 +0200, Marcin Jessa wrote:
>
>> On Fri, 29 Jul 2005 03:01:14 -0700
>> jnemeth@victoria.tc.ca (John Nemeth) wrote:
>> 
>>> On Dec 18,  6:53am, Marcin Jessa wrote:
>>> } 
>>> } I was also wondering if anyone is working on kernel implementation of CAR
>P.
>>> 
>>>      Personally, I think we should do something like VRRP, which is
>>> standards compliant, instead of using NIH stuff that will have limited
>>> compatibility.
>> 
>> CARP is free implementation of VRRP which CISCO claims rights to.
>
>How can it be free, then? Patents are granted for ideas, not for
>implementations.
>
Reading a patent takes training; it's not simply "the idea" that's 
patented.  A patent contains, among other things, a series of claims.  
Each claim contains a series of elements.  In order for something to 
infringe, it has to contain *all* of the elements of any one claim.  Of 
course, a good lawyer tries to write claims that cover all interesting 
variations of the idea, but it's not always easy.  Suppose you designed 
and patented a new cryptosystem.  You had about twenty claims, each of 
which described the encryptor and decryptor.  Someone who built only 
encryptors but not decryptors wouldn't infringe -- their devices would 
not have all of the elements of any single claim.  (That is not a 
contrived example.  If you look at the old (expired) RSA patent, you'll 
see that (if I recall correctly) all but one of the claims does mention 
both ends.)

		--Steven M. Bellovin, http://www.cs.columbia.edu/~smb