On Tue, 2002-04-09 at 06:00, alexander@alexanderbraun.de wrote:
though IANAL: is it possible to attack patent no. 5,265,261 due to terms of prior art? I only read the abstract but it seems to me it is a clear description of tcp/ip but with an initial date from 1993. and i don't see any difference to 5,437,013.
I think, but don't quote me ;), that we're actually only talking about one patent here - 5,437,013. 5,265,261 appears to be an earlier version, and at first sight Microsoft appear to have narrowed their claim from the previous version. There is also supposed to be an even earlier original claim, which has now been completely disclaimed - if anyone can find it, it would be interesting (I don't think the UKPTO passed it?).
What the patent appears to be is a claim on raw-mode network access, along the lines of libpcap or something - an efficient way of marshalling streams of data from applications and the network stack to the wire so that it actually makes sense. I would be astonished if there was no good prior art on this; the claim is from ~1993. Even if there is no prior art, I can't believe that it's even vaguely original.
So, it looks like that Samba doesn't infringe on this patent (unless it requires raw access for broadcasts or something?) as-is. But it also looks like this patent could be dangerous for applications beyond those that implement/use SMB - it looks like even tcpdump would infringe :(
Cheers,
Alex.