> I'm not sure if this is a valid analogy, but this reminds me of how > (I've been told) IBM PCs in the 80s were reverse-engineered by IBM-clone > makers. They'd have someone look at some BIOS microcode or something, > and write a spec from it.
IANAL The network protocol equivalent (and what someone told me the SAMBA team did), to have a "clean room" implementation is to have one person do a tcpdump of the wire protocols, figure out the transactions bits/bytes, write it down as a specification, and someone else write an equivalent implementation. Test, and repeat as needed to get everything working. You also have to document the process to be able to demonstrate later that you did not use insider information. For OpenAFS, almost all of the interested parties are already "contaminated" by having seen the IBM code itself, so would not likely meet a "clean room" reimplementation test. And, in any case, patents (in the US at least), can still encumber independent implementations (it is the invention that counts for patents). Perhaps "In re Bilski" will change all that for software, but that is still a pending case. IBM learned that lesson for PCs with the PS/2 MCA bus (patents on everything, and (just by accident) specified a card size which required fab capability to get enough stuff on the board (no more simple 74xxx logic gates on the card like you could do with the ISA cards), raising the barrier to entry for a few years). Gary _______________________________________________ AFS3-standardization mailing list [email protected] http://michigan-openafs-lists.central.org/mailman/listinfo/afs3-standardization
