Subj : Re: Implementing lock-free data structures To : comp.programming.threads From : Joe Seigh Date : Thu Mar 10 2005 05:11 pm On Thu, 10 Mar 2005 13:24:29 -0800, SenderX wrote: >> I don't feel like publishing the general rules at this point because it >> would basically be a guide for patenting all the lock-free implementations >> no one has gotten around to yet. Because the US patent office *will* >> issue patents for them. Sun already has a patent on lock-free ilnked >> list used with one form of GC. > > Yeah. I too am "sort of thinking" about protecting one of my proxy gc's ( > not published anywhere ) that can chew on extremely high amounts of > concurrent writes without running the risk of draining all the system > resources. All of my other collectors blow the stack in C++, or just plain > run out of memory in C when a couple of million writes rapidly and > repeatedly come in from 10 or so threads over extended periods of time. How > could I effectively protect my new collector without getting a patent? > Well, besides getting a patent, the only things you can do are publish or keep it secret. If you publish it, it can be patented in theory. If you keep it secret, it can still be patented by anyone else who thinks of it also and then you can't use your own idea after that. The problem I'm talking about is taking patents on further "improvements" upon your orginal idea. These are sometimes called blocking patents. There's potential for blocking patents here. -- Joe Seigh .