On Thursday 27 April 2006 11:19 pm, Andrew Lentvorski wrote: > Tracy R Reed wrote: > > Andrew Lentvorski wrote: > >> Correct. Every VLSI design group I know of has a blanket ban on using > >> anything GPL since it might later cause source code for a design tool > >> to have to be released. > > > > This is rather scary. Shouldn't you be checking the license of any code > > you use to see if you are allowed to use it, GPL or not? > > Ummm, yes. That is actually the whole point. If its BSD, I can use it. > If its GPL, I can use it *but only if I never distribute the tool*. > > In VLSI, sometimes you need to distribute the tool. Maybe to a > subcontractor, maybe to a foundry, but it passes to the outside. > > Once that happens, the GPL takes force. This has nothing to do with > "accidental" inclusions. This has to do with boxing yourself into a > corner which you can't get out of or which will require lots of work. > > The easiest way to avoid getting hounded is simply ban GPL code for your > software development environment. > > > Generally the GPL folks just say stop using our > > code or open your code. > > And what happens if a GPL library is the cornerstone of that code? The > best example of this is Computational Geometry libraries. They are > complex pieces of code which would benefit from being open since the > user community is so small and bugs are so subtle. However, all of them > are GPL or commercial-hostile licenses. If I am a company, I will > refuse to allow the programmers to use *any* of this code since it will > become the cornerstone of the code. Thus, the company generates another > buggy, incompatible computational geometry library and then puts a > restrictive license on it because they spent money to develop it. > Lather. Rinse. Repeat. > > GPL works well for code which has a potentially large starting user > base. BSD works much better if the user base is small and needs to be > nurtured. If Berkeley had adopted the GPL for SPICE2, the small user > base would never have coalesced into a coherent whole. It would have > retarded the growth of VLSI for years. The fact that SPICE2 was a BSD > license meant that an entire industry of circuit simulators appeared > almost overnight. Berkeley may not have gotten much for it (and I > disagree, but that's for another day), but the entire economy and an > entire industry benefited to a huge degree. > > -a
Good arguments and useful distinctions. But I have one quibble. You wrote: >Thus, the company generates another > buggy, incompatible computational geometry library > and then puts a restrictive license on it because > they spent money to develop it. > Lather. Rinse. Repeat. This is your opportunity to push a BSD license and/or form a community around this cornerstone library. Indeed that community may be a big help in promoting this. For some reason (creeping, no galloping old age, senility?) I am reminded of an old physicist friend of mine Fred Wolf, who in the early 70's wanted to see the world. Fred was at that time a young theorist with some ideas about particle theory and an Assistant Professor at SDSU. So he wrote (the old fashioned way, snail mail) to physicists in university departments all over the world. He said, "I will be in visiting near your university next year. If you are interested I would love to talk with you and your colleagues. Fred harvested several dozen letters inviting him to speak at universities all over the world. With these letters in hand he went to a government funding agency and said, "Look I am getting all of these invitations from all over but I don't have the funding to followup." Voila, a short proposal later Fred was funded and became a world traveler. I guess the moral is that all it takes is desire and some clever promotion ... BobLQ colleagues about Yada Yada ... " -- [email protected] http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
