dld@F.GP.CS.CMU.EDU (David Detlefs) (12/21/89)
There has been incessant discussion in this group of the form: NON-FSF-CONVERT) If I use GNU library code such as getopt or regexp code in my program, it falls under the GPL. This is wrong! FSF-CONVERT) No, it's good! Besides, if you don't like it, just don't use it! NON-FSF-CONVERT) That's bad! FSF-CONVERT) That's good! NON-FSF-CONVERT) Bad! FSF-CONVERT) Good! ... I'd like to phrase this question differently: I don't care whether it's bad or good, morally speaking; rather, is it *effective*? If we view the short-term goal of the FSF as the following: To advance the state of the computer art by targetting specific well-understood, generally useful utilities, such as editors or compilers, and producing very high quality no-cost-to-user versions. This should drive commercial competitors out of these markets unless they offer significant technical innovation. They will hopefully move on to building new tools that are not yet so well-understood -- i.e., actually advancing the state of the art. Now, I don't claim that this is the actually the FSF's (i.e., RMS's) goal -- it is too pragmatic, in contrast to the GNU Manifesto's overtly morality-based arguments. However, I would claim that it is what most people who contribute to the FSF perceive as its goal. If we accept the above statement as the FSF's goal, does requiring the use of library code to draw the using work under the GPL enhance the chances of reaching it or not? I think it hurts; it may keep commercial compiler vendors in business longer than necessary. I give seminars on C++ at industrial sites; when I discuss the available C++ compilers I always say good things about G++, but I then have to warn about the legal ramifications of using libg++, and the difficulty of finding independent, unencumbered libraries you can buy. Gcc/G++ with unentangling libraries for free would be pretty tough competition for compiler vendors. If they went out of the compiler business, their employees could get on to working on something new and difficult. For what it's worth, -- Dave Detlefs Any correlation between my employer's opinion Carnegie-Mellon CS and my own is statistical rather than causal, dld@cs.cmu.edu except in those cases where I have helped to form my employer's opinion. (Null disclaimer.)