Path: utzoo!utgpu!jarvis.csri.toronto.edu!clyde.concordia.ca!uunet!snorkelwacker!apple!oracle!news From: wbailey@oracle.com (Bill Bailey) Newsgroups: gnu.misc.discuss Subject: Re: GPL yet again... Message-ID: <1990Jan4.164338.6516@oracle.com> Date: 4 Jan 90 16:43:38 GMT References: Sender: news@oracle.com Reply-To: wbailey@oracle.com Distribution: na Organization: Oracle Corp Lines: 47 In article , storkel@hijol.shell.com (S. Storkel) writes: > Suppose I write a program and it requires a version of getopt. What > happens if I wrap all of my object code into a big archive and > distribute it with a note saying "You must link the code with a > version of getopt before it will run. Try something like: ld ...". If > I distribute something in this fashion and Joe User happens to link in > Gnu Getopt it doesn't seem like I'm breaking the GPL. If you don't use any of the Gnu specific extensions to getopt() then I don't see how your code could fall under the GPL in the above scenario. That is, if the ATT version of getopt() could just as easily be linked in, then it is ludicrous to think that someone could suddenly find his code copylefted just because someone else decided to link his libraries with gnu code. The programmer who distributed his object libraries might know nothing about gnu, might never have seen any gnu code, might never have even heard of gnu in the first place. So how could his code fall under copyleft? If FSF'ers claim that it still falls under copyleft, then I assert that they are on very shakey legal ground which would never hold up in court. But on the other hand, isn't this a tremendous loophole in the GPL? Moving to the gcc and gnu C library issue, can't I as a programmer ship my object libraries with a Makefile for on-site linking and say "please link with any ansi C conforming C library." Again, I might never have even heard of gnu much less seen any gnu code so how can I be held accountable to copyleft if someone to whom I sell my object libraries happens to link them with the gnu C library? And, for the record, I agree with Storkel that forcing copyleft on people just because they link with gnu's C library is subversive to at least one facet of the gnu project. It will seriously detract from people's willingness to use gcc. Thus people will continue to use expensive, proprietary compilers. This won't advance the state of the art and it won't do away with people re-inventing the software wheel. This whole "linking with the gnu C library" issue is confusing to everyone. I just wish FSF would come out with a clear decision about using gcc, the gnu C library and copyleft which either states: a) using gcc and/or the gnu C library on your code means your code falls under copyleft. b) you may use gcc and the gnu C library without having to worry about copyleft. The ONLY way your code can become copylefted is if you DELIBERATELY copyleft it yourself. -bill