On Fri, 11 Feb 2005 13:57:47 +0100, Josef Dalcolmo wrote: > You can distribute GPL'ed code in binary form, you just have to make the > sources available as well. And, yes I would use this as a test: if your > program needs gpl-ed code for some of it's functionality, you have to > licence your program according to the GPL - unless you distribute the > GPL'ed parts separately and your program is still basically functioning > without the GPL'ed code.
The problem with this is what I've called the "patch hole" in another context [1]. The problem with this definition is that I can *always* distribute GPL'ed parts separately and re-combine them arbitrarily upon execution, and it's not even particularly hard. Write your code with the GPL'ed code embedded. At the end, before you distribute, extract it and record the extraction so your program can "rewind it"; you're left with nothing in your code that is GPLed. Later, the user will go get the GPL software, and you software "rewinds" the extraction process, and the user is left with something that is byte-for-byte identical to what you weren't allowed to distribute by the GPL.... so what good was the GPL? (Compiling issues can of course be extracted away, which is what a linker does.) If this is all the protection that the GPL provides, than it is utterly useless. But truly nailing down what it means is even harder. Nobody really knows what the GPL means when it gets down to it; the entire copyright-based model is broken and unrepairable in a software context. It's like nailing jello to a wall, you just can't hold it up there. [1]:http://www.jerf.org/writings/communicationEthics/node10.html#SECTION000105000000000000000 -- http://mail.python.org/mailman/listinfo/python-list