On Mon, Jul 12, 2004 at 10:59:14AM +0100, Matthew Garrett wrote: > >I believe the more reliable DFSG#3 is to the point. It requires the > >right to make derived works--any derived works, not "derived works, > >as long as you're willing to give the initial developer a copy". > > I'm not sure I buy this. Surely the GPL #7 effectively restricts the > scope of derived works I can make and distribute to those that don't > infringe upon actively enforced patents? Isn't that a more onerous > restriction?
I acknowledge that the DFSG isn't so specific that these tests follow directly from it. It could be read so strongly that the BSD license fails (discriminates against the field of removing copyright notices!), or so loosely that almost anything passes. (That's one of the benefits of these tests; they help find where the DFSG is actually read.) There have been long (and mostly fruitless) debates about the interpretation of DFSG#3, of the form "requires *all* derived works" vs. "requires at least some derived works". Neither extreme is really correct, since judgement and interpretation really are required. In any case, I don't think I've seen any cases where it was actually a good idea to try to apply DFSG#6. While it could be used fairly sanely if a license actually said "can not be used in the field of nuclear weapons research", that's very rare. Most discrimination is a side-effect of other restrictions, and many restrictions can be interpreted as discrimination. The GPL carefully and deliberately discriminates against the field of proprietary software development, even though it does so in a roundabout way. (Being involved in that field from time to time myself, I can vouch for its effectiveness, having had to reimplement a few wheels as a result.) > >If not, then it seems like it's a poorly-executed version of the GPL's > >"apply restrictions to linked works by treating the result as a combined > >work", and DFSG#3 also applies. (Taken at face value, it's also license > >contamination, DFSG#9.) > > With the exception of the desert island and dissident scenarios, are > there any cases where the effects of QPL 6 are worse than those of the > GPL's requirements? That's hard to answer, since those scenarios are specifically designed to check the effects of clauses like QPL 6. If you mean "what real- world cases exist that the extreme test cases of the DI and CD test?", there are plenty, mostly in the "private modifications" category. See Message-ID: <[EMAIL PROTECTED]> (http://lists.debian.org/debian-legal/2004/07/msg00174.html). -- Glenn Maynard
