On Wed, Mar 06, 2002 at 11:38:42PM +0100, Ceki Gülcü wrote: > > Jeff, > > Kudos for having the courage to proceed with this. Comments inline.
:) It's not easy or fun. > At 08:46 07.03.2002 +1100, Jeff Turner wrote: ... > >Why prefer the ASL to a copyleft license (eg GPL)? > >-------------------------------------------------- > > > >This is an slightly distasteful topic for most Apache developers. The license > >is simply not a central part of the Apache philosophy. Apache is > >about creating communities that create great software. The ASL is a > >minimum legal necessity that allows us to do this, nothing more. It > >promotes no political axe-grinding, and has no great philosophy that > >needs defending. The ASL, in fact, presents such a small > >conversational target that any licensing debate inevitably becomes > >"what is wrong with license X". That inevitably leads to > >misunderstandings, holy wars and bad feeling, It's not productive, > >and not fun, and why we find licensing debates distasteful. > > The license is very much part of the Apache philosophy. It may even embody the > essence of the philosophy. The license says, basically, "do what you wan't, but don't sue us, don't abuse our name, and give credit where credit is due". That isn't much of a philosophy ;) It hints at the underlying importance we attach to the Apache name, but that's all. > No need to be apologetic about discussing licensing. Not apologetic, just reflecting a general lack of keenness for licensing debates, because they usually end up in unproductive GNU-bashing. > A good license is more valuable than a million lines of code. I maybe > exaggerating but only a little. > > >In particular, it's not fun rubbishing the GPL. The reader is > >encouraged to read the GNU's philosophy pages > >(http://www.gnu.org/philosophy/). It is wonderful, high-minded stuff > >that most programmers instantly resonate with. Opposing RMS's vision > >of Free Software at first seems to be like kicking a puppy. > > > >But let's kick it anyway. It turns out that the puppy soon grows up > >to be a bulldog, biting and tenaciously hanging on to any code it > >can. Due to the GPL's extensive scope and 'viral' linking rules, > >GPL'ed code cannot be incorporated into proprietary software. It must > >all be copylefted, or none of it can be. > > A bulldog? :-) Something with teeth :) But yes, bad analogy; will be removed. > >In many cases, we at Apache find the GPL's virality a hindrance in *our* goal: > > to (not in) *our* goal? agreed > >creating communities that create code. This is because large parts of our > > that write code? okay > >"community" are selling custom solutions, not shrink-wrapped products > >sold in volume for general consumption. Essentially, selling > >software-based services, not software. When you're selling a service, > >releasing the code makes no sense to *anyone*. The code is mostly > >customer- or sector-specific, so is not reusable, and of little > >interest to fellow developers. The customer *certainly* doesn't want > >you publicising their code, breaking confidentiality agreements and > >potentially exposing security flaws to the world. > > Hmm, are you sure we are only selling services? I dunno. I claimed that "large parts of our community" are selling services, not software. I don't know how true that is. I *suspect* it's true; that there are more consultants here than people banging out commercial code. I could be completely wrong. That's why it's so hard and dangerous to claim to speak for anyone but oneself. > Exposing security flaws to the world is very debatable. Most > cryptographers consider "security-by-obscurity" as bad practice. I > would drop the exposition argument. Yes that was very much in my mind :) The detractors of "security through obscurity" are usually talking about large commercial software. When you have custom code written in a hurry on a tight budget, security holes inevitably arise, and security through obscurity is better than nothing. Though your first impression is how most people will see it, so I agree it should be removed. > I found the ethics argument in the Reese-Stenberg article to be very > powerful. > > The original author has no *absolute* right on extensions and > improvements. The fact that I wrote 100 initial lines of code gives > me no right, moral, ethical or otherwise to impose a license on the > 10'000 lines that you subsequently write. I certainly have no rights > on 10'000 lines of *unrelated* code! Indeed! But arguments of morality are even more treacherous than arguments of pure pragmatism. GNU proponents would surely argue that the means justifies the end. The goal of Software Freedom warrants a bit of arm-twisting. > >Thus, to adopt a copyleft license like the GPL would alienate the > >service-oriented portion of our community. We want the widest > >possible audience, not for "market share", but because the diverse > >input results in software with "hybrid vigour", wide applicability > >and the kind of tough-as-nails quality we strive for. > > The service orientation again. We can't know the exact motivations of > developers for authoring open source code. Service-oriented software, > maybe but maybe not. The service-orientation argument is correct, > just not exhaustive. Right, and the text needs to make that clearer. I chose the "service orientation" argument (over the ethical argument, say) because it ties in directly with Apache's inclusive, community-oriented goals. I wanted to emphasise that the issue is "software Freedom" (GNU) vs. "community and quality" (Apache). > >Thus, we encourage users to adopt non-copyleft licenses like the ASL > >for "everyday" code, as it increases the chances of code sharing and > >cooperation, ultimately leading to better software. > > What is meant by "everyday" code? It's a cop-out to avoid the implication that we consider the ASL a 100% replacement for the GPL. I don't think that Linux, say, should abandon the GPL. I want to say, "the ASL is more appropriate for the stuff *we* do", nothing more. > >For further information, please refer to the well-researched and > >well-written O'Reilly article entitled "Working Without Copyleft", at > >http://www.oreillynet.com/pub/a/policy/2001/12/12/transition.html A > >good general reference of open source licenses is Bruce Perens' book > >"Open Sources: Voices from the Open Source Revolution" at > >http://www.oreilly.com/catalog/opensources/book/perens.html > > The "Working Without Copyleft" article is remarkably good. The point > about the FSF controlling the LGPL is another very significant point: > > The Free Software Foundation controls the license. They can release a > new version of the license, which then will automatically apply to our > software. Although we do not expect the Free Software Foundation of > making changes that deviate from the spirit of the current versions, > they could make clarifications that are contrary to our > intentions. For example, they may clarify that the result of > aspect-oriented weaving is subject to the terms of the LGPL, whereas > we had intended that it is not. Another concern is who will be in > charge of the Free Software Foundation 10 years from now, or what > happens if the Free Software Foundation is discontinued? [LGPL, > section 13] > > Licensing is a blackhole for time. It also happens to be central element in > open source development. Its importance cannot be overstated. Thanks for the constructive feedback, support and lack of flaming :) --Jeff > -- > Ceki Gülcü -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
