On Tue, Nov 02, 2004 at 06:30:42PM -0500, John Cowan wrote: > > but I can distribute it under the X11 > > license, so the person I send it to is no longer forced to include source > > (or to grant patent licenses, and so on). > > Just so. AFL original and derivative works may be redistributed under > any license ranging from X11/MIT to a proprietary license. The AFL is > intended to belong to the BSD/MIT/Apache family of licenses, placing > almost no restrictions or obligations on the licensee, except: > > 1) Mark your changes and don't delete other peoples' change marks. > 2) Don't sue me for liability, warranty, or patent infringement.
Watch out: it only says "if you sue me for patent infringement, you lose your license", not "you can't sue me for patent infringement". (There's no way to do the latter, of course.) This means that only people who care about using your software are discouraged from suing you--people who are just out to prevent competition or extort license fees (who are the real problem, in my opinion) don't care. (I think you understand this already--I'm pointing it out since it's important that people using these clauses understand their limitations.) > 3) If you do sue me for something else, sue me where I am. General (but not unanimous) feeling on d-legal is that choice of venue is not free. (This is another topic that would subvert the thread, though.) > And even these modest restrictions apply only to the original work, > not to any derivative works. So what's the point? Why spend the cost of a longer license (and all of the related costs, such as mailing list threads as people struggle to comprehend the license :) to say "don't remove the change list", if you can still do so if you simply distribute the result under a different license? It seems like a lot of the stuff in this license is essentially optional--it says "please do this, but if you don't want to, feel free to use a different license"--they're just requests, made in an extremely complicated way. Requests don't belong in licenses. > > For what it's worth, I think the general feeling here is that licenses that > > need to do things which require a contract to be formed, and not a simple > > copyright license grant, are "probably non-free". It also brings in a > > bunch of bags of worms: "consideration", for example, and to my (very > > poor) understanding, contract law differs across jurisdictions far more > > widely than copyright law. > > Quite right so far as it goes, but contracts have advantages, and are > used by most corporate licenses (the MPL is a contract, for example). Most corporations do this to place heavily onerous restrictions on users that have nothing to do with copyright law, which requires them to use contracts. I don't believe this ability to restrict people in more ways is an "advantage", from the perspective of free software. > The trouble with bare licenses is that they may be revocable -- nobody > knows, but certainly a bare license to trespass on land, for example, > is revocable at will. The AFL is a unilateral contract: *only* the > licensor makes promises. Most people making this claim are saying that they're revokable because of lack of consideration, which implies the person is confusing copyright licenses with contracts. The only hint that it might have some validity is that I havn't been able to find any statement from the FSF or their councel (eg. Eben Moglen) saying that it isn't true, and some googling does show people claiming that they've asked about it and received no response--glaring silence is very suspicious. However, it seems that a similar possible problem exists with contracts: that they're not binding without consideration. I've read some text from people claiming to be lawyers explaining that things like "agreeing to the warranty disclaimer" and so just don't cut it as consideration. (I don't claim that there's real credibility in that, since I havn't been able to track them down. If anyone has any credible and human-understandable explanations of what generally qualifies as compensation--which is quite possibly different in different parts of the world, or even different states--please link.) It seems to me that the "free software contracts aren't enforcable due to lack of consideration" claims are more likely than the nebulous "we're not sure, but some people seem to think copyright licenses are revokable" claims (at least the contract concern is based in a specific legal principle), making me skeptical of using the former to solve the latter. > As I said above, the AFL is intended to be a maximally permissive license > that still provides some modest protections for both licensor and licensee. > The licensor is protected from certain kinds of lawsuits; the licensee > is assured of actually getting the original work (DFSG #2) and that > the licensor has the right to license it, as well as the full MIT-style > set of freedoms (note that by one reading, at least, the MIT contains > an implicit patent license in the verb "use", which is not relevant to > copyrights but is relevant to patents). But the licensor isn't assured of that, since the "licensor" isn't actually forced to give it to him--since he can just use a different license. It seems like this license says "you (the licensor) may bind yourself to give the source, but you don't have to", and I'm not seeing the point. In terms of freeness, if Debian is allowed to simply remove the AFL, distribute the work under the X11 license and pretend the AFL doesn't exist, then it's clearly free, though this implementation is so unusual among free software licenses that I wouldn't assume that an author placing his work under the AFL agrees with that--in other words, I'd look for confirmation from the author that this is what he intended by placing his work under it. -- Glenn Maynard
