From:
<xua...@ba...> - 2008-01-29 04:55:01
|
Kal Ahmed wrote: > > Hi Xu=C3=A2n > > =20 > > The reason I originally chose Apache 1.0 was that I wanted a license > with the minimum of restrictions on who could use the code. I > specifically did not want GPL in there as that forces a particular > software distribution model on the licensee (i.e. they have to make > their source available). > I understand that. :-) > > =20 > > IANAL but I would have thought that a project under GPL should be able > to use a library licensed under Apache as long as the library source > was available=E2=80=A6no ? > Unfortunately, for practical matters, no. At least http://www.fsf.org/licensing/licenses/ says: Apache License, Version 1.0 <http://www.apache.org/licenses/LICENSE-1= =2E0> This is a simple, permissive non-copyleft free software license with an advertising clause. This creates practical problems <http://www.gnu.org/philosophy/bsd.html> like those of the original BSD license, including incompatibility with the GNU GPL.= Actually, there are at least 2 types of "use". 1. The first type is "write your own software such that it references interfaces provided by the other software". This is almost always legal, even if the other software is not open source. 2. The second type is "distribute your own software along with the other software". This is only possible if at least one of the licenses for each of the component projects allows this. And the GPL does not allow this bundling for its GPL component unless the GPL, or a compatible license, applies to the other software. And this is where the compatibility issue comes into play. > Also, I=E2=80=99m not sure if the effect of having multiple licenses wo= uld be > to reduce FUD or increase it. > It is quite common nowadays to to be dual-licensed or multi-licensed in general. See http://en.wikipedia.org/wiki/Dual_license . One prominent example is the Firefox browser, which is multi-licensed under MPL v1.1, GPL v2, LGPL v2.1. Type "about:license" in your browser to see more about this. ;-) I think, if a user of TM4J actually wants to join TM4J with some other open source software, having an explicit statement "You can use TM4J under license X" (where X is also the license of the other open source software) actually creates the inverse of FUD, i.e. confidence that this is allowed, legal and intended, without the need to consult license compatibility matrices, which may not even exist for that particular combination of licenses. The effect for the user is that he|she can choose among the licenses. The more licenses, the more permissive. The effect for the developer is that he|she has to agree not only to one license, but to all licenses. That typically means that attribution and all the other requirements are retained as long as the project is not forked off and the forker decides to choose a subset of the possible licenses for his|her fork. > > The one other thing that is worth thinking about with a license is > attribution =E2=80=93 I think that if others are using our stuff the on= ly > thing I would like to see is an acknowledgement of that =E2=80=93 that = was one > of the things Apache 1.0 provided. > Maybe the requirement for attribution is one "restriction" on the user which makes Apache 1.0 incompatible with GPL v2. However, GPL v3 allows for an attribution restriction.=20 > > My preference would be for a single, permissive and unrestrictive > license that says something like =E2=80=9CTake this stuff, do what you = want > with it. If it=E2=80=99s good think about giving it back. If it=E2=80=99= s bad don=E2=80=99t > sue us. If you use it, please acknowledge our work.=E2=80=9D > Then, although it retains the FUD problem of not explicitly permitting to use another license, like multi-licensing would, maybe we should use a 3-clause BSD license (which is single, permissive, unrestrictive, and well known to be compatible with copyleft licenses) and add the statement, separately from the core license but next to it, "If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.", like it is done in the bzip2 license. For example, Firefox respects such requests and has following notices in "about:license": Optional Notices Some permissive software licenses request but do not require an acknowledgement of the use of their software. We are very grateful to the following people and projects for their contributions to this product: * The zlib <http://www.zlib.net/> compression library (Jean-loup Gailly, Mark Adler and team) * The bzip2 <http://www.bzip.org/> compression library (Julian Seward) * The libpng <http://www.libpng.org/pub/png/> graphics library (Glenn Randers-Pehrson and team) Note that this is even more permissive than using multi-licensing. Because in the multi-licensing case, using TM4J in closed source projects (i.e. when using TM4J under the original TM4J license) strongly requires attribution without exception; attribution is optional only when using TM4J under one of the other free software licenses. So if you want to require attribution in general while making it optional only if TM4J is distributed as free software under a copyleft-license (i.e. where it is unlikely that the TM4J heritage is obscured as availability of the combined source code is required when attribution is optional), multi-licensing is a good and known way to achieve this. > Cheers > > =20 > > Kal > ciao, Xu=C3=A2n. :-) > > =20 > > *From:* Xu=C3=A2n Baldauf > [mailto:xua...@ba...] > *Sent:* 26 January 2008 16:02 > *To:* Kal Ahmed; tm4...@li... > *Subject:* TM4J license broadening > > =20 > > Hi Kal, > > I'm thinking that it would be good if TM4J could be licensed under > additional licenses besides the Apache 1.0 license. One reason is that > the Apache 1.0 license is GPLv2-incompatible and GPLv3-incompatible, > limiting the linkability of TM4J with other open source software > projects. That's why I'd like to propose that TM4J should be licensed > under following licenses: > > * the original TM4J license (which is an Apache 1.0 license) > * the Apache 2 license > * the GPL v2 license > * the GPL v3 license > * the LGPL v2.1 license > * the MPL v1.1 license > > as well as any later versions of these licenses. > > The rationale behind this long list of licenses is that to ensure > linkability with other open source software projects once and forever. > The "as well as any later versions of these licenses" is an intended > loophole to allow for adaption to future legal developments in the > spirit of the current licenses which we do not yet know (i.e. a GPLv4, > an Apache 3 license, ...) by the respective license standard setting > committees. This makes the explicit listing of some of these licenses > unnecessary (because some of these licenses are compatible with > others), but I'd like the minimum-license-list to be as explicit as > possible such that any doubt cast on license compatibility does not > create fear, uncertainty or doubt regarding license applicability. > > > To achieve this license broadening, consent is needed from nearly > every author, you being the most prominent one, that's why I start > with you. (As time goes by, contacting authors becomes more difficult, > that's why I better start now than later.) > So, what do you (and others) think about this proposal? > > > ciao, > Xu=C3=A2n. > > (Message was resent due to tm4...@li... > <mailto:tm4...@li...> not accepting or at least > withholding posts from spam-thwarting e-mail-addresses.) > |