On 28/08/11 01:31, Russ Allbery wrote: > Ximin Luo <infinity0@gmx.com> writes: > >> The cost of the initial setup means that people don't do this in >> practise. I have not come across a single debian/copyright file with the >> full MPL text in it. Having a unified shared package for licenses makes >> this a lot easier. > > I think this has more to do with the fact that most work covered by the > MPL is covered by multiple licenses, there are questions as to whether the > MPL is even a free software license, and the maintainers of some such > packages have stated that they don't consider the MPL to be a license in > effect for Debian's purposes. I believe that is, for example, the reason > why the MPL is not included in the iceweasel debian/copyright file. (In > fact, I suspect most of the reference rather than inclusion of the MPL is > because people are just copying the practice of the iceweasel package > without giving it much thought.) > Thanks for the explanation; I didn't find this in the places I looked. Still, the fact that "the right way" takes much more effort than the multitude of "wrong" ways is not a good thing. >> This is flawed logic. Because there is no standard for this, and because >> this is a side issue from actual package work, every maintainer needs to >> work out this solution for themselves, and does things in a slightly >> different way. > > I agree that the divergence and re-solving of the same problem is > unfortunate, but I don't think that's a very strong argument for inclusion > in common-licenses. It's a good argument for improving the documentation > and resources available to packagers, such as wider dissemination of the > DEP-5 copyright format. > Well at the moment only humans can edit this format. Making humans pass around huge swathes of text isn't a great idea, practically speaking. I'm curious why you guys didn't decide to e.g. specify that License blocks simply point to a file that contains, and only contains, the verbatim license (which the source package would usually have.) >> You missed my point. Verbatim text in copyright may be mechanically >> extractable, but not easily verifiable. It's hard in the general case to >> verify that a license block called "MPL" actually contains the full >> correct MPL text, both for machines and humans. > > It's trivial to reverse the quoting for DEP-5 format. It's somewhat > harder if the license was reformatted in the process of quoting it. But, > regardless, writing code to verify that the maintainer did not err in > assembling the debian/copyright file is not something that I've heard > people previously want to do. Are you aware of some initiative here that > Policy work would make easier? > Rewrapping can be quite tempting for short blocks. Also, some people have the preference of removing line-breaks to have one paragraph per line (e.g. /usr/share/doc/openjdk-6-jre/copyright). Also, when you do something like, copy-and-paste a GPL2 block then changing the words to fit LGPL2, it's easy to make mistakes. I just see this as extra cruft and redunancy that doesn't benefit anything. There's no loss to just pointing readers to a shared system file, whereas you would gain easier maintenance of the file, better readability, etc. If you were to write a program that could report the copyright status of every single file on the system, it would be weird if you showed a slightly different GPL3 for different files. Even if you parsed a license text to a canonical form, I doubt this would be a visually pleasing form, or even one that has a coherent logical structure. e.g. Steve suggested collapsing whitespace - but this loses (e.g.) paragraph information. Anyone who actually wanted to write such a program, would have it present a standard GPL3 when it detects that a file is under GPL3. That seems the most natural way to do it. > I want to be sure we're solving problems that we actually have, not > problems that we anticipate that we could have but that have not yet been > encountered. > > The MPL 1.1, to be clear, does continue to have one of the better claims > of any license that's been put forward for inclusion in common-licenses > recently. But I'm still finding Ian's argument at: > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=487201#64 > > fairly persuasive. > Realistically, I don't think anyone is going to choose the MPL simply because they see it in /usr/share/common-licenses. Also, it seems the anti-MPL stance is just from a few people, rather than the Debian project as a whole, e.g. [1] says it's fine. It would be equally wrong to omit the MPL for the same reason. [1] http://wiki.debian.org/DFSGLicenses#Mozilla_Public_License_.28MPL.29 -- GPG: 4096R/5FBBDBCE https://github.com/infinity0 https://bitbucket.org/infinity0 https://launchpad.net/~infinity0
Attachment:
signature.asc
Description: OpenPGP digital signature