On Thu, 2 Nov 2006 11:55:23 +0100 A Mennucc wrote: > severity 396646 normal > retitle 396646 clarify and tidy up copyrights > thanks
I disagree with this severity downgrade. A file lacking permission to modify and to redistribute is a serious bug, IMHO. > > ciao Ciao! > > On Wed, Nov 01, 2006 at 11:23:43PM +0100, Francesco Poli wrote: > > There seems to be still a licensing issue, though... :-( > > According to its debian/copyright[1], mplayer includes a file named > > mmx.h > ..... > > Where's the permission to redistribute (DFSG#1)? > > Where's the permission to modify (DFSG#3)? > > that file is part of a library that was made public; That is to say? A library that was published? This does *not* necessarily mean that there's any permission for the general public to redistribute and/or modify. > unfortunately > all the original web pages (as found by Google) > http://min.ecn.purdue.edu/~rfisher/Career/vita.html > http://shay.ecn.purdue.edu/~swar/libMMX/Index.html > http://eetpc20.bd.psu.edu/~rfisher/ > are currently unavailable; anyway you can get info from the Google > cache. > > you may also download the library from > http://fresh.t-systems-sfr.com/linux/src/libmmx-990416.tgz > I read it all around: it is true that there is no place where the > authors explicitely state "you can redistribute this code at wish"; That is the issue I pointed out, essentially. There's no explicit permission to redistribute and/or modify. With the current copyright laws, this means that there's no such permission at all... :-( > but at the same time the authors > -) put the .tar.gz in their web pages This does not imply any permission to redistribute, let alone to modify. > -) wrote a public-domain-like statement in the code Could you quote it in its entirety? > -) wrote 'Copying-policy: PD' inside libmmx.lsm Which, unfortunately, is essentially meaningless: PD could be the acronym for anything[1], and public domain has anyway no clear and unambiguous meaning across jurisdictions. [1] Prohibited Distribution? ;-) > -) go to great length in explaining how to use their code in other > applications I really do not think that they intended to give > permission to use the code as public domain, but not permission to > (re)distribute it. They could be happy to distribute their code to anyone, and allowing him/her to link his/her programs or libraries against it. At the same time they could want to forbid redistribution of and modifications to their code. I don't think that publishing a library with good documentation necessarily implies that the library is published in a DFSG-free manner. > > Moreover, according to > http://www.google.com/codesearch?hl=en&lr=&q=Dietz+Fisher&btnG=Search > code from that library is also part of > xine, VLC, avidemux, mythtv, ffmpeg, gstreamer; > since it is not considered a problem in any of those, then it must be > OK in mplayer too. Err, it's the other way around: since it seems to be a problem for mplayer, it can be a problem for xine, VLC, avidemux, mythtv, ffmpeg, gstreamer, as well. This should even more encourage you to solve the issue, as it would be beneficial to other packages too. Question: why does not xine's debian/copyright (at least in the sarge version) say anything about mmx.h? Is this an overlook by xine package maintainer(s)? Or was mmx.h purged from the debianized xine? > > > BTW, there's another issue: the debian/copyright[1] file states: > > > > | Name: GSM 06.10 library > ..... > > The term "use" is vague and could be interpreted in a strict sense > > as meaning only "to run" or "to execute" the library. > > This license should be clarified: persuading upstream to change > > the first line in > > > > ] Any reproduction and use of this software, with or without > > modification, ] is permitted provided that this notice is > > > > would suffice to make it unambiguously DFSG-free. > > I will do that. Good. > > ---- > > ps: > when I wrote debian/copyright, I was overzealous: I included > copyright statement of any bit of code and header around; I think you did the right thing, as mandated by Debian Policy (version 3.7.2.2, section 4.5), which essentially states that debian/copyright must be "a verbatim copy of [the package] copyright and distribution license". The entire license, not the main one only! Hence, when there are different licenses that cover different parts of the package, all of them must be quoted verbatim in the debian/copyright file... > but I then found this reference > http://games.slashdot.org/comments.pl?sid=199749&cid=16359141 > that I summarize here as [...] > So it seems that small bits of header code is not copyrightable. It could be uncopyrighted: since I didn't examined the file mmx.h, I cannot comment myself on its copyrightability. But, take into account that, in order to be on the safe side, we must be very cautious before declaring that something is uncopyrighted (unless it's pretty clear that it actually is). As far as the typical content of a header file is concerned, there are several different legal theories, and not many court cases to dig... The FSF official theory seems to assume that they usually indeed are covered by copyright (the linking restrictions of the GNU GPL license are based on this assumption); others disagree. But anyway, as I said, to be on the safe side, you must assume that those files are (generally) copyrighted. > (I do not know if this would apply to mmx.h , though : that contains > some real function code) In that case, it seems even more clearly copyrighted... -- But it is also tradition that times *must* and always do change, my friend. -- from _Coming to America_ ..................................................... Francesco Poli . GnuPG key fpr == C979 F34B 27CE 5CD8 DC12 31B5 78F4 279B DD6D FCF4
pgpueoBAbxdgu.pgp
Description: PGP signature
