> On 2004-08-09 13:35:30 +0100 Freek Dijkstra wrote: > >> As an end-user, it's far easier to just compile it all myself >> [...] then to change the code of netatalk to have it link to >> gnutls. > > Fine. If you choose not to help others, that's your choice. I don't > like it.
I could have just compiled all myself, and stopped there. That solved my problem. I instead choose to spent some time and change it for those who only use packages and do not have the time to find the problem and compile software on their own. My apologies if I sound blunt because I did not commit myself to making changes I do not see the need for. I may do it after all, now that Matthew suggested that it doesn't require much changes, but I personally belief OpenSSL is just fine. To keep on-topic: Netatalk is no derivate work. However, it is not clear if the compiled program, which is linked with but the compiled program is. Technically, for DHX (password encryption support), netatalk does compile against openssl by linking against the header files found in /usr/include/openssl (it adds -I/usr/include/openssl to CFLAGS) Aparently, in simular cases (with Courier in the 2002 thread), this was taken as incompatible with this note on the GNU FAQ: However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable. I was, and still am surprised (and yes, also disappointed) by this. > We want copyright permission from the copyright holders for this act > not covered by their licences in combination. Either openssl or > netatalk could give an exception. They haven't give it, for whatever > reasons. Regardless of what we think of their reasons, we don't have > permission... OpenSSL does give that permission: http://www.openssl.org/support/faq.html#LEGAL2 (last paragraph of answer) Netatalk is willing to give it, but can not: It is practically unrealistic to ask every possible contributor (including samba and libiconv contributors) to make this exception to the GPL licence. http://sourceforge.net/tracker/index.php?func=detail&aid=890674&group_id=864 2&atid=358642 It is sad that despite all copyright holders are more then willing to co-operate, there still is something holding them back. Please attribute the fault to whoever you like, but not the copyright holders. > If you want to argue that copyleft is wrong for software, > this is not the list for that and I'm not sure what is. Maybe ox-en or > gnu-misc, but maybe not. As for suggestion that GPL is wrong. Well, that it too harsh, but you are right. I should take this part of the discussion to gnu-misc. For the record, I don't think copyleft is wrong. On the contrary. However, I DO argue that licences which limit people in using (other) open source software are not optimal. If that is indeed what GPL does, I dislike it, and prefer other (better?) licences, like maybe BSD or Apache (APSL). Sorry if I sound too pessimistic here. Regards, Freek Dijkstra
