To lwn: as I have continued to correspond with other Debian users on the subject of my first letter my views have gradually changed and become clearer. So if you decide to publish anything of my writing since the first letter, this letter would probably be the best one to publish (plus the note about the url of the English translation of the interview that got this whole discussion started).
On Sun, 28 May 2000, Tobias Peters wrote: > In the May 25 lwn, Alan W. Irwin writes: > > > I suspect this bad situation is a leftover from the old flame wars that > > used to erupt between GNOME and KDE supporters. > [...] > > What is ironic about the exclustion of KDE from Debian now, is that the > > Qt-1 library is actually officially supported by Debian! > [...] > > now that both Qt-2 and KDE-2 are coming out under free licenses. > > The fact that debian has an official Qt package should show you that the > absence of KDE is not a result of flamewars. > > Both licenses being free is not enough. The debian people actually read > the licenses and try to understand what they mean. They came to the > conclusion that a GPL'ed program must not be linked with a QPL'ed program; > the GPL forbids this. Now they are waiting for KDE to change their > licenses to "GPL + special exception: linking to Qt is allowed." You and others have brought up the possibility of KDE changing their license. I cannot speak for KDE, but I suspect they are reluctant to change from the GPL. Furthermore, even if every current and historical KDE developer were willing to modify the license of their work, you have to acknowledge it would be a logistical nightmare to organize such a change since so many have contributed to that project. So my judgement is this is a practical impossibility. (A similar practical impossibility is one of the protections against changing the kernel source from GPL.) You and others have also brought up the incompatible license argument. The official Debian argument on this issue is presented at http://www.debian.org/News/1998/19981008. However, if you analyze the argument *as put forth there* it seems to me to be in a grey area when dealing with a GPLed binary package like KDE which only depends in a dynamic way on the Qt shared library. In this situation, all the headers, etc., are distributed with Qt so the binary KDE has absolutely nothing in it that belongs to Qt. Now I have emphasized above "as put forth there" because it may be that the Debian argument is just plain badly presented. I have a lot of respect for Richard Stallman, and he has also made a specific judgement on this issue (scroll down in http://www.gnu.org/philosophy/license-list.html to the discussion of QPL) which absolutely supports the Debian position. However, I cannot find the reasoning behind his judgement, just his assertion of it. But given that RMS has come down on their side, I am willing to acknowledge the fact that it is unlikely Debian will change their position on distributing binary KDE unless RMS does. I can see why neither Debian nor KDE wants to change here, and it is also a practical impossibility for KDE. Unfortunately, that leads to the conclusion that Debian will never distribute the binary versions of either KDE-1 or KDE-2 except in the unlikely event that RMS changes his mind. This leads to your next point which I absolutely agree with. > > Nothing however prevents them from ditributing debianized KDE source > packages, and I must admit that I don't understand why they don't do this. > The user could compile (link) and use them due to his "fair use" option. > It would make an end to all these "They don't like KDE" arguments. They > once did the same with pine (which may only be distributed as source or > unmodified(!) binary). > I believe Debian source distribution of KDE is a way out of this mess. Source distribution does have the downside that the user must compile on their own machine, but modern PC's are getting so powerful that it only takes a minute or so to compile most packages. (Large collections of packages like KDE would take substantially longer, but this is simply a background task that Linux is well-suited to handle.) The upside of source distribution is the binary compiled specifically on your machine is more efficient and is guaranteed to have good library compatibility. Also, a security bonus is there is no way the binary on your machine can be tampered with if you have created it yourself from clean source. As well, much of the point of freedom software is access to the source. Thus, for all these reasons it makes sense for all distributions (not just Debian) to start emphasizing source more than they have in the past. One current problem for Debian (unlike rpm-based distributions) is it does not have a standard source-package format. My understanding is that the current Debian leadership has acknowledged this as a Debian weakness and is working on the problem. My advice to them is they couldn't go wrong by quickly adopting a standard debianized source tree format that is consistent with the debuild command. Once such a tree is set up, the debuild commands allows you to create the binary deb(s) with one easy command. Once a standard source package format was adopted for Debian, then I think KDE source distribution could be a wonderful poster-child of that standard format. Official adoption of the KDE source package by Debian would benefit all Debian/KDE users by the increased quality assurance that all official Debian packages enjoy simply from their interaction with the technically advanced Debian community. Official adoption also would benefit the Debian community by giving better interaction with the large pool of talent that has gone into the KDE effort. However, this scenario only works if a significant fraction of the Debian community welcomes source distribution as a way to support the KDE effort without violating any licenses. This leads to the concern expressed in my original letter about whether the Debian community is still prejudiced against KDE as a result of the old flame wars. I have hope on this issue because with the exception of one correspondent who made it clear that he was completely prejudiced against KDE in any form, the others seemed as reasonable as you on the issue of distributing KDE as source. Alan W. Irwin email: [EMAIL PROTECTED] phone: 250-727-2902 FAX: 250-721-7715 snail-mail: Dr. Alan W. Irwin Department of Physics and Astronomy, University of Victoria, P.O. Box 3055, Victoria, British Columbia, Canada, V8W 3P6 __________________________ Linux-powered astrophysics __________________________