Slashdot Mirror
QT/GPL licensing trouble
Bitscape writes "LWN reports that Corel's gui packager, which uses both QT and lib-apt, fell under a licensing conflict which makes it illegal to distribute. While the author of lib-apt has agreed put an exception in its license for this case, is it a taste of things to come for people who use code released under the ever-expanding mess of incompatible licenses?
"
I think the project your referring to is the Harmony project which is the GNU effort to make a source compatible lib. IIRC, that effort was reinstated some months ago. It sounds very exciting. It's the only UI lib I know of that doesn't have problems with UI updates outside of the event thread. A multithreaded UI could be very cool, indeed.
--
http://cheeser.blog-city.com
Suprise suprise... alot of people here have been dismissing RMS for being "too radical" for saying the GPL is the only truly "free" license. Now you can appreciate, in it's full ugliness, why he's been advocating the distinction between free software and open source.
There is a difference, and you just read about one of them. The GPL or a BSD-style license would not have these issues. The fact that a special exception was required says that the authors are ameniable to change (they didn't have to allow this you know), but wasn't the whole point of this movement to prevent somebody from holding that over your head in the first place?
--
GPL, of course, is the standard license for most Open Source software, and for good reason - it does the best job of protecting the source from abuse by others. However, its "viral" nature makes it hard to cooperate with other licenses.
My biggest cooperation concern is MPL, the license for Mozilla's code base. I have great faith in the wisdom and foresight of the Mozilla developers, and i expect it to become the gold standard for browser engines. More importantly, its modular design makes it an easy plug-in for other projects needing a browser interface. And there's the rub... how readily can the Mozilla code be linked to a GPL'd program? I know both the Mozilla and Gnome people are working on the licensing issues, and i hope they'll come up with something that can be extended to use with most other GPL'd software as well.
The other license that worries me is SCSL, associated with Jini and future versions of Java. Again, i think Java is killer technology, and i want to see it grow and mature in cooperation with Linux. But until Java's APIs stabilize, the Open Source variants such as Kaffe (I don't consider SCSL "Open Source") will not be able to catch it in quality. And i want my Java NOW, dammit! So... what sort of licensing hell awaits GPL'd projects that want to use Sun's classes? Yikes!
With the addition of mature Mozilla and Java technology, along with improvements in Linux desktop/WM software, i believe Linux is poised to take the desktop world by storm. I just hope licensing conflicts don't delay it for years to come, and force endless redundant development in order to be compatible with GPL.
---
Maybe that's just the price you pay for the chains that you refuse.
Hand me that airplane glue and I'll tell you another story.
I think the situation has changed since this debate originally arose.
The QT and GPL licenses certainly are potentially in conflict, but whether they are actually in conflict depends on current practice.
IIRC, the GPL contains a clause which allows software to be linked with non-GPL packages if those packages are routinely supplied as part of the OS. Hence you can GPL a Windows application, and you can link applications with Xfree86 which is not GPL.
Now every major Linux distribution is shipping QT and KDE, doesn't QT now fall under this same exception?
If people on slashdot spent half as much time coding as they did worrying about what other people are doing in relation to GPL (et.al), this would be a moot point. It is one thing to extoll the virtues of true GPL; It is another thing entirely to piss and moan about what a few select individuals (companies) are doing. If you truely believe GPL creates a superior product, get them where it hurts; CREATE a better product. Period.
o ns, etc.
I'm aware I might get flamed for this, but so be it. I tire of the "me toos", RMS "rah rah" crowd, knee-jerk-reaction-crowd-to-geek/privacy-instrusi
Everything has tradeoffs.
If you use GPL, the commercial companies don't like it. They have a hard time figuring out how to sell it.
If you use BSD, the rip-off artists love you, but they don't share back.
If you use MPL, one knot-hole acts as a filter for all possible changes, and one knot-head can really set up impedance.
Personally, my preference is GPL. I understand why the commercial ventures don't like it, but some of their reasons for not liking it are the same as my reasons for liking it.
OTOH, Sun wrote almost all of Java, so it's properly their ball and court. Play there if you want to.
And Netscape wrote a complete starter for Mozilla. The team may have later decided to dump much of the code, but even then Netscape was paying for much the development work. So it's their ball and court.
Troll-tech seems to be trying to be accomodating, but this is also the software that they are earning their livlihoods from. They wrote it. They are trying to get it into wide use so that it will be used commercially (which is when they make money).
As a combination end-user/developer my preferences are for GPL, GTK, etc.
Licensing is is the essence of why I would rather use Python than Java, but it sure would be nice if Python had Java screen painters/class libraries/printing/IDE's (on the windows side, anyway). OTOH, Python has better database access and integration with C. But at least on windows Python (or possibly tkInter) crashes too often to use it, so there really isn't a choice.
I think we've pushed this "anyone can grow up to be president" thing too far.
I have to make something absolutely clear about GPL, because I'm sick of hearing this "viral license" nonsense.
It is wrong wrong wrong and extremely impolite to accuse GPL of being a viral license because an application that links against or is based on a GPL'd app must be GPL.
Someone was kind enough to write a piece of software and give it away. Why shouldn't the auther be able to make the reasonable request that all future uses of that software be equally free? I for one don't want the code I give away turned into commercial products without any benafit to me.
It's not "live and let live" when you base code on my work. That's my work too in there, and I deserve a say in how it's used.
And that's why you should use the LGPL for your next library, ladies and gentlemen.
(see "Why you shouldn't use the Library GPL for your next library" if you don't know what I'm talking about.)
Seriously, free software has seen the successes it has because it shows about the right amount of cooperation with proprietary software. You can't reach into the sources of a GPL application, change it, and sell it as a proprietary application - but you can connect to a GPL server with your proprietary client, or run your proprietary app on your GPL kernel.
As far as I'm concerned, the LGPL fixes a potential bug in the GPL - one way that two separate pieces of software can talk, dynamic library linking, might not be allowed, so we explicitly permit it. Yet RMS seems to think this bug should stay. As far as I can see, this artificial distinction will only make calls from proprietary software into free software (or vice versa) somewhat less convenient: you have to go through CORBA or some similar gateway.
I think that use of the LGPL (the "Liberal General Public License", as I now dub it in riposte) would most usefully increase - that all new programs should start using this license, since it seems increasingly that today's standalone program is tomorrow's library.
--
Xenu loves you!
RMS doesn't want people to stop using LGPL, period. As outlined by the link you provided, he only wants the LGPL to be used when its of strategic advantage (that is, it benefits free software more than it hurts it) to usher in the software in question into being a de facto standard across the board. the GNU C Library is a perfect example of this. Everybody has their own version of a C Library. glibc isn't anything special in this regard. LGPL it, however, and you'll have everyone using it because it's easier than brewing your own. It also ensures that more programs calling on that library will be compatible here and elsewhere without much modification.
What RMS doesn't want to see, however, is software that gives free software a definitive edge (that is, if it does something much better than existing software, or something completely different -- and naturally you'd want the OS to be GPL'ed, duh) being LGPL'ed (which is why I'm still wondering why the Berlin Consortium chose the LGPL.. the proprietary Unices -- and free ones -- currently have X. Berlin would give free Unices an edge if it were GPL'ed.. oh well), because that doesn't help free software.
~ Kish
I never said that all software should be free (either as beer or speech), and anyone who writes code has the right to choose how they share it with others. And i praised Mozilla for a smart corporate license.
The *point*, which you obviously missed, is that the merely semi-free licenses like Qt's are at a serious disadvantage in the "marketplace of ideas" when competing with GPL or BSD licenses. Developers are more likely to work with code they can trust... they can't trust Qt. Which is why, in the end, i expect to see Qt marginalized by GTK, not for GTK's technical superiority, but because the license terms are better for Free software developers. Free speech is better than free beer.
Remember, the ONLY reason TrollTech opened up the Qt license as much as they did was because the community was abandoning KDE for Gnome, almost entirely for licensing reasons. If Qt's license hadn't changed, i think KDE would be nearly irrelevant now. As it is, i think Qt's license didn't change enough, and it only delayed the problem.
Of course, if you think i'm just another fanatical RMS clone, then you won't be able to see beyond your prejudices to the deeper point.
---
Maybe that's just the price you pay for the chains that you refuse.
Hand me that airplane glue and I'll tell you another story.
What I find troubling is the ease at which a special exception was granted for Corel. I am hoping this has nothing to do with the relationships Corel has been developing with Debian, because if so this is totally wrong. The library being GPLed restricts it from being linked to almost anything, presumably for a reason. Now a special exemption is granted for just this one product that wants to link to it? This makes no sense at all. What happens down the road when twenty other special exemptions have to be granted, do we see the license extended by another length of the GPL? And this exemption is being granted only to a commercial entity.
In my opinion what the free source movement needs is the equivalent of cross-licensing, and no, GPL is not cross-licensing when it cannot freely be used in BSD licensed or X licensed code.
So a corporation like Corel can just call up Debian on the phone and instantly get an exemption while an unknown student would not be able to freely study the library code to re-use in his
projects throughout his latter career? I predict no one else on Slashdot will see how wrong this inequity is, and that is the problem.
As a example of how absurd the current situation is, there are maybe a half dozen incomplete attempts to reproduce Microsoft's Windows headers. No one seems to see the big picture that what is needed are one set of headers that are good enough so that serious projects would consider not having to shell out the licenses to buy Microsoft's development tools. No, entities such as Cygnus want to keep the code under whatever license is restrictive enough (GPL) so that if lightning strikes they can make a fortune by selective commercial licensing, so therefore we have perpetual beta products.
Perhaps the energies Perens (et. al) have not been entirely fruitless. But they're not hugely sucessful either. If you could harness all that slashdot discontent and direct it into focused coding, rather than pissing and moaning, a great deal more would be accomplished.
Which brings me to my second point. What does it matter if some little company in California creates a closed source product? Whose freedom does this impinge on? With the exception software patents, everyone else is essentially free to code as they please. The point being that the two can exist simultaneously. Yet RMS has been known to advocate prirating commercial software.
Futhermore, GPL software has, for the most part, been totally unfriendly to geeks. Not only in "userfriendliness" and GUIs, but also in terms of software functionality and purposes. In general, it is software that appeals to geeks and geeks alone. To advocate RMS's idea of freedom, is to say: The geeks' right to code free of non-free-software influences, exceeds the right of the average user to enjoy software that meets their needs. I disagree.
Both free and non-free software have certain unique advantages over one another. Not only can they coexist, but they're strengthened by one another -- they push one another to mature and expand in scope. So I come full circle. Let both do their best to succeed; let the cards fall where they may. As a matter of optimizing the results though, free software should worry about what is going on within its own community. Bolster and explain free software, but don't try to tear down anything that is not free.
I just got an email from Elizabeth O. Coolbaugh at LWN about this article, which has several falsehoods (you really can't link BSD or other licenses to GPL code, and it totally disregards the LGPL), and she said it is a misleading article and will be corrected. She also said it was "Obviously put together too quickly" ;) This is the reason Gnome for example uses almost entirely LGPL instead of GPL. The article took a rare exception to this. Using the GPL instead of LGPL for a library expressly means you want to deny linking to any non-GPL code (including BSD). ;-)
Daniel M. Duley
mosfet@jorsm.com Sorry, I don't have an account and Slashdot hasn't sent me a password yet
You know, a lot of people - especially slashdotters, like to flame me when I make a prediction, like Corel's gonna make a big screwup, or NetSol's gonna do something, or what have you.
Once again, I've been proven right though. Can't contest it. I said a few months back when this was originally announced that Corel's gonna have licensing issues, especially with Debian. Corel doesn't understand the DFSG (Debian Free Software Guidelines) and doesn't care so much about them.
This immediately puts Corel into conflict with a good many packages in Debian, pretty much. Told you so. *smirk*
What bothers me, however, is that while I'm not some GPL hardliner, or DFSG hardliner even - the author is compromising his license.
There's one license I won't tolerate - and that's a compromised one. The author owns the software, and how he chooses to license it is his decision. No author should be forced to bow to anyone because they cannot create a derivitive work with a compatible license. This is Corel's problem, and they should be working to rectify it properly - by writing software that can be licensed compatibly, not by basically forcing the author into compromising his license.
Look at it this way; should I put a new stereo in my car, just because big corporation XYZ says so? That's how I'm looking at this - you may view it differently, but IMNSHO, that's not how it is. Here, a maintainer is basically being told to compromise on his license because a corporation says so. I don't believe for a second that this is right. Even for the good of the distribution as a whole, it simply is not right.
If Corel's going to consider DFSG a hinderance that the maintainers and authors have to work around, then I think that Corel had best get off it's butt, find a new distribution that doesn't care about author's rights and the DFSG, and start from scratch. This is nothing short of bullying, almost Microsoft-style, to make sure they get their way so they can turn a profit.
Yes, it's going to increase Debian's user base - assuming Corel isn't found guilty in the Canadian's SEC-thing investigation - but it's now doing it at the cost of author's rights and freedoms. They've already had one licensing blunder - this makes two - how many more will there be? Eleven? Thirty? Fifty? For all we know, the entire distribution could end up being totally incompatible with the GPL or DFSG. Maybe Corel will change their license, or maybe they won't. Time will tell on that.
But at this point, it's my belief that the entirety of Corel's base - which is Debian - has now been compromised, as well as Debian as a distribution alone. Sure, Corel can throw in fancy things like WordPerfect 2000 and Corel DRAW! and other neat applications that would fall under 'non-free' at best. But they're doing it at an incredible cost. I really have a hard time accepting the compromise of a license as 'reasonable' in any situation whatsoever. If the author wants to license their software in a specific way, then that is their right to do so. They licensed it how they wanted it, they can change the license, but never should they be forced to change the license.
We already know Qt/Qtlib/KDE is non-DFSG compliant. Corel said 'we don't care.' Corel blatantly ignored the restrictions of the GPL and other licenses when they released the beta. Now they're basically forcing authors to change their licenses so they can attempt to save their finances.
Maybe Corel should have just gotten RedHat. At this rate, their ethics and intelligence are about the same level. Maybe even as bad as Microsoft, who frequently bullies programmers who use their products. It's trendy to hate Microsoft, though, so I'm sure this will get moderated down as 'flamebait' or 'troll' by the people who have never actually authored anything.
Ah well. Some days, it's just not worth getting out of bed, I guess. Much less out of bed to program. Maybe I should put a clause in my programs that says they can't be used by Corel for profit. See how they handle that one. *sigh* Yet another day of SNAFU. Situation Normal, All Fucked Up. Maybe someday we'll see some honest and decent corporations, but I guess it's going to have to wait for now - the almighty dollar calls again.
-RISCy Business | Rabid unix guy, networking guru
your company here.
shelby != ford
This line from the article made me a little tense:
Please think seriously before you create a unique license for your software product. Please use the GPL or the BSD style license if you can.
I don't want to overreact. I understand that the author is simply trying to encourage the use of established Open Source licenses. In fact, it has been my personal experience that many developers who fear the GPL don't entirely understand it ("The GPL causes me to sign over my copyright to the FSF. I can't modify my own program and sell it closed-source to someone later.")
But I want to make a simple point -- the developer has an absolute right to determine the license under which his/her product is released. If Joe Programmer releases the most killer Linux app ever, but his license agreement requires you to hop on one foot to the mailbox and mail him a check once a month, then you either start hopping or you don't use the program. You can send Joe nasty emails, you can publicly flame him on slashdot, but none of that changes Joe's right to use his General Hopping License. If you can't handle the GHL, Joe's product is useless to you.
Save the whales. Feed the hungry. Free the mallocs.
Actually, IIRC, the clause refers not to "routinely shipped libraries" but instead to "essential system libraries"...to me, the distinction is vital.
There is absolutely nothing "essential" about QT for Linux...I can rm that tree, and my system will reboot just fine. You can't however simply remove the Windows\System directory, and expect anything to work properly.
While accurate, you miss the context of why that original clause was amended to the GPL. RMS wanted to make certain that software authors could write GPL'd Motif software, which at the time was (and some could arguably write still is for commercial vendors) the most popular GUI widget library available for UNIX and it's clones. RMS didn't want to prevent Free Software authors from writing useful code, so he carved out a niche for Motif and told everyone to go ahead a write free Motif code.
Now I ask you, since Motif is completely closed source while QT meets most "Open Source" guidelines by at least Eric Raymond's perspective, how is this clause usable with Motif and yet unavailable for QT? Or better put, how is a closed source Motif allowed to use the "Essential system libraries" clause while an "Open Source" QT can't, when they perform exactly the same system level services?
Beats me. RMS, can you chime in here???
Actually, it was more meant for the proprietary libc's that came with the Un*x systems that GNU was bootstrapped on. When the Motif issue came about, people kind of looked the other way (RMS included), but it was an uncomfortable situation. Emacs (the quintessential RMS work) has never used Motif, even though the crufty widgets it used in lieu aren't great to work with (Xawe, I think).
Maybe my memory is fuzzy, but I distictly remember RMS stating the this exception was OK for use with Motif. And I'm certain that emacs is #ifdef'd for use with Motif, because I've built straight emacs linked against Motif before. There's a
[time taken to check] Yup... I'm right about Motif support, here's configure output from the latest emacs-20.4: [snip for brevity] The thing is, Motif was never a very viable development tool for free software developers. You have to statically link, people on non-proprietary Unicen usually don't have the development libraries, etc. So, people could look the other way because there was never any real danger of it starting a trend.
Qt is viable. That it is Open Source isn't terribly meaningful, because the GPL doesn't mention Open Source in any way -- and it shouldn't, because the Open Source Definition is hardly a legally robust document.
So, are you suggesting that RMS and his lawyers are going to draw a line in the sand and test the GPL in court here??? This seems not only dangerous but also an action with little gain. QT may not meet complete "Free Software" requirements as far as RMS and other total GPL proponents are concerned, but if the FSF decides to fight this issue out they better be ready to deal with compilation on proprietary UNIXen as well.
I think the most important point to make is that the enforcement of all of these licenses must be consistent across the board. If RMS, Linus, the GIMP developers, whoever, start playing games like "Well, you can derive from my source, but (s)he can't," just like how Sun's SCSL is structured, then we've denegrated the GPL and Free Software for all. It seems to me that if RMS includes support for Motif right in the latest emacs itself then GPL proponents really don't have a leg to stand on when claiming that QT and other GPL'd source can't be linked together; at least not if they want their argument to stay consistent.
I think RagDot is wonderful. I can offend both the feminazis and the stereotypical slashdotter (leftish, anti-capitalist, etc) in one foul swoop. Rag works in so many different ways....
1) Its rag like, think newspaper
2) Think Women (bitching and moaning)
- Think Red (need I explain?)
-Think Commie (well duh)....
...I said it, now watch me get flamed... hehe =)