Slashdot Mirror
Warwick Allison Of QT And KDE Fame
Konqi the Dragon points to this BeOpen interview with Warwick Allison of Troll Tech's Qt Library, writing: "Warwick is also a longtime KDE contributor with interesting things to say about GPL, QT Embedded, KDE, The Mythical Man Month, Distributed Development and scaling projects." If Warwick's words make you say "hmmm" (for whatever reason), you might also want to revisit earlier Slashdot stories about Embedded QT and Trolltech. [Updated noon GMT 10 July 2000 by timothy] Thanks to jdfox, who pointed out that I had perpetuated the misspelling of "Warwick" in the headline. Sorry, Warwick!
Looks like Mister Allison isn't a big believer in business being able to make a profit from an OSS business model. I'm not sure this is a healthy thing for the OSS movement, especially when voiced by someone with some visibility like Warrick.
Now, i can't naysay how he sees thing. Currently the money is definitly more likley to come from a company hiring an Open Source programmer who's got a good track record writing free software, but I'm hoping (and betting on) ESR's services-based model. Allison says it breeds non-user firendly code. Hogwash. Only if your company or programming team is unscrupulous or lazy.
I am hoping that most leading OSS companies like VA and Red Hat don't fit into that particular category. We've all had enough of unscrupulous and lazy software companies. Or at least I have.
So yeah, Warrick is right about why he got a job, but by saying that's the way it should work, he may be pronouncing self-fulfilling prophesy.
I hope not. We all have too much to lose if OSS can't make the big leap to big business.
Beware the Whyte Wolf.
With a gun barrel between your teeth, you speak only in vowels...
You know, i'd like to see some hard numbers as to just how many people have actually purchased a Qt License from Troll Tech.. And, more specifically, if that number is sufficiently large enough to really warrant keeping Qt something other than GPL'ed.
The only thing standing in the way (as I see it) of KDE being the dominant desktop for Linux is this persistant, irritating, annoying, pointless debate over the nature of the Qt libraries. Its a boat anchor that has been dragged behind KDE for far, far too long. Get it over with, guys.
Its only by pure luck that GNOME development has been centered more on building pointless foo-foo options as of late, rather can concentrating on basic usability issues. Both efforts have boat anchors. One has to do with politics, the other has to do with direction.
Want KDE to win? Make Qt free. Game over.
Bowie J. Poag
Bowie J. Poag
Because those weren't available on all platforms when Qt was written (it has always been cross platform, earlier versions had both macro- and template-based containers, for those compilers that didn't support templates back then).
Besides, QString is UniCode (while std::string is not, since it's just std::basic_string) and reference counted, so you get a performance bonus as well.
As for bidirectional scripts: there's some support in KDE I think, but as I don't speak any of the involved languages I'm not too sure about it, ask someone. Heck, write to Trolltech and ask them to pay you for writing it!
-- KDE programmer and computer science student in Klagenfurt, Austria.
I actually agree with the sentiments of most of what you say, but whats this?:
"... is because there's a relatively small number (7) of people working on it"
Where on earth did you get that figure from? Have you ever read either the source code, the contributors list, or looked on lkml to see who submits patches? It is _far_ more than seven.
-- MartinG To mail me: echo kewyjlcxyzvjfxbqwh | tr bcefhjklqvwxyz
When I read posts like yours I'm not sure whether to laugh or cry at how religously people can take the decisions of other people.
What Warrick Allison states is a very valid opinion which myself and many others have had. I've often asked myself, "if I was a developer working for a company whose livelihood depended on being able to sell support for the software I wrote wouldn't I be less motivated to write easy-to-use and bug free code as someone who worked for a company whose lifeline wasn't based on selling support contracts?"
I've never been in that situation but from the attitudes of most OSS developers today usability isn't a key word they are interested in. Frankly I feel that the whole selling support concept only adds the current esoteric nature and general poor documentation of Open Source projects. Second question "Why spend time developing comprehensive training manuals, online help and documenting APIs if you are trying to get users to buy a support contract?". Laziness is a well documented trait of good programmers and bad ones as well, so why would one not expect that programmers would take the easy way out especially when it would be beneficial for their employer?
PS: Exactly what do we have to lose if OSS doesn't make the leap to big business? I'm in it cuz I like it not for the IPO money, aren't you? I hope you realize that VA Linux is a hardware company that sells relatively overpriced servers which has nothing to so with making money on supporting OSS. VA Linux subsidizes its OSS forays by marking up their server prices, which to me seems like an alternate business model to ESRs.
PPS: This isn't a troll, if you disagree, post a response instead of moderating this down and making this place seem more like a mad house for raving zealots instead of a forum for rational discussion.
It is encouraging to see a success story about making money in the open source world. I believe that the industry should move towards open source as much as possible, and I completly understand how companies withstand the urge to go open source.
I have planned on going the same route that the author took in my software developement career. I am planning on doing some development work on some open source projects as an experience and resume builder so that I can fill in the part that talks about experience on my resume. It is a perfect place for new programmers to gain useful experience programming, but I have concerns about a market that is totally open source.
For all products to be open source is scarey. Now I know that open source doens't necesarily mean free. It means letting everyone know what your software is doing, possibly fixing it for their own use. This type of use is great. But what should a company do if they are making some true innovations. Release that software for the world to look at.
For example, Oracle may or may not have the most scalable and fastest database. But lets say that their programmers have developed some new methods that enable there database server to be faster and more scalable than the competition. How do the open source gurus claim that opening that software up would help Oracle. This is precisely the part that I have yet to understand. It is the same as patenting a new device. It lets the inventor or inventing company protect there investment for awhile so that even doing the research to begin with is worth it. The patent gives the garantee that the innovation is protected.
The article didn't help clarify my view of the issues with open source either. There method of dealing with the problem is by having some products open source and free and other products closed source and for sale. This doesn't really answer the question of how to deal with open source software and make money, unless this happens to be the final solution.
For example, Microsft opens the source to DirectX, IE, etc.. but keeps the source on NT and Office. Of course, this isn't really the core of what they do, it would be like giving out some toys, but not the real thing. Do people think this is the answer to the open source questions, or do people really think everything can go open source?
cp -R
But users of Qt and KDE should realize that the Troll Tech QPL license is bad for open source and for Linux. If Linux had been licensed like Qt, it would have never caught on. And if KDE succeeds at displacing other Linux desktops, it will largely spell the end of Linux as a competitive, open source client desktop operating system: if you have to pay Troll Tech under the conditions they require you to pay, as a developer, you might as well go with BeOS or Windows, pay less money, get more development tools, and (in the case of Windows) cover a much bigger market segment.
However, I disagree with your statement that "Troll Tech has been very nice to the Linux community". Troll Tech was merely business savvy and opportunistic. Adoption of Qt by KDE was one sweet deal for the company: they made software available to people who otherwise wouldn't pay for it anyway, they got a huge user community, lots of naive university students started hacking free software in it, and they got lots of feedback, bugfixes, and improvements, and adoption by several big companies in return. Without the exposure from KDE, Qt would have remained just one of many mediocre, tool-poor cross platform toolkits with a tiny user community. Troll Tech's "donation" paid handsomely for them. The open source community has nothing to excuse for or be thankful for.
Altogether, I agree: don't ask Troll Tech to free Qt, or KDE to change their license. Instead, just don't use those systems. Develop and support truly free desktop software instead.
1. If you want unicode, use something like:
typedef std::basic_string<short> Qstring;
2. Almost all std::basic_string implementations are
reference counted also
3. Reference counting can be bad in multithreaded
4. std::basic_string has been around for YEARS
5. The main question is the containers, std containers/algorithms are one of the best things about c++
I know it will cause a little pain to migrate, but it'll be worth it in the long run.
http://rareformnewmedia.com/
Woah--you are saying that Troll Tech can "take back" Qt for Linux whenever they please? And you are suggesting that we build a large open source infrastructure on that kind of basis? Something that can be rendered useless at the whim of a small private company or whoever acquires them?
If we based the predominant Linux desktop on that kind of foundation, someone like Microsoft could come in, buy Troll Tech for a miniscule few $100 million, and put a quick end to any client aspirations of Linux. Microsoft kills technology that threatens their position regularly by buying out the competing companies.
If there weren't any other reasons, that alone should be enough of a reason not to touch Qt with the proverbial 10 foot pole.
It's not a loophole in the QPL. The problem is that the QPL is not compatible with the GPL. Code under the GPL that is compiled/linked with QPL code can't legally be distributed.
The reason is that the GPL requires that the whole work (in this case the GPL'd code plus QT) be redistributable under the terms of the GPL. The QPL does not permit distribution of modified versions of the software, only distribution of the modifications in a separate form (patches etc), whereas the GPL does allow redistribution of modified version of the software. Since the whole work cannot be distributed under the terms of the GPL, it cannot be distributed at all.
The fact that noone is enforcing it does not make it any less illegal.
One part of this article has actually been on my mind for a little while now. How do you base a business around selling a service for the product you sell and still encourage usability? If we were to take RedHat for example, it would seem that an ideal market for them would be tons of folks with RH Linux running, and making it as hard as hell for them to figure it out on their own. Is this really a good model for Linux to adopt in the long run?
Certainly, no matter how many help files are included or intuitive features implemented into any OS there will still need to be some type of support structure. It just concerns me that ease of use would actually be contrary to a business model.
Another real quick example would be going back to Warrick's comments on Sendmail. My personal experience with Sendmail is limited at best, but from what I've seen there's a ton of stuff in there. I'd guess that it would take me several hours, if not days, to get this up and running properly being that I have no previous experience with it. On the other hand, I have set up IMail* on an NT Server for my company which I had up, running, and configured in about 20 minutes. That's 20 minutes without reading over any manuals or having any previous experience with it other than a basic understanding of POP and SMTP.
I don't mean to be picking on Sendmail as much as I am, especially considering it most likely has a lot more functionality than something like IMail. Thing is, IMail has a financial incentive to provide ease of use, where as Sendmail actually would lose money making things easier.
My apologies for not providing better examples here. The point I'm really trying to get at is whether or not selling software as a service rather than a product is a viable option for the future. Will it prove itself to be good for the computer industry as a whole? Is it still too early to tell, seeing as how the Linux community is just now really starting to focus on usability?
* I still use IMail as my company's E-Mail server after installing it a couple of years ago. Great stuff for doing mail on NT, and at $1000 it's actually one of the cheaper ones out there for that platform. Very stable and fast.
The line must be drawn here. This far. No further.
Your first question is easy: STL isn't supported on all compilers and OS platforms correctly yet. Why let such a trival thing like dependance on STL be a show-stopping trip up for cross platform independent code?
"But gcc supports it!"
Unfortunately, gcc isn't installed everwhere nor are the underlying support libraries.
On the other hand, simple C++ features seem to be supported almost everywhere. I have no problem with abandoning nifty STL features to insure stability and correctly working platform independent code.
Furthermore, destroying Qt is easy without ever triggering the "KDE Free Qt Project": stop the commercial support, and do half-hearted development. Then, KDE/Qt would be a toolkit with one of the most restrictive "open source" licenses in existence, completely unsuitable even for commercial users that have the money to pay. Hundreds of commercial developers would be left stranded and would have to port their code to other GUI toolkits.
I think it's completely naive to believe that the QPL and "KDE Free Qt Foundation" protect anything. And I cannot help but think that Troll Tech is disingenuous rather then incompetent when they put together this string of legally flawed licenses and agreements.
In any case, this discussion has gone off track. I responded to a posting by someone who claimed that (I'm paraphrasing) "Troll Tech could take back Qt at any time, so we shouldn't annoy them". I don't see how anybody who has that kind of belief can advocate building a large free software infrastructure on such software.
We can argue whether that belief is justified. I think it is partially justified, given the license and the loopholes. You may disagree. That's life.