Debian and KDE developers have announced an ambitious initiative to provide an enterprise Linux desktop operating system based on the KDE interface. Perens, an important open-source community leader, though, sees this move as pulling effort away from his similar GNOME-based plan, UserLinux.
Linux Developers Spar Over Enterprise Desktop Plans
2003-12-17 Linux 29 Comments
GNU/Linux/OSS always preaches the positive nature of choice. By very nature of their positioning, this stance is expected.
Any technology camp/social movement which grows from singularity will have similar tenets.
So is it surprising that while one standard is still being created, another already springs out? Hardly.
Best of luck to LSB and freedesktop and other entities that promote standards and unification; it’s not nature to this beast.
“There’s More Than One Way To Do It” may be the Perl slogan, but I fully agree that UserLinux should only support one desktop.
*nix has always been fragmented. Just because it is now happening to the Linux clone doesn’t mean that this is something new.
I usually agree very little with Perens (for personal and ethical reasons) but I think he is right to focus on businesses and the desktop. That is clearly the right focus. Whether it will gain traction in the long term is not clear at all. Eventually there will HAVE to be a reason (ie value-added) to go with one distribution or another.
What _is_ very clear is that GNOME’s use of LGPL (rather than GPL) is the key thing attracting business interest in the USA. Europe, though, has a firm KDE bias across the continent – should be interesting to watch this play out. My 0.02 – there will be more and more incompatibilities as companies stake IP claims. This was the history of commercial BSD and will likely be repeated. “The more things change…”
This is what democracy is about: putting your money/time/effort where your mouth is. This does not mean that Mr Perens is not right in making a choise it is just possible that he is prooved to be wrong. The great thing of it all is that whatever happens everyone comes out a winner. Gnome has to prove itself, KDE has to prove itself, desktop integration gets a shot in the arm. Mind you effort IS pulled away from Enterprise Linus but then again, not everyone would want to be associated with it in the first place and now a lot of convergence in happening..
Before everyone jumps on the I Love GNOME(TM) bandwagon OSNews is so fond of, you should read this blog post by Aaron Seigo.
Make sure you read the articles posted at the begining first.
While Qt’s developer Trolltech is a member of the Canopy Group it seems wise not to put all our eggs in the KDE basket.
I think Bruce is trying to say that market forces, at least here in the US, have pretty much aligned behind GNOME and against KDE for the Corp. desktop. KDE, despite being ahead in many areas, just isn’t making the headway here in the states it has in Europe. With the purchase of SuSE by Novel, this is only going to make GNOME usage stronger (despite what Novel says about still supporting KDE in SuSE). On the otherhand, there is no reason in the world the KDE folks shouls just give up, they have every right to fight for their desktop in the Corp. environment. My personal opinion: GNOME, while not as pretty and nice as KDE in some spots, is the way Corp. linux developers will be going. They made their choice and the trend appears to be going in the direction. Also keep in mind there is wiff of nationalism here. GNOME is view in some quarters are a more American distro (despite the fact it’s made of of contributions from all over the world) while KDE is much more European (German to be specific).
Trolltech isn’t a member of Canopy group. They are listed on the Canopy website as “portfolio companies” — companies that they invest in. The only mention of Canopy on Trolltech’s website, on the other hand, is the a page that describes how little of TrollTech Canopy actually owns — less than 6%. The vast majority of Trolltech (65%) is owned by the TrollTech employees. To tell the truth, I’d trust TT a lot more than a publically-owned company like Novell, whose only interest is in its stock-price.
Note, also, that Canopy lists another company “Linux NetWorx” on their website. I can find no mention of Canopy on the company’s website. Linux Networx was the company that built a Top-5 Linux-based super-computer for LLNL (big government research lab). Do you think that they have a plan to destroy their business by putting a $700 per-CPU SCO-tax on Linux? Canopy is full of shit, and its pointless listening to what you say.
I was reading pretty much the whole discussion on the UserLinux mailing list. For the last little while I have been trying to decide whether to work on GNOME or KDE. The decision is a difficult one, but during the discussion many good reasons were brought up to go with KDE and Qt.
But no matter the decision, it was made for the wrong reason. It basically came down to whether closed-source software could be made with no-cost tools or not. That was the only reason Bruce could provide. Now, my co-worker has his own company. It is basically a one-man show. However, he does not hesitate for a second to buy the tools he needs. The price to pay for Qt for closed-source development would – to him – be not an issue at all.
Now, Bruce certainly has lots of experience in this industry. That’s why I don’t understand why he bases his decision on a reason that is not an issue – at least for someone like my co-worker, supposedly someone who wouldn’t be able to afford the licensing costs.
All that is proposed to make linux more corporate can be merged back into Debian. I just don’t see what will make Userlinux so different from Debian. I thought it would include non-free stuff, but apparently the white paper states otherwise, claiming it is the task of ISV’s to include non-free stuff and keeping Userlinux free from license issues. I imagine Userlinux being something like this: Debian with the anaconda installer and a task including Bruce’s choices and some tweaks with KDE just an apt-get away. What was the problem again?
If all QT-libs are just an apt-get away, why discuss the favoring of GTK in the base install of Userlinux?
If QT/KDE isn’t included in base UserLinux it wouldn’t be certified and wouldn’t necessarily get supported by the industry. Well that means KDE supporters need to find industry support themselves and could easily be fully Userlinux compatible as it’s base is free.
No matter how many people try to explain it away, the qt license is still an issue. I like KDE more than Gnome and KDE would be the dominant desktop if not for the licensing issue. How about the one-man shareware developer? Remember shareware? For most companies, big and small, $2k per developer license is not that big of a deal, but what does this license entail? Do you have to pay $2k per developer per release or what? The best thing to happen would for somebody like Novell or IBM to buy trolltech and LGPL it.
> How about the one-man shareware developer? Remember shareware?
Sure. Name three shareware packages for Linux done in any of the other toolkits.
It’s not like there aren’t any other options.
> Do you have to pay $2k per developer per release or what?
No, it’s a one time payment. Only the embedded product has additonal royalties per unit, as obviously usual in that business
The licensing terms are clear. Its $1500-$2000 (depending on edition) which includes a 1-year support contract. During this support contract, you get all updates for free. After the first year, you can continue to get support and updates for $480-$720 (again, depending on edition) per year. To even small companies, this price is tiny compared to the cost of a developer, and with the high quality of the toolkit and the professional support, Qt could easily pay for itself.
The cost is probably bit out of reach for shareware, but I’d wager that shareware is been obsoleted by open source anyway.
How much Shareware are of any real use?
Open Source/Freeware has more or less killed that market anyway and if $2000 is a too large investment your software probably won’t have much to offer either.
“I like KDE more than Gnome and KDE would be the dominant desktop if not for the licensing issue.”
I thought KDE was already the dominant desktop. It always seem to win in polls.
It is true that the cost of a developer license is not the prohibiting cost.
Suppose that eg java, which allows commercial applications be developed and distributed royalty free, starts including SWT in addition to AWT / Swing. Now suppose the desktop environment is KDE. To get a 100% integration, it is favourable to have an SWT implementation using KDE technology, in this case Qt, DCOP etc.
Now suddenly, Java can no longer be used for commercial applications, since it is GPL, unless Sun negotiates a deal with Trolltech that allows them to do that. It is very clear that they wouldn’t provide any easier deal than developing for Qt standalone.
It is very clear that in the case of the core system libraries and components, you cannot have a royalty requirement for commercial applications.
If the linux kernel being GPL, required all user applications running with it either being GPL, or the developer paying a $1000 royalty to Linus to allow developing and running commercial applications with it; would it be the success it is today?
Why is it that Perens is all for competition, except when it’s competition to his own projects? The people working on KDE wouldn’t be working on Gnome to begin with. Gnome doesn’t appeal to everyone, just like KDE doesn’t appeal to everyone. To the older Unix crowd, neither Gnome nor KDE may appeal. Perens needs to get over it. People are going to work on what they want to work on, even if it’s seemingly duplicate effort. There are enough differences between KDE and Gnome that duplication is only on the surface. The reality is they are as different as Microsoft Windows is from from the OS-X desktop environment.
As for businesses/enterprise, most have already made their choices. Despite Sun and RedHat’s embrasing Gnome, most companies that I’ve had experience with are using KDE and Suse. Companies like seeing other corporations behind the products they buy. Trolltech being the owner and primary developer behind QT means it will be there for a long while. On top of QT, KDE is polished and easy to support. Aesthetically it’s much more pleasing to the eye than Gnome and it is better designed and more tightly integrated.
Now that Novell owns Suse, I expect to see further acceptance and market penetration of KDE and QT into corporations that have been traditional Novell customers inside the US.
I don’t understand, how the OSS community generally dislikes the MS monopoly yet some of its members push Qt and by doing so a new kind of monopoly.
Imagine the computing world in 10 years and lets just for a moment assume KDE konquered the desktop. Everyone who wants to release software that is based on Qt (which would be the obvious choice for a GUI by then) and does not want to choose a GPL compatible license is forced to pay TrollTech. Note that the conditions and the price for Qts commercial license are almost arbitrary changable by TrollTech. Now *that* is something I don’t want to see happen in the future and that’s the reason I would favour Gtk+ over Qt and GNOME over KDE.
Besides the technical advantages of Qt over lets say Gtkmm are marginal. Just because KDE may be technically superior to GNOME does not mean that the Gtk+ toolkit is the cause for GNOME’s faults.
I agree with most of your points. No matter how you put it your still beholden to a small company that can decide how much their going to charge you for that future license. Corporations just aren’t going to like that.
The problem with Gtkmm is that historically(as is the case with most language bindings), it’s always been a bit behind the release of gtk+, maybe a bit buggy, and the libraries aren’t shipped with it. The main guy behind gtkmm addressed these issues, which was covered by osnews. Also, historically, unix developers haven’t really embraced c++ for numerous reasons. Remember the c++ standard didn’t even come around until ’98.
Gtkmm really is a nice toolkit and doesn’t rely on a preprocessor like MOC. I know the reasoning behind qt’s decision to do a preprocessor as it goes back to not having a c++ standard(I think) when it was started.
but it just has some serious down sides.
it feels just like OS X did, when OS X first came out….just laggy.
now nothing is gonna touch an optimized xp system for that “i’m running in executive services mode” (microsoft move graphics from userland to something closer to their kernel called “executive services”)
that said, OS X has come along ways, plus they go way out of their way to make their windows dragging/resizing look flawless.
KDE? well i just leave it in wireframe mode. and i turn off all the effects. result? kde feels fast. knife through butter.
gnome? ….well, i have a lot of nice things to say about gnome. but the only thing i really want to say at this point is nautilus is a freaken dog, and metacity opaque windows-on-all-the-time is just tripe.
even with accelerated drivers from ati or nvidia’s website, using linux or freebsd, gnome always looks like crap when manipulating windows. and it certain doesn’t approache the potential snappishness that kde has….and it’s so far out of the ball park from xp.
note: my rant here is mainly about how it feels to navigate the gnome gui. most of the sluggishness comes from metacity and nautilus.
back to rant: i don’t care what tweaks i’ve done, or what platform, freebsd, redhat, debian, yellowdog(on a freaken mac)…sos.
video tearing when dragging windows around in gnome. would it hurt to sync the animation with the refresh rate…or how about some double buffering.
today was a bad gnome day. just got tired of my laptop feeling like 500mhz dog (it’s a 2ghz p4).
put kde on, it was like i could breath again.
Instead of knowingly pissing off a large portion of the user community a “leader” like Bruce should be finding ways of bringing the community together. If supporting KDE and Gnome in the same distro is too difficult then he should lead the way in better integration between the two. I use both QT and GTK apps on my system and i’m not going to use a distro that restricts the choice of apps I can use just because my toolkit is not supported
Anyway this whole thing is moot because Userlinux is going to go nowhere.It is designed to fill a gap that does not exist and no others will base distro’s araound it if they are going to be so divisive.
Its impossible to please everyone, therefore you shouldn’t even try. Microsoft dosn’t try to pleae everyone. They just force people through monopoly power.
The biggest mistake people make is tryingto please everyone. Look at Mandrake. They tried to please dekstop enthusiasts, Windows Converts, geeks, enterprises and everyone else. They ended up with a mediocre product for many uses. Its a swiss army knife yes, but for specialised applications, people choose other distros. Most people hosting web pages run Redhat instead. Many people wanting a desktop distro run something like Lindows, Lycoris, Xandros, and actually prefer SUSE. Geeky types love Debian and Gentoo. A lot more swear by Redhat/Fedora. This leaves Mandrake broke, under credit protection.
Linux is about differentiating, and USerLinux does that. I probably will not run it, I prefer Redhat/Fedora for GNOME, but I think to argue and say Bruce Perens doesn’t want competition is silly. He wants to compete with SUSE, Redhat, Mandrake and all already. HE does not have to duplicate their choices. If anyone feels Mandrakes treatment of KDE is better, then go Mandrake. If you feel no implementation is adequate then scratch your itch and make your own adequate distro. Address your own concerns there, and do not worry about what Perens is doing.
So statements like “Why is it that Perens is all for competition, except when it’s competition to his own projects?” do not make sense at all.
Debian is there for your taking. Just compile your KDE and make a distro. Wait a minute, Xandros, Lindows, MEPIS all do that. So what really is the problem here? Is it because it is GNOME?
Anyone who produces OSS with Qt has absolutely nothing to fear from using it. Once it has been released as GPLed code, it can never be made close source or proprietary. Newer versions can be made proprietary but the existing free version can be forked.
However, it would be extremely foolish for Trolltech to close the source of Qt since doing so would result in an immediate fork. If the company goes bankrupt, Qt will revert to an LGPL-like license. OSS developers have nothing to fear from Qt or Trolltech.
As far as shareware or cheapskate developers go, I’d wager nearly all of them would use the free Qt to work on their applications and then once they are basically ready, they’ll pay for the proprietary Qt for the final phases.
KDE’s core components are all LGPL.
So much for excuses not to use Qt/KDE.
I’m a happy Gnomer, and I can’t stand reading everywhere that “KDE is technically better than Gnome”. It may be true, but I’d like someone to explain me why and in which area is this sentence true instead of just trolling without arguments.
Just to clear up something, if Trolltech goes under, Qt will go to a BSD like license. Not LGPL. This would seem to have some not too savoury implications too. Qt could then be made into some other license. Microsoft could even fork Qt then. I do not know how being under a BSD license would affect continued development though. I wouldn’t want to call the agreement the best thing since sliced bread. Anyway, with Mono/GTK#, most of those advantages will be eliminated, if not reversed.
The realy good thing about GTK is that it is by the community and for the community, and for a specific purpose. This is allowing it to improve in ways the community sees fit. Stuff like Mono/GTK# is only the beginning. I would rather Trolltech made it LGPL, but that is their decision and choice, but it will haunt KDE for a while too. Once Mono reaches maturity (2004), we will possibly see the (main) advantage of Qt/KDE gone and we will see what happens from there.
just 4 posts up from yours, in loving detail is an explanation.
i’m not a kde fan boy. i actually prefer the way gnome looks.
right now, on my debian system, i use gdm to login to KDE, GNOME or FLUXBOX.
tweaking all of them to get it just the way i like them, the result is always the same.
fluxbox – lightning fast, of course there is little gui to speak of.
kde – not quite an optimized-xp fast, but close.
gnome – (with metacity and nautilus), slower then the ALL the rest.
i’ve done things like use the metacity wireframe patch, turn off file previews, turn off antialiasing, using the simplest theme…
doesn’t make a difference.
it looks nice, but form follows functionality.
i’ll take an ugly but fast desktop anyday.
fortunately with kde or fluxbox (or mac osx) i don’t have to settle.