Qt 3.0.0 Beta6 out, QCom gone

Normally a new beta wouldn't be dot news, but the new Qt 3.0.0 beta6 release is notable for one reason: QCom is gone. In a mail sent to qt-interest, Trolltech explains, "The feedback we received on this module during the 3.0 beta phase has been mixed. Many users think this module lacks the intuitiveness and compactness that they have learned to expect from a Qt API. Therefore, we have made the difficult decision
to withdraw the QCom API from the Qt 3.0 release. We will continue to develop this API until it is evolved enough for our customers, and will include the improved version in a later release.". They also note that the new Qt3 plugin functionality is still available under a much simplified API, see the changelog for full details. IMHO, this is all quite reasonable and is probably for the best -- even on the dot, people turned out to be quite opinionated about QCom, although most of the heat seemed to be directed towards COM itself. As a final aside, theKompany is pushing its Korelib as an alternative option.

Dot Categories: 


by not me (not verified)

Don't respond to these trolls, Shawn (especially that Wiggle guy), you're beating _yourself_ up! Just do what you need to do, and know that there are some rational people out here who like theKompany and what you guys are doing.

I'm looking forward to seeing the results of your joint venture with Hancom, it looks interesting.

Best of luck to you and everyone at theKompany!

by jamal (not verified)

someone want a creator of GPL apps, to change it to LGPL , so he can use in his commercial apps *to get some money* without paying.
is this the scenario above ?
what a niiiiice world :)

by deman (not verified)

I have a story.

Once upon a time, a man and his son went for a travel. They brought along with them their only lovely donkey. At first, the son riding the donkey while his father walking.

The went thru a village. The people there said " How rude are you son, you suppose give priority to your farther to ride the donkey. You are stronger than him". Upon hearing that, the son went down and his father rode on the donkey.

The went thru another village. The peoples there yelled "What an irrensponsible father, you suppose give the donkey to your son. You'd sacrifice for your son". Passing the village, the father went down and the son said "Now let we both walk since if anyone of us ride it, people think we are wrong".

Then after a while they went thru another village. The people there scream at them "How come you don't ride the donkey. It's for you to ride.". Upon hearing that, both the father and the son rode the donkey.

They went thru another village, and the people there yelled "Stupid! You beind so rude to the donkey!".

The son said "Father, what is right and what is not? Everything is wrong in peoples' eyes".

The father replied "Yes, what is right is what we believe. Whenever we believe in what other believe, we are wrong. When we are trying to satisfied all, we also wrong. Nothing right in us for every person".

To conclude, I believe that Shawn should do what he himself thought as right. Forget of those who just talk but never contribute something to the community. All they want is "free beer". They are parasites in our community.

- deman

by Filippo Erik Negroni (not verified)

ditto :-)

by kidcat (not verified)

I will try to get the maintainer of the fortune mod to axept this little story... good words at the right place!


by Wolfgang (not verified)


now that hopefully the license discussion is behind us...
Where is the 0.8 version of korelib? As it was said before the version out on
Web is 0.0.1 and very preliminary. I'd be really interested to see it. I am
a Python fan so your comments on the script capability make me very curious.

by Shawn Gordon (not verified)

I'm not sure if I said it here or somewhere else, but basically Korelib is the foundation upon which all of the embedding technology for HancomOffice is based. We are working on an updated release that has the required bits for this, it is a week or two off and then we will update the web pages and the other stuff. We are in serious crunch mode right now to hit this November release date with the whole suite. Sign up to our announce list on our web site under 'press' to get updates as they occur.

by John Sinnott (not verified)

Shawn -

Hope you get to read this, after all it is at the bottom of a LONG thread.

Try not to get frustrated with people like Wiggle. He, like many in the extreme "free" software crowd, are so narrow minded that nothing will ever please them. There is a certain bias from many of them towards QT and KDE that no amount of logical reasoning and hoop jumping will ever change. Do what is right for your Kompany first :) Trust me, your business model right now has many of rest us very impressed. I truly hope it is successful.

-John (looking forward to Aethera and Kapital 1.0)

by Wiggle (not verified)

>Try not to get frustrated with people like Wiggle. He,
>like many in the extreme "free" software crowd, are so narrow
>minded that nothing will ever please them.

Oh yes, II'm a free software extremist who is in favour of a
mix of Free software and closed-source development - just not scam
artists like TheKompany.

by Niftie (not verified)

Hey Wiggle, why don't you stop using GPL software altogether? Don't use Linux, don't run a standard FreeBSD (which includes gcc among others), etc. Just use your BSD software and Windows.

If people like to release Free software under the GPL, let them do it! It is their choise. People like you who try to restrict Free software shouldn't be trolling in discussions about Free software!

Run Windows if the commercial Microsoft license pleases you!

by Wiggle (not verified)

>Hey Wiggle, why don't you stop using GPL software altogether?

Why? Do you understand anything about this?

A license has to reflect the use the software is put to. Why do you think the Vorbis guys changed from LGPL to BSD for the encoder/decoder. The GPL is great, in its place... but here it is not being used for freedom, it's being used to force fees out of people.

by Niftie (not verified)

> The GPL is great, in its place... but here it is not being used for freedom,
> it's being used to force fees out of people.

The main idea behind the GPL is to force people; it is usually used to force people into releasing their code as Free software. In this case, at least there is a choise. Either release the source code under Free software conditions, or buy a license under commercial conditions. Freedom *and* choise!

In any case, since the code is released under the GPL it will be Free forever, even though some people may have bought the code and used it in closed circumstances.

by dbl (not verified)


How is TheKompany a scam artist? I think they're one of the best things to happen to KDE. There's been a few KDE applications put out by them, or sponsored by them.

Granted, Vivio isn't all that useful, imho (I like Dia, personally). But, now that I've got KDE 2.2.1 installed, I'll try kUML.


Everyone's entitled to their own opinions, I suppose...

by Amazed of London (not verified)

If you want people to take Korelib seriously as a component architecture, can I suggest that you modify the licensing terms to EXACTLY the same as Qt, and market it as an extension to Qt?

That is, triple-license it, allowing Free Software authors on Free *nix to choose between GPL and QPL, whilst proprietary software authors on any platform pay for a commercial license? Perhaps (although this is not necessary) also releasing a non-commercial version for authors of non-commercial software on Windows?

Trolltech have spent a lot of time and effort thinking about the ramifications of their particular licensing schemes, and it seems to work well. It allows Trolltech to contribute to KDE, support all forms of Free Software (gaining a great deal of PR and goodwill in the process), and at the same time make money from commercial licensing.

Any criticisms of the Korelib licensing scheme then are also criticisms of the Qt licensing scheme, and this allows TheKompany to shield itself behind the enormous amounts of goodwill generated within the Free *nix community by Qt.

Programmers who are using Qt as a platform will already have considered the licensing issues involved with their particular type of software. If the licensing terms for Korelib are identical in every way to Qt's, then using Korelib becomes a total no-brainer.

Just a thought.

by Asif Ali Rizwaan (not verified)

Why is KDE News not showing news, reviews, etc. lately perhaps 5 days?

by Navindra Umanee (not verified)

Because you're not submitting any? We can certainly use all the help we can get... Personally, I'm away on the sunny island of Mauritius, I'm not at the computer a fraction of the time, and when I am, my internet connection is often uselessly slow. But we all have our excuses.


by Carbon (not verified)

Simple, the developers are all too busy playing with KDE 3 while they still can! Message, source, and even binary incompat is a precious commodity, and they aren't wasting a single minute of it! ;-D

by not me (not verified)

OK, the first change that should be made is to the name. QCom will never win the hearts of programmers who hate COM as long as it is still named QCom. It just emits bad vibes or something.

by KDE User (not verified)

Hear! Hear! Now Miguel, there's a bonobo who can think up some good names.

by Carbon (not verified)

QAintCom ?

by jorge castro (not verified)


Keep doing what you're doing. You got programmers you gotta pay and you gotta make a living. I'm perfectly happy with what theKompany has given to the community - thanks.

Wiggle - You're an idiot, if you don't like it, USE SOMETHING ELSE. If you don't like the license, USE SOMETHING ELSE. But you better hurry on your LGPL Korelib clone if you're going to make it in time for KDE3.

by LMCBoy (not verified)

When I configure qt-3.0beta6 like this:

% ./configure -xft -qt-gif -qt-jpg

I still get -DQT_NO_XFTFREETYPE in the Makefiles. Am I doing something wrong, or is xft disabled in the beta? (I have TrueType fonts working with Qt-2.3, and I have ensured that I have libxft and X11/Xft/XftFreetype.h).


by A.C. (not verified)

It should be obvious by now that Wiggle is yet another identity that Miquel uses on boards like this and linuxtoday to troll. The style is exactly the same as a popular "ac" that use to post these exact kinds of things. He got found out and shut down there. I would assume that he would either be to weak or busy with MONO to keep messing around with KDE. It's really sad that he was one of the original KDE developers but has turned into such a spaz.

by Niftie (not verified)

Are you seriously saying that the person behind Wiggle the Troll is the head developer of the GNOME project? This is a pretty strong claim. Can you back it up with some references?

Niftie Noufty Noouh!

by Roberto Alsina (not verified)

Well, Miguel spent several days on #kde saying stuff like "Why are you helping this evil project?", so I put nothing beneath Miguel.

Ok, an alligator without a hat *could* be lower than Miguel, but I am not betting.

by ac (not verified)

How do you know this was Miguel on #kde ?
Maybe it was just his dog ?

I dont think that thinking every troll is a desparate Gnomer is the right attitude.

Please be more nice to each other. Open source has enough problems ( i.e. venture capital running out ), we dont have to spread hate among competing groups.

We should ignore the ones who cant cope with our moral standards. This is the
easiest thing to do, it
1) saves time.
2) To the naive, it doesnt look as if someone has pay real attention to his
arguments. Just look at the thread he has caused ~100 messages. He must have
had something to say, didnt he?

Arguing is a god thing to do, it keeps you from starting to get squared, but not every argument is sane .... ( It seems I cant reach a nice conclusion, so I go to sleep )

by Roberto Alsina (not verified)

> How do you know this was Miguel on #kde ?

The linux world was much smaller then. I had already been on IRC with him.
And he replied to his own email address, so unless he was a very clever fake, he was himself.

by Beubeu (not verified)

I think you have watched too much X-Files or other serials of this kind, and I think Miguel is clever enough to spend his time doing some more interesting stuff ! Anyway, while KDE popularity is increasing, the number of stupid comments such as those of Wiggle will grow up... (maybe we should add an option "ignore stupid trolls", and, for those who like trolls, an option "ignore constructive comments" :-)

As for technical aspect, I would really appreciate if someone could give me some information on technical aspects of Korelib, especially in comparison with COM, KParts and Bonobo.

by Shawn Gordn (not verified)

There is a lot of confusion that Korelib is a Kparts replacement. This is not true, it doesn't do the UI plug in part, they are complimentary pieces, and are both used in Aethera for example. I've asked my programmers to write up a short synopsis to explain the difference and the uses. Maybe the dot will run it as a story if we are nice to them, I saw Navindra asking for content recently :), and I think that given the amount of discussion here, it coule be a worthwhile piece.

by RR (not verified)

I have to say that is has been most fun ever to read discussion above. This wigle guy seems to have serious "my way only way" problem, or was it gnu way
only way, don't know don't care.

In my short life i have noticed at those people who have this obsession of being right, no matter what, are usually those who are often most wrong.

And if he really is some head guru of "other" project,he has even less reasons to keep noise, if he is right, the opposite will win eventually, no matter who says and does what and when.

Anyway thanks for the kde people for wonderful desktop. Even tho this is posted from win2000 i still have linux lying around here somewhere. =)

by Wiggle (not verified)

>Anyway thanks for the kde people for wonderful desktop.
>Even tho this is posted from win2000 i still have linux
>lying around here somewhere. =)

Ahhh.... nothing sums up KDE supporters like that statement.

Ignorance, rank hypocrisy and total non-involvement - all in one rotten package.

by RR (not verified)

Oh well i am so sorry my comment didnt please you. I will try to do beter next time. :p

Besides how do you know where i am involved in? oh yeah i am running win2000 in work i must be totally clueless, woe me

end of thread hopefully, naah have your last word i know you want to say it anyway.

by Amazed of London (not verified)

It should be plain by now that whoever this Wiggle character really is, they are only here to troll.

Remember that trolling is only done to provoke a reaction. Ignoring a troll cuts off their air supply, and eventually they die.

So, as often seen on bridges: Please don't feed the trolls. (Well, maybe you can send something to the ones who make Qt, but they're fluffy trolls.)

It matters not one bit really what Wiggle thinks or posts. KDE is the de-facto standard desktop for Linux. With all but one of the major distributors shipping it as the default, I can't see that changing, either. If Wiggle doesn't like that, then I suggest that they get involved with Enlightenment, because e17 looks like the most interesting and viable (not to mention sexy) alternative.

BTW, some people might find this article on The Register interesting, for an industry viewpoint that's outside the Linux desktop turf wars:


by Andi (not verified)

Instead of independently realizing the widget-library (under LGPL) and independently to determine direction of development of the project (KParts, instead of QCom), developers KDE are on the bit TrollTech :-(

Because of it constant copying a code is necessary them...

It is favourable TrollTech, but not to developers KDE.

by Me (not verified)


by Brad Neuberg (not verified)

Hi. I am a former Java CORBA developer. I have also played around with GNOME Bonobo, Windows COM components, KParts, OpenOffice's Universal Network Object's (UNO) infrastructure, and Mozilla's XPCOM architecture; you can say I'm really into component technology. This is just one developer's opinion of what KDE and QT should do when it comes to a strong component architecture.

First, since my primary background is CORBA, I'll talk about that first. I worked with CORBA for about 2 years, on and off work projects, and I can say that it is not the way to go. It has nothing to do with performance concerns, which the ORBit project briliantly solved. The chief issue is that Raw CORBA, sans Bonobo, is really quite hard to develop with both mentally and while programming. First it takes a developer an eon to come up to speed with the million different acronyms and ideas in CORBA (BOA - Basic Object Adapter, POA - Portable Object Adapter, plus twenty thousand other acryonyms); this assumes that the developer is already quite sohisticated with both object oriented programming plus advanced object patterns. CORBA code, especially when dealing with C code accessing a CORBA ORB, is almost completely unreadable in my opinion. It really destroys any hope of self-documenting code. Further, keeping the generated CORBA skeletons and stubs in sync across a huge project is very difficult and time consuming, even with advanced makefiles; basicly CORBA interfaces and generated code can be quite brittle. For some languages, such as Java, many files are created for a single CORBA class; various Helpers and Holders, for example. I used to get a headache just dealing with the many different ways of accessing, storing, and dealing with my CORBA classes, for example. The only thing worth saving in CORBA, in my opinion, is IIOP (Internet Inter-Orb Protocol), the standard that makes it possible for CORBA ORBs to communicate with each other over the Internet.

The Bonobo project attempts to answer these problems, but it is my opinion that it is not successful. It adds a layer above the CORBA objects such that you don't have to deal with IDL (Interface Definition Language) or any CORBA concepts; instead you deal with property bags and pass these around. While this does make it much easier for the individual programmer, the chief problem is that turning everything into property bags destroys interface based programming and is akin to variables hanging off an HTTP statement. As soon as you want to go 'clean' and design nice IDL interfaces you have to delve into CORBA again. So my chief criticism of Bonobo is that by making CORBA easy with property bags it throws the baby out with the bath-water.

Windows COM was nice, just several years ago. It makes it possible to do some really powerful stuff, but the completely unreadable GUIDs combined with all of the complex OLE and Automation interfaces when you want to do anything GUI related are unspeakably complex. Windows COM _does_ get it right by using something known as typelibs. Typelibs are a binary file that holds descriptions of the components you are dealing with, such as the method names and so on. This means that you don't need to generate stubs and skeletons, so code management becomes much easier and can be more 'automagic' based and linking can be done more at runtime. My chief criticism of Windows COM is that it gets really complex when you want to do higher-level GUI-type work, has developed several years of 'interface cruft', and is of course proprietary without payola porting toolkits.

Mozilla XPCOM is nice; its cross-platform, and can be binded to by several different languages. It's sort of a nicer Windows COM; it doesn't go further than that. It also doesn't address cross-machine communication. One problem is that it is not that 'battle-hardened' and optimized, since it is newer than other component techs.

The two nicest component technologies I have dealt with are OpenOffice's UNO components and KDE's KParts technology.

I'll start with OpenOffice's UNO components. The UNO technology is several years old and has been used to support several releases of a huge project, and is therefore more solid in my opinion. It uses typelibs rather than generated skeletons and stubs, so it is much easier to work with. When writing code that accesses an UNO component, it is very readable and understandable, which I believe is important. It has very strong 'programing in the large' support, which are easy to use, such as supporting interfaces, services, and aggregration support. OpenOffice already has an extensive set of API's that can be called as UNO components, and these are readable and easy to use, getting rid of the complexity that Windows COM introduces when you want to do higher-level GUI things. It is also cross-platform, and seamlessly supports OLE and Windows COM transparently to the programmer. Writing an UNO component is also very easy.

KParts is one of the easiest component technologies I have ever seen, and it gives you alot of bang for the buck. The use of XML to do user-interface merging is brilliant, in my opinion, getting rid of one of the biggest sources of GUI component complexity. Creating a KPart component is very easy, and by using the DCOP IDL compiler you can easily do interprocess and cross-machine calls. One downside to KParts in my opinion is that it is not cross-platform. Further, while I don't know much about it, binding different languages to KParts doesn't seem that easy. Also, the inter-machine communication protocol doesn't seem accessible outside Unix.

I believe the perfect component technology would merge KParts and OpenOffice's UNO components. Merging more of KPart's ease of use with OpenOffice's strong UNO infrastructure would be killer. UNO makes cross-language binding easy, as well as bridging over different component technologies easy; it has the name Universal Network Object's for a reason. It already supports CORBA IIOP for communication with Bonobo and Java ORBs, as well as OLE functionality cross-platform. I think this would be a killer combination. Throw in support for XML-RPC and SOAP using UNO's bridging technology, and you would have an easy way to do cross-platform (since this works with Solaris and Windows), cross-desktop (since this would interact with Bonobo through the Bonobo bindings occuring at OpenOffice), and easy to script both locally and over the network (using XML-RPC and SOAP), and you would have a killer answer to .NET in my opinion.

I'd love to hear people's opinions on this (both heated and non-heated :).

Brad Neuberg

by Chris Bordeman (not verified)

Maybe you would like to expand on this some and submit it as an article?

This sounds great! I'm not much of an expert on component technologies myself (read : no experience :-) but, anything which allows more integration between KParts and Bonobo, as well as allowing more diverse networking support, sounds great.

Question : would this mean that, say, OpenOffice would be easily able to load KOffice documents, just by inserting said document as an object? Or vice versa? It would seem so, since they would both be using the same component model. Then again, as I said, I know very little about this particular subject.

by Brad Neuberg (not verified)

If OpenOffice and KOffice were rebuilt to use the same component architecture that I proposed, then they could indeed by embedded in each other.

Is what you propose at all related to Apple's (dried-up) OpenDoc technology?

Bonobo isn't all about PropertyBags. PropertyBags are simply a way for components to share key/value pairs, and get notifications when they change.

BBonobo has a lot more than simply propertybags. I've programmed major bonobo applications and have only used a propertybag once. Most of Bonobo is nothing to do with Propertybags

by Brad Neuberg (not verified)

Lain, thanks for responding. Could you tell me more about what Bonobo offers other than PropertyBags? How else does it hide the CORBA underpinnings?

Brad Neuberg

Bonobo also offers:
Persistant Streams
Agregation of interfaces into one object
Controls (not quite the same as embedding)Menu/Toolbar merging, using XML like KParts
EventSources and Listeners
Printing and zooming of embeddables
It hides the CORBA underpinnings by linking CORBA_Objects to GtkObjects (GObjects in Bonobo2). This means that in basic usage of Bonobo you never have to deal with CORBA itself, you just call methods on the GtkObjects that you create and Bonobo does the work for you. If you want to create your own interfaces, then yes, you'll need to get dirty with some CORBA stuff, but again Bonobo has macros and helper functions so that all you really have to think is how to do what your interfaces do (rather than how to set up your interfaces for CORBA)

As an example, I've written a Bonobo based CDDB querier. Of the 600 lines of C that it took to write, less than 100 of them are to do with Bonobo, and of that, most of them are boilerplate.

by Daniel Mayer (not verified)

Sounds like an excellant idea. Although I don't think this would be able to find its way into KDE 3.0 unless a great deal of interest amoung developers is fostered. Even then, integration of UNO would probably push back the release of 3.0 by at least a month or two. However, I think it would be worth the wait. (This all assumes that UNO will not undergo any major changes before the 1.0 release of OpenOffice).