Posted
by
emmett
from the kde-last-train-to-trancentral dept.
[vmlinuz] writes: "Linux UK has got an interview with one of the key KDE developers - David Faure. Interesting reading, particularly with all the press that KDE/Debian is getting."
This discussion has been archived.
No new comments can be posted.
You DO realize that silently auto-starting servers is EXACTLY what many people hate about Microsoft, don't you? And don't give me a line about how secure you've made this server--the best security is NO server. -- Wanna hook MAPI clients to your Tru64/AIX/Linux server?
Oooh, this is going to cost me karma as trolling, but...
How is this all-inclusive, single-source applications/OS environment any different than, say, Windows and Office?
I know, I know: it's not monopolistic, you are able to run other GUIs, it isn't tied into the OS, etcetera.
On the other hand, there are a *LOT* of Linux supporters who really get off on flaming Microsoft for packaging Office with Windows, for the way Office buggers with the GUI, for their scripting support, for bloatware, and so on.
All accusations that could be made toward KDE, and probably Gnome. Like, when Linux finally rules the common man's desktop, is an installation package that by default installs KDE and KOffice really any different than what MS does? Isn't there a parallel with Office's non-standard dialogs and widgets, and KDE's different-from-other-GUIs dialogs and widgets (there being a complete lack of standard for Linux GUIs)? Is KDE scripting going to be any less a security risk than VBA? Isn't X-Windows+KDE+KOffice not abhorrently oversized, just like Office?
Yes, I'm fully aware that it's a different situation... but not so different that it's not more than a little ironic that what KDE is praised for, MS is condemned for.
This blurb was taken from KNews: KMail Security....Lessons from ILUVYOU? -- There was some discussion on the kde-devel mailing list about KMail's security. As it looks KMail is far more secure than most of its counterparts. The testing and discussions continue.....
Anyway, I used to run Win NT on VMware (not the latest version) in my last job. We had a license for Win Maple, but not for Linux, and sometimes I had to write some Maple stuff for teaching. Worked fine, and I certainly tried some other apps, and besides sound everything worked fine (haven't tried to fix that). When I have to use a Windows app, I like it to run in a X-Window, so I still can use my preferred system.
I will probably have to use the W98 partition of my current notebook under VMware in my new job, too. But I definitely would prefer using Wine to run them. Tried it recently again, and they did a great job since last. So, if the Win-apps you have to use run with Wine, forget about VMware, if not, it's a good 2nd choice.
Yes, I wish I could rephrase my answer on that. I was totally surprised by the question - note that this was *before* the $3K thing, since the interview happened on June 1st. I replied to "Why Qt uses the QPL and not the GPL", not really to "what to do about that alleged incompatibility" - sorry about that.
So, what do I have to say about that incompatibility ? That people should look at what they want/need. KDE is free and open-source, since it's GPL/LGPL. Qt is free and open-source, since it's QPL. Isn't that what you want to use ? Free and opensource software ? RMS may find an incompatibility between those two licenses because of the way *he* interprets them, but remember that the GPL hasn't been tried in court yet, and interpretations vary. Why does everyone here (and elsewhere), dreaming of "freedom", blindly obey what RMS says about licenses ? There _may_ be a slight incompability somewhere (I'm not a lawyer !), and I hope it will be solved at some point, but I don't see how that prevents debian from shipping KDE Qt ! Both are free to use, distribute, hack, etc. Isn't that what you want from free software ? So stop the license crap, and benefit from all the free software that is available, including KDE, while lawyers tear their hairs off upon how to interpret a line in a license.
And I really wish people would stop talking about "non-free" for something that *is* free (in all meanings of the term).
Have fun with KDE ! 1.91 will be available in a matter of hours/days! This comment posted with konqueror [konqueror.org]:)
Exactly. Scripting works by inter-process (DCOP) calls. So forget all those talks about "viruses and other scripts that can wipe out your hard disk". All you can do is control a KDE application - and no, konqueror doesn't offer "rm/" in its DCOP interface:-)
Um, try reading the postings from the article about the $3000. There were links to mail list postings and the like (http://lists.kde.org, the kde-licensing list). There is no need to rehash that in this forum.
a "script kiddie" is someone who uses canned DoSes or exploits written by someone else with no knowledge of their inner workings.
"script kiddies" dont WRITE anything. I keep seeing people on slashdot misuse this term to refer to anyone who does anything malicious and I'm sick of it. Forget the big hacker/cracker debate, don't misuse the term "script kiddie".
You wish that KDE core applications were developed out of CVS ? With the changes that happen in kdelibs quite often - to build a better API, and because of all the new stuff in KDE 2.0 (KParts, DCOP, XMLGUI, etc.) - this is almost impossible. What you gain from having apps in CVS is that they can be fixed by many people, especially those who change kdelibs. Out of CVS, it's up to the author to fix his app to cope with the changes, and _that_ can be piss him off a lot, obviously.
I realize that this is not an easy problem. But there is a better solution than hiving off all development into the CVS and core team. In short, there should be a stable (1.1.2) and a development tree (1.x.x) just like the Linux kernel. The stable tree should actually be a number of trees, with each core app having its own tree. And also a zipped snapshot, so an interested developer can grab it and immediately compile it with no fuss or muss. Only after that the developer needs to put in the extra work to install cvsup and sync up to the latest tree, the mailing list, the maintainers, etc.
The stable trees can be maintained by a completely different set of developers - there will be no shortage of volunteers And so, no additional load on the core developers.
This is not about "controlling", this is effectively about helping....
Yes, of course, but sometimes the cure can be worse than the disease. Sometimes the first cure you think of isn't the best one, even if it seems to help.
To answer the more personal comments...
The personal comments were not necessary, and I apologize.
...KDE is definitely nobody's fiefdom.
I was referring strictly to the mailing list.
...the idea of developing core apps out of CVS really doesn't make much sense. You want to "open more" the development by making it private ?
I don't see how any of my remarks could be interpreted that way.
About KDE 2 apps that are not in CVS : from 1.91 onwards, they should be able to compile and run on top of KDE. The API shouldn't change anymore. No need to wait for the official 2.0.
That's nice, but the API is going to change again in the future (I hope! Ossification is bad) and the same problems will come up again.
Finally, I'm not a paid fulltime developer... yet.
Remember the wonderful children's book "Holes are for Digging?" No? Popular in the '50's when I first learned to read.
In the same way, libraries are for linking. Dynamic libraries are for linking to ANYTHING, regardless of license. If a company, author or organization makes dynamic libraries available on a system, then ANY app may link to them. It may be different with hard linking libxxx.a libraries which are properly regarded as object files which become part of an application itself, rather than something linked to at run time.
I will link to any dynamic library on my system regardless of how it got there and regardless of artificial license restrictions and if somebody don't like it they can sue me. Hell, you can even link ANY windows app to the most proprietary closed MS libraries in existence. That's why there are there - to be linked to.
These objections to the Qt libs being linked to by GNU libs and apps are bogus. If GNU don't want its libs to be linked to, it should remove them from public distribution. It is as simple as that, regardless of legal contortions and trickery used to obfuscate common sense and prove the opposite. And, I don't think I would have much difficulty at all getting a jury to agree, if it came to that.
If free software and Linux in particular gets bogged down in these artifically created legal power struggles over purtiy, and it is nothing but a power struggle among distributions and companies hoping to profit from free software over turf and market share, then it can kiss its ass goodbye. Gnu, and free software generally, will win or lose the battle for the integrity of free software on general principles of free speech and intellectual freedom and fair use, not in nitpicking the fine print of licensing terms. That approach is unconvincing and defies common sense, whereas the other, broad-based approach is easily understandable and has integrity as a whole. Similarly, in the realm of software patents, restricting use of common altorithms and interface methods is theft. It is plain wrong, and anyone who writes code knows that. Gnu can never win its case by arguing the fine points of this or that particular patent, but must make the case for intellectual freedom and the (sacred) value of our common intellectual heritage convincingly, wholly and without excuse or unnessary rationalization. It is really a freedom of speech and freedom of association issue. If we are forced to communicate digitally, that freedom cannot be restricted or abridged with artificial licenses and patents. Even the average Joe Sixpack can understand that.
Holes are for digging, libs are for linking, and computers are for computing.
I so much agree with you ! Very well said. What about getting a slashdot account, to post at 1 instead of 0 ?:-) (I'm posting this one so that people browsing at 1 wonder what I'm replying to:) )
Close, but no cigar. It has to be: KDE_1_2_STABLE CVS, and it has to be complemented with the individual apps also posted in simple tgz format. In addition, the individual apps in, for example, kdenetwork, should be available as individual tgz's. Why should you have to download and build 6 apps when you're only interested in one?
The branch is called KDE_1_1_BRANCH, that's the post 1.1.2 tree. I don't see why you've got a problem with that.
As for downloading 6 apps, well, you only have to do that once, and you NEVER have to compile every app. For KSysV development (meanwhile moved to KDE2), I never do that, I just compile kdeadmin/ksysv, and maybe kdeadmin/doc/ksysv.
The only thing I see getting in the way is core developers are not, at this point, willing to recognize that non-core developers have a big barrier to getting started - much bigger than most other open-source apps. Even mozilla is easier to download and compile than KDE, let alone individual KDE core apps which have to be dug out with considerable effort from a rather indimidating CVS tree.
Really, in what way is downloading & compiling KDE less easy than doing the same for Mozilla? There ARE snapshots of libs and app-packages available, and the next set of those will even contain CVS information, so you just need to update a specific directory from that on.
You DO realize that silently auto-starting servers is EXACTLY what many people hate about Microsoft, don't you?
Exactly! It's a calamity. I mean it's just terrible. I turn on my box, and this thing call a 'kernel' auto-starts. I start X and this X protocol thingie starts running. I load my email client and it "auto-starts" a library to connect to the pop3 server.
When you load a program, you expect the program to load itself and function as it should. If you don't like the functionality, then use something else. Complaining may make you feel better, but doesn't do anything productive. If I want an email client that has pop3 support but not imap support, then I'd use an email client with onlyh pop3 support. I won't whine everytime someone tried to add imap support to an email client.
Really, in what way is downloading & compiling KDE less easy than doing the same for Mozilla?
Instructions for building Mozilla:
1) Download the source tarball [mozilla.org] 2) Unzip into directory of your choice 3) cd into the directory and type: make 4) Go for coffee, come back in 40 minutes 5) Type./mozilla and go for a surf
Instructions for building kmail:
0) Look for a kmail.x.x.tgz. Find a very old one. Get confused. Hunt around with RPM. Figure out it's now part of kdenetwork. Look for kdenetwork.x.x.tgz. Find a version that's way too old. Learn you have to get it out of CVS, and you have to get all of KDE with it. 1) Explore kde.org. Find the part that tells you about cvsup 2) Download and install cvsup 3) Find the instructions about how to get download KDE via cvsup 4) Decide what parts of KDE you actually want to download. Get confused. Decide to download everything 5) Start the download and go to sleep 6) Wake up, discover you got disconnected, repeat as necessary 7) Get a cup of coffee, sit down, and start figuring out how to build it 8) Guess that you have to build QT first. Try to build QT in the obvious way. It doesn't work. Go back to instructions. Learn you have to link a directory first. Do it and try again. 8) Go out for a walk while QT builds. Come back, find it's still building. Play Mahjong. 9) Try to use the same strategy to build kdebase. It doesn't work. Try some more. It still doesn't work. 10) Go back to developer.kde.org. Learn about the build script. Download it. Fiddle with it until all the path variable are correct. It finally seems to work. Out of fear, decide to build everything. 11) Go to sleep while it builds. 12) Wake up. One of the first packages failed to build - it's because somebody checked in something ugly that broke the build. Start cvsup again. Go for another walk while the tree updates. Later that day, start the build again and go to sleep. 13) It builds this time. Now figure out how to install it, without conflicting with your current KDE installation. Go for a surf. Find a howto on this subject. Read it. Fiddle with paths. Now it works. 14) Run kmail from the new desktop
OK, I may have missed a couple of steps, or a couple of details, but that's essentially what has to happen. Umm, don't you think there's a difference in difficulty and time investment between the two procedures?
There ARE snapshots of libs and app-packages available, and the next set of those will even contain CVS information, so you just need to update a specific directory from that on.
But why can't I just download a tgz of the specific package I want to compile, like almost any other open source project? Then all I need are headers for kde and qt, and I'm up and hacking. There is a *radical* difference in the amount of work required to get started. Later, when I've got some code I'm proud of, I can figure out how to submit it to a maintainer. In other words, the normal open source process.
Package snapshots only eliminate a few of the steps I listed, and you still have to figure out which packages you need, which directories to put them in, and how to set the paths. This is a *lot* of work if you've never done it before.
Mainly, what I hear you saying is: there's no problem, it's just your imagination. But it's not my imagination, I've been through this. I know what the difference is between downloading and compiling "spruce", for example, versus downloading all of kde and qt just to compile "kmail". It amounts to *days* of extra work the first time you do it. It's not my imagination.
Perhaps I should rejoin the KDE mailing list and we can continue this discussion there. *But* I still have this nagging feeling - why should I do that, when the last time I did, I just got flamed? --
Sorry to say, but either you WANT it to be difficult, i.e. troll (though I don't think so really), or you construe things to be much more complicated than the really are.
Yes, the list is longer than for Mozilla. OTOH KDE IS much more than one app, it's an environment/application framework.
This is exactly my point. I do not want to compile KDE. I want to compile kmail. So why am I force to compile 6 more apps and a good part of KDE and possibly QT? When we both know that it would be easy to provide to me the source code I need just to compile kmail? And after doing what you suggest, I wind up with the wrong version of kmail - one that runs under a beta desktop that I'm not using. Fixing bugs there doesn't fix the immediate problems. Downloading other snaps doesn't gives me an old frozen version that nobody is fixing bugs in.
Good gosh, man, can't you see that if people are complaining, there must be a problem? It takes a lot to get people to speak up on something like this. My original post was moderated up by four readers and not marked a troll by any, as it easily could have been. What does that tell you? And what does it tell you about how it will be perceived if you keep sticking to your guns instead of looking for a resolution?
By the way you left out the "fiddle with environment variables" step in your list, maybe others. --
Well, because kdelibs and qt are libraries that kmail depends on.
You won't have much luck with Mozilla either, if you haven't got gtk+ or libc.
As for "my original post was moderated up": my posts won't be moderated up because the story is too old, we all know what stories are actively moderated, and one two days old isn't among the likely list.
And what's the problem with downloading the tarball of the 1.1.2-kdenetwork? Sorry to say that, but to me you appear to be complaining for the sake of complaining.
And you don't HAVE to compile 6 apps, even if you don't use the nice "inst-app" file, you can still simply uncomment the other dirs in the Makefile after configure.
As for "fiddling with environment variables": yeah, you're completely right, setting two variables (KDEDIR and QTDIR) is just SOOOOO hard...
As for "fiddling with environment variables": yeah, you're completely right, setting two variables (KDEDIR and QTDIR) is just SOOOOO hard...
Ok, ok, you win, you win, actually there is no problem at all with obtaining compiling KDE apps such as kmail, and as I have better ways to spend my time I'll forget about it for another year, thanks for the advice. --
I've noticed several wrong impressions of KDE2.0 here. First of all, throw out all notions of KDE1.1 and whatever you think of it. Secondly, go download Konfucious (the 1.9 beta) from the KDE site. Here's what's different:
QT2.1x is a different beast than QT1.x or even QT2.0. It's faster, more stable, and less ugly. The theming support is great. It's just a more capable toolkit.
KDE is now structured differently. There is no more kpanel (RIP). It's been replaced with kicker. There is no more kfm. Half of its functionality was inhereited by kdesktop (which puts icons on the desktop and sets the background), and the other half by konqueror.
Konqueror is not just a file manager. It's a file viewer, as well as a file viewer for the network. The combination of these two functionalities means that it's also a web browser, but it does other cool things like embed a KOffice part to be a KOffice file viewer.
KDE is no longer just a library that you need to run. While there are kde libraries, there is also a dcop server that handles interconnections between programs. Running a KDE2.0 program requires you to have an active instance of the dcop server, or else the application will bomb.
The KParts technology is a beautiful thing. While it may be superficially similar to M$ technology, in reality it's much more cleaner and useful. Microsoft's technology starts to look like a horrible kludge compared to the elegance of KDE2.0. KDE2.0 also implements enough of the M$-like ideas (KOM, etc.) that major commercial ports of traditionally windows-centric applications (e.g. project Kylix) can start to happen.
KDE2.0 does not look anything like KDE1.1. KDE1.1 may have looked like the illegitimate offspring of CDE and Win 9x, but KDE2.0 has an elagance of its own, particuarly with the System theme, though you can theme it any way you want.
Best of all, it's modular, so you can replace any single component with another and have it work fine.
... when you can use Emacs? It's on every platfom I work... and it does everything (except tap dance). Hmmm, I wonder if the Win2K telnet service can be configured to use emacs as its shell...
I'm too lazy to go out and find the urls at the moment, but there are two tools that can fix your MS-Word HTML problems. The W3C [w3c.org] now offers "HTML tidy", which includes a handy option to undo word 2000's damage. There is also the "demoroniser", written by a gentleman whose name escapes me and updated by Tom Christiansen et al... check Perl.com [perl.com] for "demoroniser" (yes, the Brit / Aussie spelling.
Oooh, this is going to cost me karma as trolling, but...
How is this all-inclusive, single-source applications/OS environment any different than, say, Windows and Office?
You may be surprised at how much support you get. Disclaimer: I *use* KDE every day and I think it's great. I use lots of KDE mainline applications and lots of what I'd have to call "3rd party" applications. (though I probably use even more Gnome/GTK applications, running them from the KDE desktop.)
I have some experience compiling KDE out of CVS, have done a small amount of KDE/QT development and have participated in the KDE mailing list. My intention was to become a regular KDE contributor.
This was short lived, however, as I quickly found that the KDE mailing list, was being run as a sort of private fiefdom by this same David Faure. Any comments of the form "KDE mainline applications development should be opened up to more developers" i.e., not just be restricted to the main CVS tree, were met extremely aggressively. I didn't want to press the point because I thought "maybe he's right" and all the core applications *do* have to be tightly controlled by the core developers in order to have the highest possible quality.
But now, 6 months later, I know he's wrong. KDE development has slowed down *a lot*, and especially KDE core applications development. I've been putting up with the same bugs in kmail for almost a year now - it hasn't changed a bit, because kmail development has been inextricably bound into the main KDE CVS tree, and it's all oriented towards KDE 2 now. So if you're not compiling out of CVS, you're out of luck. The bottom line is that KDE users have to wait and wait and wait for bug fixes.
Now, I'm a developer and I should just be able to download the code, make the fixes and send them in. But in the case of kmail, you have to download practically the whole tree, and let me tell you, it isn't that easy to get it to build. I would have done that anyway, except for the agressive reception I got on the KDE mailing list. Anyway, I'd still have a problem with it, because suppose I put in a bunch of great features - It would still be months and months before anybody gets to use them, because they have to wait for the official release of KDE 2. One of the reasons developers develop for free is to see their stuff get used *immediately*, and this just doesn't happen with KDE development.
I did what open-source developers are famous for doing - I moved on to other projects that are more to my taste (e.g., Freenet [sourceforge.net]!!). Now, I'm not bitter - why should I be, I'm still getting KDE for free and it's still darn good - but I am concerned that a certain prima-donna mentality is hurting the KDE development, and because of that, hurting the entire open-source movement. I wonder how much of this is due to the fact that some fulltime, paid developers (Faure is one) don't feel compelled to play by the traditional, unspoken rules of the community. OK, I've said my piece, and I'm posting this without my +1 bonus because I'm not sure whether this is a flame or a troll or constructive criticism. --
I see how that would give you a problem, but mostly I was wondering why linuxuk.org.uk was using MS Word and/or MS SQL. Surely they aren't doing that, so what other circumstances could cause the '?' quote thing?
Right. Because the actual scripting model exists from the inside-out (WSH), which gives lamprey's like Gohip and macro-virii the foothold necessary to maniuplate common settings and turn the browser and emailer into an ad machine. Wicked stuff for people with short attention spans and a reflexive [OK] click response.:-) It seems like KDE is boldly going where Win9x has gone, without n levels of obfuscated templates between the developer and the toolkit, and is using the same kind of mechanisms/methodologies. They probably don't have much of a choice because if the technology favors the method then developers have to go with the flow...or reinvent the wheel differently; and reinventing the wheel takes time and money not readily available to people making free software.
Incorrect: only KApplication using DCOP needs the DCOP server. No developer is forced to use DCOP when they code a KDE application.
Now, have a look at which applications does logically require DCOP: mainly KOffice and Konqueror. Lemme say you, if you want to use KOffice, it will be stupid not to install KDE. If you want no to have KDE, it will be stupid to want to use KOffice. The same is true for mainly all other DCOP-based KApps.
Now, a good programmer can put some test in his code to check at launch time if a DCOP server is available and then act regardingly, it could then just display a massage box warning that desktop communication is not available, and avoid each call to DCOP stuff.
Finally, you said "It was bad enough when the kdelibs were required". Yes. Sure. All program should be statically linked with all libraries it needs, and never call.so files. This allow a very powerful and unnecessary bloat in the RAM. I think that you'd also want to have X working withou xlibs, gnome working without gnomelibs, and your whole system working without (g)libc. Go to/lib,/usr/lib, and/usr/local/lib and do some rm -rf *, then reboot your system.
We went out last night to the party Linux Format threw, and we got chatting to Phil Hands from Debian. He was saying that KDE doesn't have a licence. what are your thoughts on this?
That's odd because KDE definitely has a licence; LGPL for most of libraries and GPL for most of the applications, and Qt indeed has a different licence; it is QPL. It isn't much different for home users or for even users in a business environment. Troll Tech have put a lot of time and effort into producing a high quality toolkit and in all fairness they should get paid if you sell an application that uses Qt.
The only thing I see here is two things: I'm pretty sure P. Hands didn't say that, but as I can't prove this, I'll just shut up on this. I think the question is as bad as the answer. KDE does have a license, of course (we know things are useless without a license, ie, they are the most non-free pieces of software). Saying it doesn't have one is not very accurate. This question tried to get an opinion about the KDE licensing wrt the recent $3K someone offered. It's a pitty the question is dismissed with "yes, we have LGPL, GPL and tk under QPL". We know that, I really want to know what the KDE people have to say about the accusations of GPL violation in KDE&family. Sigh.
And, should I comment about his lattest statement? It would be really funny to see the QPL changed so people had to pay Troll Tech for selling Qt apps. I guess that would benefit Gtk so much...
Not really. I work for mandrakesoft now, and quite a lot of people here prefere gnome to KDE (although we started as RedHat+KDE). The reason for KDE in Germany is that SuSe dominates the markt there, and they have been pro-KDE from the very beginning...
Running any program is a security risk, no matter how many times it was audited. When humans are involved, flaws are bound to exist. Therefore I suggest you remove all software included your BIOS and only turn on your computer and look at a blank screen. This will be a pretty secure system.
On the other hand if you want to run something, do what it requires. Be it install an operating system, some libs, whatever. If all we're doing is bitching about huge code bases, I can say the Linux kernel, glibc, the bsds, etc are too big, programmed by humans, and insecure, therefore they are equal to anything by Microsoft.
I am very sorry to hear that:( You wish that KDE core applications were developed out of CVS ? With the changes that happen in kdelibs quite often - to build a better API, and because of all the new stuff in KDE 2.0 (KParts, DCOP, XMLGUI, etc.) - this is almost impossible. What you gain from having apps in CVS is that they can be fixed by many people, especially those who change kdelibs. Out of CVS, it's up to the author to fix his app to cope with the changes, and _that_ can be piss him off a lot, obviously. This is not about "controlling", this is effectively about helping....
To answer the more personal comments, KDE is definitely nobody's fiefdom. My answer (which I can't remember) may have sounded not nice and I apologize for that, but the idea of developing core apps out of CVS really doesn't make much sense. You want to "open more" the development by making it private ? About KDE 2 apps that are not in CVS : from 1.91 onwards, they should be able to compile and run on top of KDE. The API shouldn't change anymore. No need to wait for the official 2.0. Finally, I'm not a paid fulltime developer... yet.
I really really hope the KDE folks know what they are doing. Konqueror.. and security.. I'm sure its secure as of now. As long as it only SHOWS things as they are. I just hope they're intelligent enough to *never ever* allow scripting in Konqueror. When it comes to luser-friendlyness, linux is still behind windows. And it will always be. They *must not* go into the same trap as MS has done, and include scripting capabilities - even though it may be tempting.
Konqueror is very modular, if I've understood it correctly. That means its possible to write extensions (easily). *That* make me worry. What if someone gets this great idea of showing emails, and executing the javascript included in them? What when Linux get visualbasic script support - what happens when some idiot thinks "Hey! What a great idea! Email reader with js and vbs execution support! It's User Friendly!"(bad pun intended).
As long as linux programmers avoid these pitfalls, we should all be safe.
-- "Rune Kristian Viken" - arcade@kvine-nospam.sdal.com - arcade@efnet
> "never ever allow scripting in konqueror" Why ? What's wrong with scripting "open a new window, open that URL inside it", and even "type my login and password for that site" ? As long as _I_ run the script, where's the problem ?
For too many people, scripting == security risk, without really thinking about it. No. But automatically launching untrusted scripts (such as those received by e-mail) is a security risk, yes. That's why kmail doesn't launch scripts.
Oh come on! What do you think ORBit does in Gnome? Face it, for an *integrated* desktop environment, you have got to have some sort of process that coordinates things. You can make it ugly, you can make it elegant, but you have got to have it there. I frankly see no difference between the DCOPServer and ORBit (except the fact that one does Corba and the other uses libICE). I challenge you to come up with *any* software scheme that has many interoperating parts that does not have something that does not fulfill this task.
Since about April, I have been making some of the cvs snapshots of Kde 2.0 beta. You might be interested in my impression, which differs considerably from some recent reviews, which are glowing endorsements.
I really much prefer Kde 1.x to either Gnome or the new Kde 2.0 so far. It's fast, sleek (in my opinion heavy pixmapped themes and large, blocky icons are NOT visually pleasing) and very stable. The browser (kfm) has some limitations but is more than enough for most browsing and downloading and email with KMail. I much prefer it to Netscape, although I must admit that Netscape 4.73 is a very stable and excellent, full featured browser. It a much better browser than MSIE 5.
Kde 2.0 is very ambitious in trying to make everything a "part" or component like the MS desktop. Believe me, this project, with KOffice, is every bit as ambitious as the entire Windows desktop and application framework, including COM and DCOM. and MS Office to boot. It took MS 10 years to get that far. Kde is trying to do the same but do it right in 3 to 4 years (since Kde 1 first became usable as a desktop).
In my opinion, success so far is only partial. For example, Konqueror is very fast and smooth, really a pleasure to use. It overcomes all the limitations of the 1.x kfm. But it crashes far too often. I feel that these crashes are due to the tremendous complexity of the code, particulary the interaction between DCOP, Kparts and the parsing of urls. Until this is stabliized, I don't recommend using Konqueror except for testing. It seems that when one thing is fixed, then something else breaks in the core libraries and base system. The build process is also nowhere near finalized, and you can expect some things not to compile from day to day in what is on cvs.
Really, I cannot emphasize enough how ambitious this work is, and that it is several orders of magnitude beyond what is currently available for unix. But, can they pull it off? The code base is much larger than Kde 1.x, although the executable code size is not. Even though the onerous burden of Corba was pretty much abandoned as a design decision, I feel that the degree of complexity may still be too great. All I can ask is that more "features" not be added until the current state of Kde 2.0 beta is stabilized and any temporary hacks used to fix noticeable bugs or speed up some operations are worked into the design in a consistent way that retains the beauty of the design.
Right now I am using Gnome 1.2 as my primary desktop so I can work on some qt 2.x code without conflict with the qt 1.x based kde 1.2. I can say that Gnome 1.2 is super-stable. It has never crashed. But it is very much lacking. It is not fun to use. Keyboard support is almost entirely absent, and many apps are in a primitive state of development. The "design" from a visual point of view is not pleasing to me - the proportions are top-heavy and many things can't be configured which are easily configured with Kde. This kind of configuration means more than "themes". For example, with GMC one cannot hide the large, blocky and space hogging icons on the taskbar and there seem to be no keyboard or menu equivalents. Of course Gnome is utterly dependent upon Netscape, a commercial product, for web access and browsing. Talk about free software. Gnome would be totally unusable without linkage (in a lose sense) to this very proprietary and closed app. Still, I am pleased that Gnome is now stable and hope that it continues to improve, along with Kde. We need choice! Maybe Gnome could incorporate the really free components of Mozilla into a new browser.
I guess it is my fault for being impatient for Kde 2 to arrive in a more stable state. But I think it will be worth waiting for!
Any comments by David Faure, if he is still lurking around, on these issues will be appreciated.
Plesae excuse my ignorance on the finer distinctions of window managers and environments, but wouldn't the Linux/UNIX community be better served if these apps ran in any X windows environment?
What is I want to use an app while I'm at my Solaris machine, or at an NT machine?
What if I'm enamored with an older windowmanager?
What if I have a hacked i-opener and I don't want to run kde on it?
There was an interesting point brought up in one of the comments (by Dave Smith [mailto]) on the interview link.
Shouldn't some restrictions be put on open scripting like this? We don't want a microsoft problem coming and haunting KDE too... Perhaps you can have a database of signed scripts that are able to use the KDE system or somein...
Here's what facilitated that comment:
What elements are mainly shared between all applications in this framework?
Scripting is, in fact scripting is shared between all KDE applications. How you talk to a KDE application and how you script it is a generic system. Among all of the KOffice applications you can also find a lot of common dialogs, and a lot of common tools that are in the framework that allows the user to see a very integrated office suite, not a different set of applications. It's all about having a full, complete office suite which has the same elements for everything.
If you've got problems with the KDE development model, nobody's stopping you from forking the KMail code and integrating your changes. Personally, I think David's an awesome guy- he knows what's going on with the code, he has some really brilliant ideas about how to implement features in Konqueror, and he's never been rude that I've seen. Brusque, maybe, and possibly curt, but never rude. And about him getting paid- he gets money to develop KDE, sure, but it's not a fulltime job for him (yet), as he replied already.
It's not his fiefdom nor is it the fiefdom of any of the other main developers- but there is a definite centralized development model. IMHO, I have to say that this has worked very well. The KDE project consistently puts out good code and good programs.
As far as KMail goes, you might want to check out kdenonbeta/kmail2 and help out with that. It can use some new developers, and it's definitely not release status yet.
Yes, this is offtopic, but this interview was a bridge too far. David Faure may not speak English as a first language, but didn't the interviewer either?
"I run the CVS version of KDE2, and I remember seeing an IRC thread in one of packages regarding not running a mail composer within Konqueror, but a mail viewer, so will the composing of documents mainly be launched by an external editor? "
"In terms of KOffice, what parts of KOffice are you working on? "
These are barely grammatical, and very hard to read, certainly not reviewed by whoever wrote them. We all get fed up with web sites that abuse HTML, so why should be stand for abusing English just as badly.
If anyone from linuxuk.co.uk is reading, I hereby volunteer for a job as a sub-editor so no-one else has to read this kind of stuff...
I'm surprised to see the '?' replacing the apostrophe character in this interview - isn't that normally the case with MS web publishing tools and their "smart" quotes? I'm viewing this with Netscape 4.7 on HP-UX.
Well, not much to say. Yes, konqueror in the last release crashed quite often - but try 1.91 in a few days, it will be already *a lot* better. We are working on stability, don't worry. And no, I don't believe the code base and the complexity makes it impossible to achieve. On the contrary, it's a lot cleaner than for instance kfm's code, which was just a PITA. Some things can NOT be fixed in kfm, due to its design. With konqueror, however, everything is possible, since the design is a *lot* cleaner.
BTW: about Qt 2.x / Qt 1.x: you can have both on the same system and still use KDE 1.x:)
Although I use KDE (but at work, I'm having fun with the latest spidermonkey release of Gnome), and would like a better browser, I don't want it at the cost of some script kiddie being able to script through the pipes joining KOffice to Konquerer. I've even dusted off Netscape 4.7 and use it because IE is a scripting paradise unless you take steps to lobotomize scripting(and render the browser an intert piece of crap). Maybe someone who has intimate experience with this scripting in KDE could inject some measure of confidence about how well the sandbox for this scripting has been made...otherwise, it's just another vector for infection and exploitation.
KDE doesn't tie you to a specific window manager. Early versions of KDE2.0 (around KRASH) simply called KDE support "we'll make a big window, title it the Desktop, and you know not to put a border around it." I believe the support has gotten better for window hints, etc., but you can still run an outside window manager. You just won't get full hints - like running another window manger (such as fvwm) and GNOME.
You're right in what you say, but you don't see the point. The major difference is about what is used for embedding components. The CORBA solution (be it Mico or ORBit) means that the component is provided by another process, whereas with KParts (the current KDE solution), the component is a shared library, loaded in process, which turns out to be a lot more stable. We didn't replace CORBA(MICO) with DCOP, we replaced it with KParts. DCOP is mainly used for IPC and scripting - and yes, I agree with you that this is needed, whichever solution.
You *can* run KDE apps under GNOME, or under any standalone window manager. You just need the KDE (including QT) libraries. There are some minor drawbacks: 1) inconsistency in look and feel between QT / GTK+ apps... I find it jarring, but you may not. 2) if you use GNOME and run a KDE app, you have to have two complete widget sets in memory... which may or may not be an issue, depending on what you do and how much RAM you have, etc.
I have been working on a project called Kafka and one thing that struck me about KDE that was impressive was there are standardised actions and routines in the libraries to keep application looking and working in an integrated fashion. Will KDE still follow this path or will it try to move out like GNOME has and try to accommodate everything?
In KDE we have always tried as much as possible to put things in the libraries so that writing a application is as easy as possible, and it means that navigation is consistent. There is still the freedom to do something completely different; you can put the file menu on the right, and you can change anything you want, but we make it simple if you want to make the application compliant with what people are used to in terms of user interface.
I don't really see how KDE and Gnome are different in terms of providing the API's for standardizing user interfaces, while allowing complete freedom to do things like change the toolbar icons, put the menus elsewhere, etc. What is meant by the "standardization" that KDE offers vs. Gnome "accomodating everything." Of all the differences between Gnome and KDE, this isn't one that I've really noticed. Any thoughts? ----
Not everyone thinks the same way as the usual/. crowd about these technologies. I happen to believe that Microsoft had (I'm not going to argue over whether they thought of them first) quite a few good ideas about integration of components, COM, ActiveX, scripting, but horribly botched the implementation (as always).
What the KDE project is the vision of a true integrated desktop in a clean, non-bloated implementation. I've run KDE1.89 and 1.9, as well as from CVS on a variety of systems, and it never seemed slow or bloated. Instead, it was clean and easy to use. I enjoyed using it because of the integration, unlike Windows where I hate using it because of the botched implementation.
To put it simply, KDE2.0 is the ideas of integration and usability done right. It's a different beast than KDE1.1. KDE2.0 is actually more usable than windows, and easier to program, while implementing the ideas that Microsoft used.
but wouldn't the Linux/UNIX community be better served if these apps ran in any X windows environment?
They do. Let's take your questions one by one:
What is I want to use an app while I'm at my Solaris machine, or at an NT machine?
Well, you can run KDE on BSD, Solaris and quite a few other OSes. NT is not included - nobody has really expressed an interest, but I'm not sure how large the technical hurdles would be. That means running native on the system (Note that it's periodically broken for OSes other than the core developers during alpha/beta testing. The core devels use Linux, BSD and something else I can't remember, but might be Solaris).
What if I'm enamored with an older windowmanager?
Use it. KWM is nice, and has *quite* a few new features, many of which are integrated into KDE, but are not necessary. I've run KDE 1.x with E with no problem, and 2.0 is supposed to work (dunno what features you lose, though. Themeability woul be my first guess).
What if I have a hacked i-opener and I don't want to run kde on it?
X Windows is network transparent. There are a handful of NFS and "running apps remotely" bugs that cross the lists. They are addressed as valid concerns, so I would imagine that they will be addressed. The lightweight KDE communication mechanism DCOP (used instead of Corba where appropriate, although Corba is still used where *it* makes more sense (like the multimedia subsystem)) operates over a X Windows standard communication system. To quote the developer site: "DCOP is built on top of the Inter Client Exchange (ICE) protocol, which comes standard as a part of X11R6 and later."
In short, it is an X system that is portable to many platforms. Some of this I've verified myself, the rest I know from the mailing lists, where KDE 2.0 beta is already in use.
David Faure may not speak English as a first language, but didn't the interviewer either?
"I run the CVS version of KDE2, and I remember seeing an IRC thread in one of packages regarding not running a mail composer within Konqueror, but a mail viewer, so will the composing of documents mainly be launched by an external viewer?"
I agree this is hard to read. I got the gist of it, but it took a couple of rereadings, and a non-English speaker may not have understood.
"In terms of KOffice, what parts of KOffice are you working on?"
I think that's a perfectly acceptable colloquialism. Ok, it's got redundant information in, but so has "look at that pen of David's" (double genitive).
I hereby volunteer for a job as a sub-editor so no-one else has to read this kind of stuff...
If you get the job, I'm sure you'll do a great job of weeding out ambiguous, messy collections of words like the first example. But please don't kill off valid colloquial phrases! There's more to English than writing formal letters, and there's no reason why linuxuk.co.uk articles should have to be formal.
"but wouldn't the Linux/UNIX community be better served if these apps ran in any X windows environment?"
but they do run in any window manager. i use olvwm and run some of the kde apps/games. they run fine. and without the "desktop environment" there's less overhead which works much better on slower machines.
No, I disagree. The scripting system should not be crippled. We should look specifically at how ILOVEYOU (and Melissa et al) worked.
First, it was stupid users, apparently. I've heard that MS Outlook will automatically run the script if the e-mail is viewed in the preview panel; I've heard some Outlook users say that that's false. I've used Outlook all of about twice in my life, so I can't say. If the script is run automatically from the preview panel, then I would say that's bad engineering (and hence bad Microsoft), but it sounds like they've at least fixed the problem.
There's also the question as to whether scripts should be executed automatically by default (e.g. when you double-click or otherwise "activate" them). Personally, I don't care. Turn it off by default, sure, sounds good.
Most importantly, though, I think the users need more control over their scripting. This is one area where KDE and Gnome have been falling down (gently). By far the most annoying thing about Windows and MacOS is that the user has almost no control, either over the OS or the applications. A few years ago I would have almost literally killed for the ability to put some scripting into Netscape that said "yes, Mr. Javascript, you can open a new window, but not when I am viewing a page from Geocities". You can argue that I could have just disabled Javascript (which I did), or made something similar to Junkbuster which would automatically parse and change all Javascript, thereby eliminating the pop-up windows (hah!), but this would have been such a nice solution.
So anyway, back on topic. I think the user should be able to do some metascripting, scripting exactly how the script should behave in certain situations. This is not the same as clicking checkboxes (especially vague ones like "Run untrusted scripts"); I want some actual *control*. If KDE (and Gnome, but they're not important in this article) do not do this, then they're just going to end up as Yet Another Inferior MacOS. I think it's time for Something Actually Different From MacOS.
"KDE is no longer just a library that you need to run. While there are kde libraries, there is also a dcop server that handles interconnections between programs. Running a KDE2.0 program requires you to have an active instance of the dcop server, or else the application will bomb."
You put this in a list of features, but it is a bug, AFAIAC. It was bad enough when the kdelibs were required, but now I have to run (and probably configure) servers? Total nonsense.
For instance, let's say I don't use KDE. But occasionally I find an interesting "kprogram" on freshmeat. For KDE 1.x, I could have the libs installed and then that k-l33t prog would work OK. Now even that (extremely lame) option is closed to me.
And don't bother answering "but you only have to do it one time" because that "one time" is the first time--meaning it will never be done. The only other alternative is for distros to auto-start this dcop crap for ALL installations--a spectre so grim that it doesn't bear thinking about.
By adding this one "feature" KDE is isolating itself from the rest of the software world. -- Wanna hook MAPI clients to your Tru64/AIX/Linux server?
The problem with Microsoft is not merely having an operating system and office suite. It is using secret APIs and what-not against others. Write a couple lines of code in Windows 3.x and voila, any non-MSDOS systems cannot run it. Do you want to write a competing office suite, web browser, etc? Well you have two paths, work around what APIs Microsoft has given out to the public and bugs in those API docs (MS Office obviously has access to all APIs). Or you can not make the program.
Now, with Linux, KDE, GNOME, etc, this freakin monopoly argument holds exactly zero credibility. First, you have the source. There are no secret APIs to give one person an advantage over another. There is no necessary bribes to find out secret API calls. Second, since this is free software, there is no financial problems with software bundling and the like. We all saw what happens when you ask computer manufacturer X to remove Office/Windows. If they want to use Corel or some other office suite/operating system, you pay for the Microsoft stuff with the computer and again with your chosen operating system/office suite. Third, and finally, Konqi/Tux are much cuter than that Windows flag logo thing.
Now, if you just hate KDE because it's KDE, be outright with it. Do not invent stupid arguments like "It is/might be Microsoft-esque." No need to use that jump-to-conclusions floor mat.
There has been discussion lately on the KDE developer mailing lists about how to avoid script-based worms, trojan horses, and viruses. Thankfully, you already have the Unix security model to protect you, so don't run things as root.
Among the options proposed include a sandbox for running scripts to see what they do, before running them for real, and not allowing certain operations from untrusted scripts.
Really, though, the scripting in KDE does not raise any additional security issues than just running scripts/binaries that were sent to you in email or that you downloaded off the net.
It has also been unofficially decided that the default for automatically running scripts/macros will be set to "off" so the user will have to set it to on for scripts to run automatically.
Anyone else who has been reading the mailing lists care to comment?
From your post, there is one thing funny. And even ironical: "Isn't there a parallel with Office's non-standard dialogs and widgets, and KDE's different-from-other-GUIs dialogs and widgets". I can't retrieve this document, so there is no link, but I remember reading an archived version of Matthias Ettrich's call for programmer: one of his argument for launching the KDE project (btw, that was meaning "Kool Desktop Environment". Ettrich thought of this as a codename, because it was a little sucking, but the name stayed and the official version is that "K" doesn't mean anything) was that all widgets and dialogboxes ware behaving differently. So, he says, we need a true Desktop Environment rather than a mere Windows Manager with an united look and feel, a lone widget library (instead of Tk+Motif+Athena... There wasn't GTK+ at this time). KDE was the first project to try to use a "standard".
About the rest of your question: embedding application is aimed at reducing bloatware. With OLE2, each application (says Word for example) contain subsets of the other (says Excel). When you edit a spreadsheet in Word, it's not Excel which is used, but a small Excel clone located inside word. So Office get bloated with redundant stuff. This will be avoided in KOffice. There isn't more risk in a script than in a normal program. The risk is how it is managed.
Finally, MS was not condamned for integrating closely things in Windows, but rather for using proprietary tricks (non documented API functions, of functions non mentionned at all) to prevent their competitor from releasing product with the same integration. Hey, you prefer to use something that is seamlessly integrated or that is just an alien program ? KDE being free software (despite what says some trolls), this I-hide-my-secret behavior is impossible. And, trust me, if someone in, says, Gnome, decided to code some bridges to have Gnome Office integrated with KDE as transparently as KOffice, or vice-versa (KO for Gnome) with Bonobo/DCOP-etc translators, the KDE team will applause and thanks. In the same situation, Microsoft will modify its standard to have the bridges not work anymore.
And what package is the dcop server executable in? Have you ever looked? That's right, kdelibs. The entire kdelibs package is meant to be what's necessary to run applications. It includes libraries, of course, and a few more things that are necessary.
So, YES kdelibs is sufficient to run KDE2 apps for those that do not wish to use such a beautiful environment.:)
There's also the question as to whether scripts should be executed automatically by default (e.g. when you double-click or otherwise "activate" them). Personally, I don't care. Turn it off by default, sure, sounds good.
Or you could ask the user for confirmation when (s)he tries to run a script or program from the email client.
Donate Food for Free - http://www.thehungersite.com
Replace "KDE" with "BSD". Nobody is obliged to use your patches. I for one, wouldn't mind having routine and stable releases, instead of a new micro revision every day.
What worries me about KDE is that it has the potential to become a windows. Do we really want ONE unified X manager? yes it makes it easier to develop, but we dont want a single solid LINUX. ick! that is one of the reasons i use linux, there are many different flavors and aspects to it.
a better solution would be to have interfaces where you just load the library for the features you want.... which we pretty much have, but could have in a better way. for example, right now you can run KDE apps in enlightenment, but how about a totally minimalist X program that is not so bloated? i am thinking of moving to sawmill unsure tho...
- The GPL prohibits you from placing any greater restrictions on software you distribute.
- The QPL places such restrictions (you must give Troll a copy).
- Therefore, the licenses are incompatible.
No one is suggesting that Qt isn't free. What we're saying is that your license is logically inconsistent, and that therefore it's impossible to know what license you really want.
Essentially, you don't distribute KDE under the GPL. You *say* you do, but really you don't, because distributing KDE binaries would be illegal if so, and you haven't seemed to mind so far.
What Debian wants here is a statement of *what* your license really is.
Come on, which developer of a GPL application would have a problem with his stuff being distributed and used by people ? Stop thinking of the QPL as "the evil thing we want to avoid because it's not GPL" (or drop netscape...). Besides, there aren't really that many GPLed applications involved. This license nonsense really makes me sick. Talk about fanatism...
About the second comment: I love being quoted out of context. Re-read what I meant by that : the fact that it's interpretation is really subjective, since not two people interpret it the same way. So this "incompatibility" is far from being proven. What makes you think there is one ? Because RMS says so ? Sometimes I feel like there is a dictator, "free" software or not...
You DO realize that X is a server, right ? Then why do you run it ? The DCOP server is a server for inter-process communication, but only between apps running under the same X server. Besides, DCOP is implemented of top of ICE, which what is used for session management everywhere now. So there is no additionnal security risk than an X server, which you already do.
IANAL, but here's my opinion: The GPL enforces that what if you base your software on a GPL application, you release it as GPL. The QPL enforces the same thing, you must open-source the software based on Qt [unless you buy a license]. So this is not an additional restriction ! "Give trolltech a copy": this is the case for any open source software ! You don't only give a copy to some people, you give a copy to the whole world !
I really hate this license crap. Can't you see that both the GPL and the QPL aim to the same thing ? They both enforce that derived work is also free and opensource.
But nobody would force you to use every new micro revision. The project leaders can designate a particular release the stable one every so often, but people who want bug fixes quickly (I've waited for kmail bug fixes too, and submitted one of my own) can still get them from the main distribution.
I don't see the advantage of not releasing often - isn't that how you encourage bug fixes and developer mindshare in your project? Users who want stable, routine releases can wait for major revision numbers if they want, but that isn't really the way to encourage developers, IMHO.
I quote D. Faure: 'Troll Tech have put a lot of time and effort into producing a high quality toolkit and in all fairness they should get paid if you sell an application that uses Qt.'
Please, everytime I hear KDE developers give this line I want to cry out. What kind of reasoning is that? I that was the logic we should follow with Linux, then shouldn't the kernel developers get paid, the Xfree86 developers, the gtk developers, the GNOME developers, or for that matter the KDE developers and so on. Or is the Troll Tech people the only ones who contribute time and effort.
Face it the QPL is a crappy license which only aim was to be 'free' enough to kill of the Harmony project.
So KDE developers get a grip, it is your right to develop your software with quasi-free libraries, but stop pretenting that the Linux community should be gratefull to Troll Tech, they are just a bunch of developers like everybody else, but developers using a license which is more in the spirit of Microsoft than the Linux community.
Is this different ? Just as much as "free" and "commercial" is different. Microsoft bundles things together to sell them all and make sure you don't go and see other vendors. KDE _offers_ many bundled components, but that doesn't mean you have to use them all if you don't want to. Example: if you have the "kview" image viewer, konqueror can use it to show images embedded in a window. Fine. If you don't have it, it will launch your preferred image viewer instead, be it xv, ee, gimp or vi.:) Face it: the biggest need on Linux at the moment, in terms of software, is for a free and full-featured office suite. KOffice is on its way to answer exactly that. Should we not work on KOffice because Microsoft used MSOffice to tie people to Windows ? Nonsense.
I indeed withdraw that answer - to a very unexpected question, which caught me by surprised (this was an oral interview, not a written one, and I misunderstood the question). Please read the thread "pathetic licensing answer".
Oh come on...X is a server, you have truetype servers and clipboard servers, there are myriad little servers running all over the place in a typical *nix configuration...
That would be because everything it's displaying is a borrowed viewer being pulled in through kparts - so it needs the integration glue in place to do that.
You're kidding me! That will bring down a windows box?
Not all of them. But certainly some that I've seen (and that's Win32 versions, not old 3.x versions). Dunno exactly what you need to do it, but I'd try IE3 and Win 9x at a guess.
I understand what you're saying, an interesting argument which implies that the GPL might not hold up in court for libraries (depending on the particular court).
But the problem here is the other way round - some people wrote GPLed apps, and the KDE people have linked derivative works to Qt. Troll may not be able to control what links to QT, but they certainly have legal control over how the library is distributed. The way they license this control is not free enough to be compatible with the GPL which the original app was placed under.
Actually, no. KDE2.0 is tied to dcop, and you need to run the dcop server before you can run any KDE2.0 application. It's quite a change from KDE1.x, but it improves the integration of the desktop. (It's sort of hard to have apps working together without a glue application, don't you think?)
If you've got problems with the KDE development model, nobody's stopping you from forking the KMail code and integrating your changes.
Believe me, I seriously considered that and would have done it if I had not gotten fully committed to other projects. The need to spend several hours downloading the full CVS tree (analog modem) and fiddling with the build process was the final straw. (I've since done that, and it did in fact take a lot of time.) But why is it necessary to solve the problem by forcing a fork? Why not address the problem in the first place by having a well-defined fork a la the stable-vs-development strategy of the Linux kernel?
Personally, I think David's an awesome guy- he knows what's going on with the code, he has some really brilliant ideas about how to implement features in Konqueror, and he's never been rude that I've seen.
I must have caught him on a bad day then, and what I've seen recently supports your opinion. I still I think I'm right about the question I raised and that KDE app source code needs to be more accessible to developers outside the inner circle. It's a real problem that bugs aren't getting fixed in the stable releases and it's a problem that isn't hard to fix.
Please note that none of what I say reflects in any way on the KDE 2 development - I think that's working great, and there's fundamental progress going on there, particularly in object embedding techniques. And the attention to detail and useability in the KDE development has always impressed me. No doubt such things are in part due to David's skills. --
Lets work on just one thing at a time. Source first, licence after. As long as I can accessthe source code, I am willing to temporarily compromiseon their licence issues. If they go M$ nature then I'll move on to the next alternitive. thats how I arrived at the glory of linux anyways... It dosen't need to be about control, or profits, or brand loyalty, thats yesterday. Lets trancend this argument and tackle the real issue. Its about freedom thru code availability. mozilla build m14, hour 52, no faults found....
Yes, I second that. I think it is time that the Open Source community started concentrating its efforts in order to develop a product superior to Windows, instead of continuously playing catchup with a large number of half-finished programs.
KDE is a disease that should go away. The developers are showing their contempt to users'/programmers' rights by choosing a non-free license. And please, do not tell me that it has changed and it is free now. If they really want a free license there are plenty available: GPL, LGPL, artistic, BSD. Why do they have to keep creating problem by having their own little incompatible license?
I do not understand/.'s continuing support for KDE, at the expense of GNOME which may not be as mature yet, but is the only hope of the Open Source community.
Well, I don't want to detract any more from the otherwise good stuff in the interview than I already have. Dave, if you want to, E-mail me, and we can discuss our differences. If not, fine; and good luck with Konquerer.
I will say a few things:
I don't particularly like the situation with the licensing, either. But I can't do anything about it. KDE has chosen to give an internally inconsistent license to its users; it must accept the resulting confusion and missed opportunity that results.
We ridicule programmers who dismiss usability; indeed, KDE is about bucking that trend. Usability implies that a project listens to its users and assists them. If there is a usability issue with your license, why must it be consistently ignored? If incomplete bug fixes are not to be accepted in code, why should they be accepted in licensing?
Licensing matters. One would think that the recent incidents with the CyberPatrol crack, DeCSS, UCITA, Napster, and MP3.com would cause people to realize this.
KDE is looking better and better. I think my next computer is going to run linux and KDE. I think i'll give VMware a shot to see if i can run 98 in a window for those couple apps i just can't avoid (like VC6 for work, etc...)
You're completely wrong. I run all my KDE apps on Blackbox. The apps are wm independent but only kdelibs dependent. In the same way you can run GNOME apps on kwm if you have all the libraries in place.
Speaking for Germany KDE is the quasi-default here. If you go into well-equipped bookstores you can choose among 12 different titles dealing with KDE which are already present there. With much luck you'll find 2 books about Gnome there. In addition in most books covering linux you'll also mostly find KDE being dealt with when it comes to the desktop.
And if you read the mainstream-pc-magazines you'll also discover that KDE is being mentioned as "the" desktop for Linux there.
-- snip --- I think www.kde.org/technology.html (mosfet's) speaks quite well about this. If not, my take on it is:
- stability (that's the main one. If you used pre-canossa KOffice, you know what a real stability problem with embedding is). - wrong approach - starting servers for embedded components, leads to servers left running after you exit the main app (ok, that's a bug, but it shows wrong design IMHO. What should be one application was a set of distributed ones, hence this kind of problems). - memory usage and slow performances - but people associate this with MICO more than with CORBA in general. Well, we'll never really know, but in any case we had to do something about this. - the interoperability we were supposed to benefit from it was null. gnome uses an orbit-specific authentication mechanism which made interoperability as difficult as it is now, and nobody ever had any interest in making the CORBA things in gnome and KDE work together anyway.
In short, I always say: we had all the disadvantages of distributed computing (stability, reliability, performance issues) for something that is not distributed, in 99% of the cases - a desktop is local, most of the time.
But it was fun, writing sources that nobody could read afterwards:/. CORBA programming is like perl. Write once, read never;-)
Nah, you got that wrong. You'll still be able to download and run a KDE 2.x app even if you don't have the rest of KDE 2. dcopserver is auto-started if it's needed and it's not running, so you won't even see that happening. And it certainly requires no configuration.
Ask some true lawyers (not some license-crack-pots from debian) and you'll get the answer that permission to link against QT is implicitly given by just creating code that is obviously meant to be linked against QT.
This is true for stuff written *for* KDE. But for pre-existing GPLed stuff which the KDE people borrowed, there is no such implicit permission.
Please don't dismiss people as "license crackpots" unless you are sure you understand what they are saying.
KDE isn't a "Linux" thing.. nor is GNOME for that matter. Precompiled, ready-to-run versions of KDE are available for several Unix variants including BSD, and it's long been available for Solaris as a one-step-install ".pkg" package.
As for running on NT.. NT isn't a Unix-like OS, so the entirety of KDE running on Win32 is fairly unlikely any time soon, even with the help of the porting and runtime libraries from companies like MKS. Windows isn't Unix, MacOS isn't Windows, and your refirgerator isn't a houseplant.
The Qt toolkit, on which KDE is built, however, is available for Windows--but not for free. Several of the apps in the KDE family are indeed available for Windows, and some others could easily be made available for it. But not so for apps that take advantage of the integration that the KDE layer provides. In other words, one might be able to port individual KOffice applications to Windows pretty easily, but they wouldn't integrate with each other, because the integration requires the KDE infrastructure underneath.
Similarly, the GTK+ toolkit, the basis of GNOME's UI, is also available under Windows. This has allowed GIMP to be ported to Windows; just don't expect it to communicate cleanly with other WIndows GTK+ apps like Mozilla without special work.
But there IS a KDE_1_1_STABLE branch in CVS! It's just that nobody used it after releasing 1.1.2 (besides some bug fixing).
Close, but no cigar. It has to be: KDE_1_2_STABLE CVS, and it has to be complemented with the individual apps also posted in simple tgz format. In addition, the individual apps in, for example, kdenetwork, should be available as individual tgz's. Why should you have to download and build 6 apps when you're only interested in one?
There's no particular difficulty or excessive work in accomplishing it: a script can simply be written to extract and zip the necessary branches of CVS, and to create the necessary new CVS trees (so development of the stable apps can proceed independently from the development tree, the way it does in the Linux kernel). The script just needs to be run *once* following each stable release.
The only thing I see getting in the way is core developers are not, at this point, willing to recognize that non-core developers have a big barrier to getting started - much bigger than most other open-source apps. Even mozilla is easier to download and compile than KDE, let alone individual KDE core apps which have to be dug out with considerable effort from a rather indimidating CVS tree. --
Re:non-free (Score:1)
Conscience is the inner voice which warns us that someone may be looking.
Security (Score:1)
--
Wanna hook MAPI clients to your Tru64/AIX/Linux server?
Is this different from Microsoft? (Score:4)
How is this all-inclusive, single-source applications/OS environment any different than, say, Windows and Office?
I know, I know: it's not monopolistic, you are able to run other GUIs, it isn't tied into the OS, etcetera.
On the other hand, there are a *LOT* of Linux supporters who really get off on flaming Microsoft for packaging Office with Windows, for the way Office buggers with the GUI, for their scripting support, for bloatware, and so on.
All accusations that could be made toward KDE, and probably Gnome. Like, when Linux finally rules the common man's desktop, is an installation package that by default installs KDE and KOffice really any different than what MS does? Isn't there a parallel with Office's non-standard dialogs and widgets, and KDE's different-from-other-GUIs dialogs and widgets (there being a complete lack of standard for Linux GUIs)? Is KDE scripting going to be any less a security risk than VBA? Isn't X-Windows+KDE+KOffice not abhorrently oversized, just like Office?
Yes, I'm fully aware that it's a different situation... but not so different that it's not more than a little ironic that what KDE is praised for, MS is condemned for.
Ouch.
--
Re:Interesting point. (Score:2)
KMail Security....Lessons from ILUVYOU? -- There was some discussion on the kde-devel mailing list about KMail's security. As it looks KMail is far more secure than most of its counterparts. The testing and discussions continue.....
Re:All you need are the libraries (Score:1)
clarification (Score:1)
George
Re:depends which version of outlook you are using (Score:1)
no that's not true
believe me all those Outlook 2000 using monkeys who sent you a million ILOVEYOU's actually opened the script
MS is dumb sure, but not that dumb come on!
Re:Why is KDE So Much Better than GNOME? (Score:1)
The bus came by and I got on
That's when it all began
There was cowboy Neal
At the wheel
Of a bus to never-ever land
Why implement at all? (Score:1)
--
Wanna hook MAPI clients to your Tru64/AIX/Linux server?
Re:mmm. KDE... (Score:1)
Not just slightly offtopic;-)
Anyway, I used to run Win NT on VMware (not the latest version) in my last job. We had a license for Win Maple, but not for Linux, and sometimes I had to write some Maple stuff for teaching. Worked fine, and I certainly tried some other apps, and besides sound everything worked fine (haven't tried to fix that). When I have to use a Windows app, I like it to run in a X-Window, so I still can use my preferred system.
I will probably have to use the W98 partition of my current notebook under VMware in my new job, too. But I definitely would prefer using Wine to run them. Tried it recently again, and they did a great job since last. So, if the Win-apps you have to use run with Wine, forget about VMware, if not, it's a good 2nd choice.
try http://www.winehq.com/ [winehq.com]
echo $FAKEMAIL | sed s/soccer/football/ | sed s/" at "/@/
Re:Pathetic Licensing Answer (Score:1)
Yes, I wish I could rephrase my answer on that. I was totally surprised by the question - note that this was *before* the $3K thing, since the interview happened on June 1st.
I replied to "Why Qt uses the QPL and not the GPL", not really to "what to do about that alleged incompatibility" - sorry about that.
So, what do I have to say about that incompatibility ? That people should look at what they want/need. KDE is free and open-source, since it's GPL/LGPL. Qt is free and open-source, since it's QPL. Isn't that what you want to use ? Free and opensource software ?
RMS may find an incompatibility between those two licenses because of the way *he* interprets them, but remember that the GPL hasn't been tried in court yet, and interpretations vary. Why does everyone here (and elsewhere), dreaming of "freedom", blindly obey what RMS says about licenses ?
There _may_ be a slight incompability somewhere (I'm not a lawyer !), and I hope it will be solved at some point, but I don't see how that prevents debian from shipping KDE Qt !
Both are free to use, distribute, hack, etc. Isn't that what you want from free software ?
So stop the license crap, and benefit from all the free software that is available, including KDE, while lawyers tear their hairs off upon how to interpret a line in a license.
And I really wish people would stop talking about "non-free" for something that *is* free (in all meanings of the term).
Have fun with KDE ! 1.91 will be available in a matter of hours/days!This comment posted with konqueror [konqueror.org]
Re:Interesting point. (Score:1)
So forget all those talks about "viruses and other scripts that can wipe out your hard disk". All you can do is control a KDE application - and no, konqueror doesn't offer "rm
Re:Pathetic Licensing Answer (Score:1)
"script kiddie" (Score:1)
"script kiddies" dont WRITE anything. I keep seeing people on slashdot misuse this term to refer to anyone who does anything malicious and I'm sick of it. Forget the big hacker/cracker debate, don't misuse the term "script kiddie".
Re:Is this different from Microsoft? (Score:1)
I realize that this is not an easy problem. But there is a better solution than hiving off all development into the CVS and core team. In short, there should be a stable (1.1.2) and a development tree (1.x.x) just like the Linux kernel. The stable tree should actually be a number of trees, with each core app having its own tree. And also a zipped snapshot, so an interested developer can grab it and immediately compile it with no fuss or muss. Only after that the developer needs to put in the extra work to install cvsup and sync up to the latest tree, the mailing list, the maintainers, etc.
The stable trees can be maintained by a completely different set of developers - there will be no shortage of volunteers And so, no additional load on the core developers.
This is not about "controlling", this is effectively about helping....
Yes, of course, but sometimes the cure can be worse than the disease. Sometimes the first cure you think of isn't the best one, even if it seems to help.
To answer the more personal comments...
The personal comments were not necessary, and I apologize.
I was referring strictly to the mailing list.
I don't see how any of my remarks could be interpreted that way.
About KDE 2 apps that are not in CVS : from 1.91 onwards, they should be able to compile and run on top of KDE. The API shouldn't change anymore. No need to wait for the official 2.0.
That's nice, but the API is going to change again in the future (I hope! Ossification is bad) and the same problems will come up again.
Finally, I'm not a paid fulltime developer... yet.
Pardon me for misspeaking in that regard.
--
Holes are for digging, Libraries are for linking (Score:1)
In the same way, libraries are for linking. Dynamic libraries are for linking to ANYTHING, regardless of license. If a company, author or organization makes dynamic libraries available on a system, then ANY app may link to them. It may be different with hard linking libxxx.a libraries which are properly regarded as object files which become part of an application itself, rather than something linked to at run time.
I will link to any dynamic library on my system regardless of how it got there and regardless of artificial license restrictions and if somebody don't like it they can sue me. Hell, you can even link ANY windows app to the most proprietary closed MS libraries in existence. That's why there are there - to be linked to.
These objections to the Qt libs being linked to by GNU libs and apps are bogus. If GNU don't want its libs to be linked to, it should remove them from public distribution. It is as simple as that, regardless of legal contortions and trickery used to obfuscate common sense and prove the opposite. And, I don't think I would have much difficulty at all getting a jury to agree, if it came to that.
If free software and Linux in particular gets bogged down in these artifically created legal power struggles over purtiy, and it is nothing but a power struggle among distributions and companies hoping to profit from free software over turf and market share, then it can kiss its ass goodbye. Gnu, and free software generally, will win or lose the battle for the integrity of free software on general principles of free speech and intellectual freedom and fair use, not in nitpicking the fine print of licensing terms. That approach is unconvincing and defies common sense, whereas the other, broad-based approach is easily understandable and has integrity as a whole. Similarly, in the realm of software patents, restricting use of common altorithms and interface methods is theft. It is plain wrong, and anyone who writes code knows that. Gnu can never win its case by arguing the fine points of this or that particular patent, but must make the case for intellectual freedom and the (sacred) value of our common intellectual heritage convincingly, wholly and without excuse or unnessary rationalization. It is really a freedom of speech and freedom of association issue. If we are forced to communicate digitally, that freedom cannot be restricted or abridged with artificial licenses and patents. Even the average Joe Sixpack can understand that.
Holes are for digging, libs are for linking, and computers are for computing.
Re:Holes are for digging, Libraries are for linkin (Score:1)
What about getting a slashdot account, to post at 1 instead of 0 ?
(I'm posting this one so that people browsing at 1 wonder what I'm replying to
Re:Interesting point. (Score:1)
What about HTML 4, CSS, Java, Javascript ? :-)
konqueror does much more than kfm did.
See http://www.konqueror.org [konqueror.org] for more info.
Re:mmm. KDE... (Score:2)
--
Re:Is this different from Microsoft? (Score:1)
Close, but no cigar. It has to be: KDE_1_2_STABLE CVS, and it has to be complemented with the individual apps also posted in simple tgz format. In addition, the individual apps in, for example, kdenetwork, should be available as individual tgz's. Why should you have to download and build 6 apps when you're only interested in one?
The branch is called KDE_1_1_BRANCH, that's the post 1.1.2 tree. I don't see why you've got a problem with that.
As for downloading 6 apps, well, you only have to do that once, and you NEVER have to compile every app. For KSysV development (meanwhile moved to KDE2), I never do that, I just compile kdeadmin/ksysv, and maybe kdeadmin/doc/ksysv.
The only thing I see getting in the way is core developers are not, at this point, willing to recognize that non-core developers have a big barrier to getting started - much bigger than most other open-source apps. Even mozilla is easier to download and compile than KDE, let alone individual KDE core apps which have to be dug out with considerable effort from a rather indimidating CVS tree.
Really, in what way is downloading & compiling KDE less easy than doing the same for Mozilla? There ARE snapshots of libs and app-packages available, and the next set of those will even contain CVS information, so you just need to update a specific directory from that on.
It's a font thang. (Score:1)
It's also remotely possible they just strip out anything resembling HTML speak when they submit the page.
Okay, I'm reaching here, but it could happen.
Re:Security (Score:1)
Exactly! It's a calamity. I mean it's just terrible. I turn on my box, and this thing call a 'kernel' auto-starts. I start X and this X protocol thingie starts running. I load my email client and it "auto-starts" a library to connect to the pop3 server.
When you load a program, you expect the program to load itself and function as it should. If you don't like the functionality, then use something else. Complaining may make you feel better, but doesn't do anything productive. If I want an email client that has pop3 support but not imap support, then I'd use an email client with onlyh pop3 support. I won't whine everytime someone tried to add imap support to an email client.
-BrentRe:Is this different from Microsoft? (Score:2)
Instructions for building Mozilla:
1) Download the source tarball [mozilla.org]
2) Unzip into directory of your choice
3) cd into the directory and type: make
4) Go for coffee, come back in 40 minutes
5) Type
Instructions for building kmail:
0) Look for a kmail.x.x.tgz. Find a very old one. Get confused. Hunt around with RPM. Figure out it's now part of kdenetwork. Look for kdenetwork.x.x.tgz. Find a version that's way too old. Learn you have to get it out of CVS, and you have to get all of KDE with it.
1) Explore kde.org. Find the part that tells you about cvsup
2) Download and install cvsup
3) Find the instructions about how to get download KDE via cvsup
4) Decide what parts of KDE you actually want to download. Get confused. Decide to download everything
5) Start the download and go to sleep
6) Wake up, discover you got disconnected, repeat as necessary
7) Get a cup of coffee, sit down, and start figuring out how to build it
8) Guess that you have to build QT first. Try to build QT in the obvious way. It doesn't work. Go back to instructions. Learn you have to link a directory first. Do it and try again.
8) Go out for a walk while QT builds. Come back, find it's still building. Play Mahjong.
9) Try to use the same strategy to build kdebase. It doesn't work. Try some more. It still doesn't work.
10) Go back to developer.kde.org. Learn about the build script. Download it. Fiddle with it until all the path variable are correct. It finally seems to work. Out of fear, decide to build everything.
11) Go to sleep while it builds.
12) Wake up. One of the first packages failed to build - it's because somebody checked in something ugly that broke the build. Start cvsup again. Go for another walk while the tree updates. Later that day, start the build again and go to sleep.
13) It builds this time. Now figure out how to install it, without conflicting with your current KDE installation. Go for a surf. Find a howto on this subject. Read it. Fiddle with paths. Now it works.
14) Run kmail from the new desktop
OK, I may have missed a couple of steps, or a couple of details, but that's essentially what has to happen. Umm, don't you think there's a difference in difficulty and time investment between the two procedures?
There ARE snapshots of libs and app-packages available, and the next set of those will even contain CVS information, so you just need to update a specific directory from that on.
But why can't I just download a tgz of the specific package I want to compile, like almost any other open source project? Then all I need are headers for kde and qt, and I'm up and hacking. There is a *radical* difference in the amount of work required to get started. Later, when I've got some code I'm proud of, I can figure out how to submit it to a maintainer. In other words, the normal open source process.
Package snapshots only eliminate a few of the steps I listed, and you still have to figure out which packages you need, which directories to put them in, and how to set the paths. This is a *lot* of work if you've never done it before.
Mainly, what I hear you saying is: there's no problem, it's just your imagination. But it's not my imagination, I've been through this. I know what the difference is between downloading and compiling "spruce", for example, versus downloading all of kde and qt just to compile "kmail". It amounts to *days* of extra work the first time you do it. It's not my imagination.
Perhaps I should rejoin the KDE mailing list and we can continue this discussion there. *But* I still have this nagging feeling - why should I do that, when the last time I did, I just got flamed?
--
Re:Is this different from Microsoft? (Score:1)
Sorry to say, but either you WANT it to be difficult, i.e. troll (though I don't think so really), or you construe things to be much more complicated than the really are.
Instructions for Downloading & Installing KMail
Yes, the list is longer than for Mozilla. OTOH KDE IS much more than one app, it's an environment/application framework.
BTW: your comment about Qt needing to be "linked" certainly doesn't hold true for Qt-2.1.1, maybe you are confusing it with some earlier version?
Re:Is this different from Microsoft? (Score:2)
This is exactly my point. I do not want to compile KDE. I want to compile kmail. So why am I force to compile 6 more apps and a good part of KDE and possibly QT? When we both know that it would be easy to provide to me the source code I need just to compile kmail? And after doing what you suggest, I wind up with the wrong version of kmail - one that runs under a beta desktop that I'm not using. Fixing bugs there doesn't fix the immediate problems. Downloading other snaps doesn't gives me an old frozen version that nobody is fixing bugs in.
Good gosh, man, can't you see that if people are complaining, there must be a problem? It takes a lot to get people to speak up on something like this. My original post was moderated up by four readers and not marked a troll by any, as it easily could have been. What does that tell you? And what does it tell you about how it will be perceived if you keep sticking to your guns instead of looking for a resolution?
By the way you left out the "fiddle with environment variables" step in your list, maybe others.
--
Re:Is this different from Microsoft? (Score:1)
You won't have much luck with Mozilla either, if you haven't got gtk+ or libc.
As for "my original post was moderated up": my posts won't be moderated up because the story is too old, we all know what stories are actively moderated, and one two days old isn't among the likely list.
And what's the problem with downloading the tarball of the 1.1.2-kdenetwork? Sorry to say that, but to me you appear to be complaining for the sake of complaining.
And you don't HAVE to compile 6 apps, even if you don't use the nice "inst-app" file, you can still simply uncomment the other dirs in the Makefile after configure.
As for "fiddling with environment variables": yeah, you're completely right, setting two variables (KDEDIR and QTDIR) is just SOOOOO hard...
Re:Is this different from Microsoft? (Score:1)
Ok, ok, you win, you win, actually there is no problem at all with obtaining compiling KDE apps such as kmail, and as I have better ways to spend my time I'll forget about it for another year, thanks for the advice.
--
Thinking right about KDE2.0 (Score:5)
Who needs all these KDE apps... (Score:1)
W3C HTML Tidy and the demoroniser (Score:1)
Re:Is this different from Microsoft? (Score:5)
How is this all-inclusive, single-source applications/OS environment any different than, say, Windows and Office?
You may be surprised at how much support you get. Disclaimer: I *use* KDE every day and I think it's great. I use lots of KDE mainline applications and lots of what I'd have to call "3rd party" applications. (though I probably use even more Gnome/GTK applications, running them from the KDE desktop.)
I have some experience compiling KDE out of CVS, have done a small amount of KDE/QT development and have participated in the KDE mailing list. My intention was to become a regular KDE contributor.
This was short lived, however, as I quickly found that the KDE mailing list, was being run as a sort of private fiefdom by this same David Faure. Any comments of the form "KDE mainline applications development should be opened up to more developers" i.e., not just be restricted to the main CVS tree, were met extremely aggressively. I didn't want to press the point because I thought "maybe he's right" and all the core applications *do* have to be tightly controlled by the core developers in order to have the highest possible quality.
But now, 6 months later, I know he's wrong. KDE development has slowed down *a lot*, and especially KDE core applications development. I've been putting up with the same bugs in kmail for almost a year now - it hasn't changed a bit, because kmail development has been inextricably bound into the main KDE CVS tree, and it's all oriented towards KDE 2 now. So if you're not compiling out of CVS, you're out of luck. The bottom line is that KDE users have to wait and wait and wait for bug fixes.
Now, I'm a developer and I should just be able to download the code, make the fixes and send them in. But in the case of kmail, you have to download practically the whole tree, and let me tell you, it isn't that easy to get it to build. I would have done that anyway, except for the agressive reception I got on the KDE mailing list. Anyway, I'd still have a problem with it, because suppose I put in a bunch of great features - It would still be months and months before anybody gets to use them, because they have to wait for the official release of KDE 2. One of the reasons developers develop for free is to see their stuff get used *immediately*, and this just doesn't happen with KDE development.
I did what open-source developers are famous for doing - I moved on to other projects that are more to my taste (e.g., Freenet [sourceforge.net]!!). Now, I'm not bitter - why should I be, I'm still getting KDE for free and it's still darn good - but I am concerned that a certain prima-donna mentality is hurting the KDE development, and because of that, hurting the entire open-source movement. I wonder how much of this is due to the fact that some fulltime, paid developers (Faure is one) don't feel compelled to play by the traditional, unspoken rules of the community. OK, I've said my piece, and I'm posting this without my +1 bonus because I'm not sure whether this is a flame or a troll or constructive criticism.
--
Re:Interesting point. (Score:1)
Does that mean we're gonna have scripting faciliated inter-process communication, much like ARexx on good ol' Amiga?
--
Re:Smart/Stupid Quotes? (Score:1)
I see how that would give you a problem, but mostly I was wondering why linuxuk.org.uk was using MS Word and/or MS SQL. Surely they aren't doing that, so what other circumstances could cause the '?' quote thing?
Re:Scripting...nailing the barn door open? (Score:1)
It seems like KDE is boldly going where Win9x has gone, without n levels of obfuscated templates between the developer and the toolkit, and is using the same kind of mechanisms/methodologies. They probably don't have much of a choice because if the technology favors the method then developers have to go with the flow...or reinvent the wheel differently; and reinventing the wheel takes time and money not readily available to people making free software.
Re:Ouch (Score:1)
Incorrect: only KApplication using DCOP needs the DCOP server. No developer is forced to use DCOP when they code a KDE application.
Now, have a look at which applications does logically require DCOP: mainly KOffice and Konqueror. Lemme say you, if you want to use KOffice, it will be stupid not to install KDE. If you want no to have KDE, it will be stupid to want to use KOffice. The same is true for mainly all other DCOP-based KApps.
Now, a good programmer can put some test in his code to check at launch time if a DCOP server is available and then act regardingly, it could then just display a massage box warning that desktop communication is not available, and avoid each call to DCOP stuff.
Finally, you said "It was bad enough when the kdelibs were required". Yes. Sure. All program should be statically linked with all libraries it needs, and never call .so files. This allow a very powerful and unnecessary bloat in the RAM. I think that you'd also want to have X working withou xlibs, gnome working without gnomelibs, and your whole system working without (g)libc. Go to /lib, /usr/lib, and /usr/local/lib and do some rm -rf *, then reboot your system.
Pathetic Licensing Answer (Score:1)
That's odd because KDE definitely has a licence; LGPL for most of libraries and GPL for most of the applications, and Qt indeed has a different licence; it is QPL. It isn't much different for home users or for even users in a business environment. Troll Tech have put a lot of time and effort into producing a high quality toolkit and in all fairness they should get paid if you sell an application that uses Qt.
The only thing I see here is two things:
I'm pretty sure P. Hands didn't say that, but as I can't prove this, I'll just shut up on this.
I think the question is as bad as the answer. KDE does have a license, of course (we know things are useless without a license, ie, they are the most non-free pieces of software). Saying it doesn't have one is not very accurate.
This question tried to get an opinion about the KDE licensing wrt the recent $3K someone offered. It's a pitty the question is dismissed with "yes, we have LGPL, GPL and tk under QPL". We know that, I really want to know what the KDE people have to say about the accusations of GPL violation in KDE&family. Sigh.
And, should I comment about his lattest statement? It would be really funny to see the QPL changed so people had to pay Troll Tech for selling Qt apps. I guess that would benefit Gtk so much...
Re:KDE acceptance (Score:2)
Re:Standards, PLEASE (Score:1)
This remind me, one day, after receiving agressive comments about my english, I had thoughts about changing my /. sig to something like:
This comment is written in Microsoft English. If you can't read it, please upgrade your brain to a Microsoft Brain with Microsoft English supported.
Stupid joke aside, you are discovering that english as the new lingua franca means also bastardized english.
And sorry if I'm not clear.
Re:Security (Score:1)
On the other hand if you want to run something, do what it requires. Be it install an operating system, some libs, whatever. If all we're doing is bitching about huge code bases, I can say the Linux kernel, glibc, the bsds, etc are too big, programmed by humans, and insecure, therefore they are equal to anything by Microsoft.
Re:Is this different from Microsoft? (Score:3)
You wish that KDE core applications were developed out of CVS ? With the changes that happen in kdelibs quite often - to build a better API, and because of all the new stuff in KDE 2.0 (KParts, DCOP, XMLGUI, etc.) - this is almost impossible. What you gain from having apps in CVS is that they can be fixed by many people, especially those who change kdelibs.
Out of CVS, it's up to the author to fix his app to cope with the changes, and _that_ can be piss him off a lot, obviously. This is not about "controlling", this is effectively about helping....
To answer the more personal comments, KDE is definitely nobody's fiefdom. My answer (which I can't remember) may have sounded not nice and I apologize for that, but the idea of developing core apps out of CVS really doesn't make much sense. You want to "open more" the development by making it private ? About KDE 2 apps that are not in CVS : from 1.91 onwards, they should be able to compile and run on top of KDE. The API shouldn't change anymore. No need to wait for the official 2.0. Finally, I'm not a paid fulltime developer... yet.
Re:Standards, PLEASE (Score:1)
This does make me worry about SECURITY. (Score:2)
Konqueror is very modular, if I've understood it correctly. That means its possible to write extensions (easily). *That* make me worry. What if someone gets this great idea of showing emails, and executing the javascript included in them? What when Linux get visualbasic script support - what happens when some idiot thinks "Hey! What a great idea! Email reader with js and vbs execution support! It's User Friendly!"(bad pun intended).
As long as linux programmers avoid these pitfalls, we should all be safe.
--
"Rune Kristian Viken" - arcade@kvine-nospam.sdal.com - arcade@efnet
Re:what a jerk (Score:1)
You're kidding me! That will bring down a windows box? Ha
Nope. I can click that link all day long and my NT box won't crash. Well, not for *that* reason...
Re:This does make me worry about SECURITY. (Score:2)
> "never ever allow scripting in konqueror"
Why ? What's wrong with scripting "open a new window, open that URL inside it", and even "type my login and password for that site" ? As long as _I_ run the script, where's the problem ?
For too many people, scripting == security risk, without really thinking about it. No. But automatically launching untrusted scripts (such as those received by e-mail) is a security risk, yes. That's why kmail doesn't launch scripts.
Re:Ouch (Score:1)
Back on topic - Konqueror (Score:1)
I really much prefer Kde 1.x to either Gnome or the new Kde 2.0 so far. It's fast, sleek (in my opinion heavy pixmapped themes and large, blocky icons are NOT visually pleasing) and very stable. The browser (kfm) has some limitations but is more than enough for most browsing and downloading and email with KMail. I much prefer it to Netscape, although I must admit that Netscape 4.73 is a very stable and excellent, full featured browser. It a much better browser than MSIE 5.
Kde 2.0 is very ambitious in trying to make everything a "part" or component like the MS desktop. Believe me, this project, with KOffice, is every bit as ambitious as the entire Windows desktop and application framework, including COM and DCOM. and MS Office to boot. It took MS 10 years to get that far. Kde is trying to do the same but do it right in 3 to 4 years (since Kde 1
first became usable as a desktop).
In my opinion, success so far is only partial. For example, Konqueror is very fast and smooth, really a pleasure to use. It overcomes all the limitations of the 1.x kfm. But it crashes far too often. I feel that these crashes are due to the tremendous complexity of the code, particulary the interaction between DCOP, Kparts and the parsing of urls. Until this is stabliized, I don't recommend using Konqueror except for testing. It seems that when one thing is fixed, then something else breaks in the core libraries and base system. The build process is also nowhere near finalized, and you can expect some things not to compile from day to day in what is on cvs.
Really, I cannot emphasize enough how ambitious this work is, and that it is several orders of magnitude beyond what is currently available for unix. But, can they pull it off? The code base is much larger than Kde 1.x, although the executable code size is not. Even though the onerous burden of Corba was pretty much abandoned as a design decision, I feel that the degree of complexity may still be too great. All I can ask is that more "features" not be added until the current state of Kde 2.0 beta is stabilized and any temporary hacks used to fix noticeable bugs or speed up some operations are worked into the design in a consistent way that retains the beauty of the design.
Right now I am using Gnome 1.2 as my primary desktop so I can work on some qt 2.x code without conflict with the qt 1.x based kde 1.2. I can say that Gnome 1.2 is super-stable. It has never crashed. But it is very much lacking. It is not fun to use. Keyboard support is almost entirely absent, and many apps are in a primitive state of development. The "design" from a visual point of view is not pleasing to me - the proportions are top-heavy and many things can't be configured which are easily configured with Kde. This kind of configuration means more than "themes". For example, with GMC one cannot hide the large, blocky and space hogging icons on the taskbar and there seem to be no keyboard or menu equivalents. Of course Gnome is utterly dependent upon Netscape, a commercial product, for web access and browsing. Talk about free software. Gnome would be totally unusable without linkage (in a lose sense) to this very proprietary and closed app. Still, I am pleased that Gnome is now stable and hope that it continues to improve, along with Kde. We need choice! Maybe Gnome could incorporate the really free components of Mozilla into a new browser.
I guess it is my fault for being impatient for Kde 2 to arrive in a more stable state. But I think it will be worth waiting for!
Any comments by David Faure, if he is still lurking around, on these issues will be appreciated.
Why make kde specific apps? (Score:1)
What is I want to use an app while I'm at my Solaris machine, or at an NT machine?
What if I'm enamored with an older windowmanager?
What if I have a hacked i-opener and I don't want to run kde on it?
Thanks,
George
Interesting point. (Score:2)
Shouldn't some restrictions be put on open scripting like this? We don't want a microsoft problem coming and haunting KDE too... Perhaps you can have a database of signed scripts that are able to use the KDE system or somein...
Here's what facilitated that comment:
What elements are mainly shared between all applications in this framework?
Scripting is, in fact scripting is shared between all KDE applications. How you talk to a KDE application and how you script it is a generic system. Among all of the KOffice applications you can also find a lot of common dialogs, and a lot of common tools that are in the framework that allows the user to see a very integrated office suite, not a different set of applications. It's all about having a full, complete office suite which has the same elements for everything.
Re:Is this different from Microsoft? (Score:2)
It's not his fiefdom nor is it the fiefdom of any of the other main developers- but there is a definite centralized development model. IMHO, I have to say that this has worked very well. The KDE project consistently puts out good code and good programs.
As far as KMail goes, you might want to check out kdenonbeta/kmail2 and help out with that. It can use some new developers, and it's definitely not release status yet.
As far as the release cycle goes- we release when it's ready. Not when we're ready, not when there's a deadline, not when Another Desktop© releases a new major version, but when we're ready. Sorry if you've got a problem with that. Besides, nobody owes this code to you- if you don't like it, there is always GTK+ and that Other Desktop...
Standards, PLEASE (Score:2)
"I run the CVS version of KDE2, and I remember seeing an IRC thread in one of packages regarding not running a mail composer within Konqueror, but a mail viewer, so will the composing of documents mainly be launched by an external editor? "
"In terms of KOffice, what parts of KOffice are you working on? "
These are barely grammatical, and very hard to read, certainly not reviewed by whoever wrote them. We all get fed up with web sites that abuse HTML, so why should be stand for abusing English just as badly.
If anyone from linuxuk.co.uk is reading, I hereby volunteer for a job as a sub-editor so no-one else has to read this kind of stuff...
Smart/Stupid Quotes? (Score:1)
I'm surprised to see the '?' replacing the apostrophe character in this interview - isn't that normally the case with MS web publishing tools and their "smart" quotes? I'm viewing this with Netscape 4.7 on HP-UX.
What?s up with that, linuxuk.org.uk?
Re:Back on topic - Konqueror (Score:1)
We are working on stability, don't worry. And no, I don't believe the code base and the complexity makes it impossible to achieve. On the contrary, it's a lot cleaner than for instance kfm's code, which was just a PITA.
Some things can NOT be fixed in kfm, due to its design. With konqueror, however, everything is possible, since the design is a *lot* cleaner.
BTW: about Qt 2.x / Qt 1.x: you can have both on the same system and still use KDE 1.x :)
Scripting...nailing the barn door open? (Score:1)
Re:Why make kde specific apps? (Score:2)
Re:Ouch (Score:1)
We didn't replace CORBA(MICO) with DCOP, we replaced it with KParts. DCOP is mainly used for IPC and scripting - and yes, I agree with you that this is needed, whichever solution.
All you need are the libraries (Score:1)
You *can* run KDE apps under GNOME, or under any standalone window manager. You just need the KDE (including QT) libraries. There are some minor drawbacks: 1) inconsistency in look and feel between QT / GTK+ apps ... I find it jarring, but you may not. 2) if you use GNOME and run a KDE app, you have to have two complete widget sets in memory ... which may or may not be an issue, depending on what you do and how much RAM you have, etc.
Standardization: KDE vs. Gnome (Score:1)
I don't really see how KDE and Gnome are different in terms of providing the API's for standardizing user interfaces, while allowing complete freedom to do things like change the toolbar icons, put the menus elsewhere, etc. What is meant by the "standardization" that KDE offers vs. Gnome "accomodating everything." Of all the differences between Gnome and KDE, this isn't one that I've really noticed. Any thoughts?
----
Re:Is this different from Microsoft? (Score:3)
Not everyone thinks the same way as the usual
What the KDE project is the vision of a true integrated desktop in a clean, non-bloated implementation. I've run KDE1.89 and 1.9, as well as from CVS on a variety of systems, and it never seemed slow or bloated. Instead, it was clean and easy to use. I enjoyed using it because of the integration, unlike Windows where I hate using it because of the botched implementation.
To put it simply, KDE2.0 is the ideas of integration and usability done right. It's a different beast than KDE1.1. KDE2.0 is actually more usable than windows, and easier to program, while implementing the ideas that Microsoft used.
Re:Why make kde specific apps? (Score:3)
They do. Let's take your questions one by one:
What is I want to use an app while I'm at my Solaris machine, or at an NT machine?
Well, you can run KDE on BSD, Solaris and quite a few other OSes. NT is not included - nobody has really expressed an interest, but I'm not sure how large the technical hurdles would be. That means running native on the system (Note that it's periodically broken for OSes other than the core developers during alpha/beta testing. The core devels use Linux, BSD and something else I can't remember, but might be Solaris).
What if I'm enamored with an older windowmanager?
Use it. KWM is nice, and has *quite* a few new features, many of which are integrated into KDE, but are not necessary. I've run KDE 1.x with E with no problem, and 2.0 is supposed to work (dunno what features you lose, though. Themeability woul be my first guess).
What if I have a hacked i-opener and I don't want to run kde on it?
X Windows is network transparent. There are a handful of NFS and "running apps remotely" bugs that cross the lists. They are addressed as valid concerns, so I would imagine that they will be addressed. The lightweight KDE communication mechanism DCOP (used instead of Corba where appropriate, although Corba is still used where *it* makes more sense (like the multimedia subsystem)) operates over a X Windows standard communication system. To quote the developer site: "DCOP is built on top of the Inter Client Exchange (ICE) protocol, which comes standard as a part of X11R6 and later."
In short, it is an X system that is portable to many platforms. Some of this I've verified myself, the rest I know from the mailing lists, where KDE 2.0 beta is already in use.
--
Evan
Re:Standards, PLEASE (Score:1)
I agree this is hard to read. I got the gist of it, but it took a couple of rereadings, and a non-English speaker may not have understood.
I think that's a perfectly acceptable colloquialism. Ok, it's got redundant information in, but so has "look at that pen of David's" (double genitive).If you get the job, I'm sure you'll do a great job of weeding out ambiguous, messy collections of words like the first example. But please don't kill off valid colloquial phrases! There's more to English than writing formal letters, and there's no reason why linuxuk.co.uk articles should have to be formal.
Re:Why make kde specific apps? (Score:1)
but they do run in any window manager. i use olvwm and run some of the kde apps/games. they run fine. and without the "desktop environment" there's less overhead which works much better on slower machines.
Re:Interesting point. (Score:3)
First, it was stupid users, apparently. I've heard that MS Outlook will automatically run the script if the e-mail is viewed in the preview panel; I've heard some Outlook users say that that's false. I've used Outlook all of about twice in my life, so I can't say. If the script is run automatically from the preview panel, then I would say that's bad engineering (and hence bad Microsoft), but it sounds like they've at least fixed the problem.
There's also the question as to whether scripts should be executed automatically by default (e.g. when you double-click or otherwise "activate" them). Personally, I don't care. Turn it off by default, sure, sounds good.
Most importantly, though, I think the users need more control over their scripting. This is one area where KDE and Gnome have been falling down (gently). By far the most annoying thing about Windows and MacOS is that the user has almost no control, either over the OS or the applications. A few years ago I would have almost literally killed for the ability to put some scripting into Netscape that said "yes, Mr. Javascript, you can open a new window, but not when I am viewing a page from Geocities". You can argue that I could have just disabled Javascript (which I did), or made something similar to Junkbuster which would automatically parse and change all Javascript, thereby eliminating the pop-up windows (hah!), but this would have been such a nice solution.
So anyway, back on topic. I think the user should be able to do some metascripting, scripting exactly how the script should behave in certain situations. This is not the same as clicking checkboxes (especially vague ones like "Run untrusted scripts"); I want some actual *control*. If KDE (and Gnome, but they're not important in this article) do not do this, then they're just going to end up as Yet Another Inferior MacOS. I think it's time for Something Actually Different From MacOS.
Re:Standards, PLEASE (Score:1)
While that is certainly true, I meant to say "non-native English speaker".
Ouch (Score:1)
You put this in a list of features, but it is a bug, AFAIAC. It was bad enough when the kdelibs were required, but now I have to run (and probably configure) servers? Total nonsense.
For instance, let's say I don't use KDE. But occasionally I find an interesting "kprogram" on freshmeat. For KDE 1.x, I could have the libs installed and then that k-l33t prog would work OK. Now even that (extremely lame) option is closed to me.
And don't bother answering "but you only have to do it one time" because that "one time" is the first time--meaning it will never be done. The only other alternative is for distros to auto-start this dcop crap for ALL installations--a spectre so grim that it doesn't bear thinking about.
By adding this one "feature" KDE is isolating itself from the rest of the software world.
--
Wanna hook MAPI clients to your Tru64/AIX/Linux server?
Re:Is this different from Microsoft? (Score:1)
Now, with Linux, KDE, GNOME, etc, this freakin monopoly argument holds exactly zero credibility. First, you have the source. There are no secret APIs to give one person an advantage over another. There is no necessary bribes to find out secret API calls. Second, since this is free software, there is no financial problems with software bundling and the like. We all saw what happens when you ask computer manufacturer X to remove Office/Windows. If they want to use Corel or some other office suite/operating system, you pay for the Microsoft stuff with the computer and again with your chosen operating system/office suite. Third, and finally, Konqi/Tux are much cuter than that Windows flag logo thing.
Now, if you just hate KDE because it's KDE, be outright with it. Do not invent stupid arguments like "It is/might be Microsoft-esque." No need to use that jump-to-conclusions floor mat.
Scripting Issues (Score:1)
Among the options proposed include a sandbox for running scripts to see what they do, before running them for real, and not allowing certain operations from untrusted scripts.
Really, though, the scripting in KDE does not raise any additional security issues than just running scripts/binaries that were sent to you in email or that you downloaded off the net.
It has also been unofficially decided that the default for automatically running scripts/macros will be set to "off" so the user will have to set it to on for scripts to run automatically.
Anyone else who has been reading the mailing lists care to comment?
Re:Is this different from Microsoft? (Score:1)
From your post, there is one thing funny. And even ironical: "Isn't there a parallel with Office's non-standard dialogs and widgets, and KDE's different-from-other-GUIs dialogs and widgets". I can't retrieve this document, so there is no link, but I remember reading an archived version of Matthias Ettrich's call for programmer: one of his argument for launching the KDE project (btw, that was meaning "Kool Desktop Environment". Ettrich thought of this as a codename, because it was a little sucking, but the name stayed and the official version is that "K" doesn't mean anything) was that all widgets and dialogboxes ware behaving differently. So, he says, we need a true Desktop Environment rather than a mere Windows Manager with an united look and feel, a lone widget library (instead of Tk+Motif+Athena... There wasn't GTK+ at this time). KDE was the first project to try to use a "standard".
About the rest of your question: embedding application is aimed at reducing bloatware. With OLE2, each application (says Word for example) contain subsets of the other (says Excel). When you edit a spreadsheet in Word, it's not Excel which is used, but a small Excel clone located inside word. So Office get bloated with redundant stuff. This will be avoided in KOffice.
There isn't more risk in a script than in a normal program. The risk is how it is managed.
Finally, MS was not condamned for integrating closely things in Windows, but rather for using proprietary tricks (non documented API functions, of functions non mentionned at all) to prevent their competitor from releasing product with the same integration. Hey, you prefer to use something that is seamlessly integrated or that is just an alien program ?
KDE being free software (despite what says some trolls), this I-hide-my-secret behavior is impossible. And, trust me, if someone in, says, Gnome, decided to code some bridges to have Gnome Office integrated with KDE as transparently as KOffice, or vice-versa (KO for Gnome) with Bonobo/DCOP-etc translators, the KDE team will applause and thanks. In the same situation, Microsoft will modify its standard to have the bridges not work anymore.
Re:non-free (Score:1)
That I'm a B1FF or that I'm cool?
If I'm a B1FF then you've just been trolled. If I'm Cool then thanks.
I suppose your moniker says it all about you too.
Conscience is the inner voice which warns us that someone may be looking.
Re:All you need are the libraries (Score:1)
So, YES kdelibs is sufficient to run KDE2 apps for those that do not wish to use such a beautiful environment.
Re:Interesting point. (Score:1)
There's also the question as to whether scripts should be executed automatically by default (e.g. when you double-click or otherwise "activate" them). Personally, I don't care. Turn it off by default, sure, sounds good.
Or you could ask the user for confirmation when (s)he tries to run a script or program from the email client.
Donate Food for Free - http://www.thehungersite.com
Re:Is this different from Microsoft? (Score:2)
KDE becoming windows? (Score:2)
a better solution would be to have interfaces where you just load the library for the features you want.... which we pretty much have, but could have in a better way. for example, right now you can run KDE apps in enlightenment, but how about a totally minimalist X program that is not so bloated? i am thinking of moving to sawmill unsure tho...
Even More Pathetic Licensing Answer (Score:2)
- The GPL prohibits you from placing any greater restrictions on software you distribute.
- The QPL places such restrictions (you must give Troll a copy).
- Therefore, the licenses are incompatible.
No one is suggesting that Qt isn't free. What we're saying is that your license is logically inconsistent, and that therefore it's impossible to know what license you really want.
Essentially, you don't distribute KDE under the GPL. You *say* you do, but really you don't, because distributing KDE binaries would be illegal if so, and you haven't seemed to mind so far.
What Debian wants here is a statement of *what* your license really is.
Re:Pathetic Licensing Answer (Score:1)
This license nonsense really makes me sick. Talk about fanatism...
About the second comment: I love being quoted out of context. Re-read what I meant by that : the fact that it's interpretation is really subjective, since not two people interpret it the same way. So this "incompatibility" is far from being proven. What makes you think there is one ? Because RMS says so ? Sometimes I feel like there is a dictator, "free" software or not...
Re:Security (Score:1)
The DCOP server is a server for inter-process communication, but only between apps running under the same X server. Besides, DCOP is implemented of top of ICE, which what is used for session management everywhere now.
So there is no additionnal security risk than an X server, which you already do.
Re:Even More Pathetic Licensing Answer (Score:1)
The GPL enforces that what if you base your software on a GPL application, you release it as GPL.
The QPL enforces the same thing, you must open-source the software based on Qt [unless you buy a license]. So this is not an additional restriction !
"Give trolltech a copy": this is the case for any open source software ! You don't only give a copy to some people, you give a copy to the whole world !
I really hate this license crap. Can't you see that both the GPL and the QPL aim to the same thing ? They both enforce that derived work is also free and opensource.
Yes, KDE's license *is* GPL/LGPL, nothing else.
Re:Is this different from Microsoft? (Score:1)
But nobody would force you to use every new micro revision. The project leaders can designate a particular release the stable one every so often, but people who want bug fixes quickly (I've waited for kmail bug fixes too, and submitted one of my own) can still get them from the main distribution.
I don't see the advantage of not releasing often - isn't that how you encourage bug fixes and developer mindshare in your project? Users who want stable, routine releases can wait for major revision numbers if they want, but that isn't really the way to encourage developers, IMHO.
Lame excuse for Qt license (Score:1)
Please, everytime I hear KDE developers give this line I want to cry out. What kind of reasoning is that? I that was the logic we should follow with Linux, then shouldn't the kernel developers get paid, the Xfree86 developers, the gtk developers, the GNOME developers, or for that matter the KDE developers and so on. Or is the Troll Tech people the only ones who contribute time and effort.
Face it the QPL is a crappy license which only aim was to be 'free' enough to kill of the Harmony project.
So KDE developers get a grip, it is your right to develop your software with quasi-free libraries, but stop pretenting that the Linux community should be gratefull to Troll Tech, they are just a bunch of developers like everybody else, but developers using a license which is more in the spirit of Microsoft than the Linux community.
Re:Is this different from Microsoft? (Score:1)
Microsoft bundles things together to sell them all and make sure you don't go and see other vendors.
KDE _offers_ many bundled components, but that doesn't mean you have to use them all if you don't want to.
Example: if you have the "kview" image viewer, konqueror can use it to show images embedded in a window. Fine. If you don't have it, it will launch your preferred image viewer instead, be it xv, ee, gimp or vi.
Face it: the biggest need on Linux at the moment, in terms of software, is for a free and full-featured office suite. KOffice is on its way to answer exactly that. Should we not work on KOffice because Microsoft used MSOffice to tie people to Windows ? Nonsense.
Re:Lame excuse for Qt license (Score:2)
Please read the thread "pathetic licensing answer".
Re:non-free (Score:1)
Re:Security (Score:2)
Re:All you need are the libraries (Score:1)
Re:what a jerk (Score:1)
Not all of them. But certainly some that I've seen (and that's Win32 versions, not old 3.x versions). Dunno exactly what you need to do it, but I'd try IE3 and Win 9x at a guess.
The problem is the other way round (Score:2)
I understand what you're saying, an interesting argument which implies that the GPL might not hold up in court for libraries (depending on the particular court).
But the problem here is the other way round - some people wrote GPLed apps, and the KDE people have linked derivative works to Qt. Troll may not be able to control what links to QT, but they certainly have legal control over how the library is distributed. The way they license this control is not free enough to be compatible with the GPL which the original app was placed under.
Re:All you need are the libraries (Score:1)
Re:Is this different from Microsoft? (Score:2)
Believe me, I seriously considered that and would have done it if I had not gotten fully committed to other projects. The need to spend several hours downloading the full CVS tree (analog modem) and fiddling with the build process was the final straw. (I've since done that, and it did in fact take a lot of time.) But why is it necessary to solve the problem by forcing a fork? Why not address the problem in the first place by having a well-defined fork a la the stable-vs-development strategy of the Linux kernel?
Personally, I think David's an awesome guy- he knows what's going on with the code, he has some really brilliant ideas about how to implement features in Konqueror, and he's never been rude that I've seen.
I must have caught him on a bad day then, and what I've seen recently supports your opinion. I still I think I'm right about the question I raised and that KDE app source code needs to be more accessible to developers outside the inner circle. It's a real problem that bugs aren't getting fixed in the stable releases and it's a problem that isn't hard to fix.
Please note that none of what I say reflects in any way on the KDE 2 development - I think that's working great, and there's fundamental progress going on there, particularly in object embedding techniques. And the attention to detail and useability in the KDE development has always impressed me. No doubt such things are in part due to David's skills.
--
Pinpoint freedom (Score:1)
Re:non-free (Score:1)
KDE is a disease that should go away. The developers are showing their contempt to users'/programmers' rights by choosing a non-free license. And please, do not tell me that it has changed and it is free now. If they really want a free license there are plenty available: GPL, LGPL, artistic, BSD. Why do they have to keep creating problem by having their own little incompatible license?
I do not understand
Re:Even More Pathetic Licensing Answer (Score:2)
Well, I don't want to detract any more from the otherwise good stuff in the interview than I already have. Dave, if you want to, E-mail me, and we can discuss our differences. If not, fine; and good luck with Konquerer.
I will say a few things:
I'll shut up now.
Oh please... what makes this "insightful"? (Score:1)
mmm. KDE... (Score:1)
Anybody else tried this? (slightly offtopic)...
Re:Why make kde specific apps? (Score:1)
Re:KDE acceptance (Score:2)
In addition in most books covering linux you'll also mostly find KDE being dealt with when it comes to the desktop.
And if you read the mainstream-pc-magazines you'll also discover that KDE is being mentioned as "the" desktop for Linux there.
A.C.
Re:One question though... (Score:1)
-- snip ---
I think www.kde.org/technology.html (mosfet's) speaks quite well about this.
If not, my take on it is:
- stability (that's the main one. If you used pre-canossa KOffice, you know what a real stability problem with embedding is).
- wrong approach - starting servers for embedded components, leads to servers left running after you exit the main app (ok, that's a bug, but it shows wrong design IMHO. What should be one application was a set of distributed ones, hence this kind of problems).
- memory usage and slow performances - but people associate this with MICO more than with CORBA in general. Well, we'll never really know, but in any case we had to do something about this.
- the interoperability we were supposed to benefit from it was null. gnome uses an orbit-specific authentication mechanism which made interoperability as difficult as it is now, and nobody ever had any interest in making the CORBA things in gnome and KDE work together anyway.
In short, I always say: we had all the disadvantages of distributed computing (stability, reliability, performance issues) for something that is not distributed, in 99% of the cases - a desktop is local, most of the time.
But it was fun, writing sources that nobody could read afterwards
--- snip ---
you might also want to read this one:
http://lists.kde.org/?l=kde-core-devel&m=957230
Re:Ouch (Score:2)
You'll still be able to download and run a KDE 2.x app even if you don't have the rest of KDE 2.
dcopserver is auto-started if it's needed and it's not running, so you won't even see that happening. And it certainly requires no configuration.
Re:non-free (Score:2)
This is true for stuff written *for* KDE. But for pre-existing GPLed stuff which the KDE people borrowed, there is no such implicit permission.
Please don't dismiss people as "license crackpots" unless you are sure you understand what they are saying.
This is insightful? KDE runs on all modern Unixes (Score:2)
As for running on NT.. NT isn't a Unix-like OS, so the entirety of KDE running on Win32 is fairly unlikely any time soon, even with the help of the porting and runtime libraries from companies like MKS. Windows isn't Unix, MacOS isn't Windows, and your refirgerator isn't a houseplant.
The Qt toolkit, on which KDE is built, however, is available for Windows--but not for free. Several of the apps in the KDE family are indeed available for Windows, and some others could easily be made available for it. But not so for apps that take advantage of the integration that the KDE layer provides. In other words, one might be able to port individual KOffice applications to Windows pretty easily, but they wouldn't integrate with each other, because the integration requires the KDE infrastructure underneath.
Similarly, the GTK+ toolkit, the basis of GNOME's UI, is also available under Windows. This has allowed GIMP to be ported to Windows; just don't expect it to communicate cleanly with other WIndows GTK+ apps like Mozilla without special work.
Re:Is this different from Microsoft? (Score:2)
Source.
Re:Is this different from Microsoft? (Score:2)
Close, but no cigar. It has to be: KDE_1_2_STABLE CVS, and it has to be complemented with the individual apps also posted in simple tgz format. In addition, the individual apps in, for example, kdenetwork, should be available as individual tgz's. Why should you have to download and build 6 apps when you're only interested in one?
There's no particular difficulty or excessive work in accomplishing it: a script can simply be written to extract and zip the necessary branches of CVS, and to create the necessary new CVS trees (so development of the stable apps can proceed independently from the development tree, the way it does in the Linux kernel). The script just needs to be run *once* following each stable release.
The only thing I see getting in the way is core developers are not, at this point, willing to recognize that non-core developers have a big barrier to getting started - much bigger than most other open-source apps. Even mozilla is easier to download and compile than KDE, let alone individual KDE core apps which have to be dug out with considerable effort from a rather indimidating CVS tree.
--