Not logged in
Log in now
Create an account
Subscribe to LWN
Pencil, Pencil, and Pencil
Dividing the Linux desktop
LWN.net Weekly Edition for June 13, 2013
A report from pgCon 2013
Little things that matter in language design
If you chose to ignore that, it is not the fault of the KDE hackers.
KDE 4, distributors, and bleeding-edge software
Posted Jan 29, 2009 0:37 UTC (Thu) by malor (subscriber, #2973)
If it's not done, you don't make it a .0 release. You call it an RC, or, more honestly in this case, a beta.
They knew that. Everyone knows that. But they deliberately called it 4.0 anyway to get testers. They've clearly and unambiguously admitted this: they thought calling it a beta wouldn't attract enough testers. Instead of being honest and asking for more help, they lied to get the help instead. A .0 release would be more broadly taken up, and too bad for the users.
This is purposeful deception, no matter what they put in the README. They tried to force contribution via deception, rather than asking for voluntary participation. That's about as unethical as you get in free software, short of stealing code outright.
Posted Jan 29, 2009 0:55 UTC (Thu) by dlang (✭ supporter ✭, #313)
if you read the article and the links in it, you will see that they made the release so that developers would have a stable platform to develop against (thus calling it a developers release)
they could have re-named everything from KDE 4.0 to KDE-infrastructure 4.0 and then make KDE depend on KDE-infrastructure, but however they did that split there would have been something one one side of the line that belonged on the other side.
x.0 == feature complete, end of story
Posted Jan 29, 2009 21:59 UTC (Thu) by khim (subscriber, #9252)
if you read the article and the links in it, you will see that
they made the release so that developers would have a stable platform to
develop against (thus calling it a developers release)
Yup. And such things have a name too: KDE 3.99.0 will be fine for such a
release. Or they can use MySQL approach and called it KDE 4.0-alpha (with
next revisions being KDE 4.1-beta and KDE 4.2-gamma). Instead they decided
to use version number which clearly says: it's finished! It's not polished,
it's not yet debugged enough - it's first rough release, but it's
feature complete! KDE 4.0 was quite far from being feature complete,
Posted Jan 29, 2009 19:09 UTC (Thu) by dkite (guest, #4577)
The fundamental social contract with free software is you can distribute it and change it. Some make it available for free as in beer.
There are no implied warranties, you own both pieces when they break.
That is the first part of the fundamental social contract.
The second part is that if you want something, you write it. If you can't, you support in some way those who can. If you can't or won't, you wait for them to do it.
Nothing more nothing less.
If you paid a distributor for a product, you can reasonably expect a bit more. Like a money back guarantee.
You can be upset or disappointed all you like, but I for one don't owe you anything unless you pay me for something. Neither does anyone who contributed to KDE.
Don't start attempting to impose your expectations as a 'fundamental social contract' unless you have a signed copy.
I'm serious. Developers don't owe you or anyone else anything at all. If you don't want to use what they write, that is your choice.
Posted Jan 29, 2009 21:48 UTC (Thu) by malor (subscriber, #2973)
Posted Feb 6, 2009 2:17 UTC (Fri) by pyellman (guest, #4997)
It's impossible to escape the conclusion that there was a cadre of people out there -- not novice users at all, but really, power/expert users -- who, despite knowing full well how the KDE team had characterized the 4.0 release, insisted as a matter of highlighting a "higher" principle (the ".0" principle) on installing it and treating it as what they insist a .0 release should be, and then using any negatives from that experience as ammunition and fodder to attack the KDE team -- all to grind into the face of the KDE team (and the rest of us) -- the consequences of messing with their beloved ".0" definition.
Posted Jan 29, 2009 2:06 UTC (Thu) by jspaleta (subscriber, #50639)
Was the developer only nature of the release mentioned in the release announcements on kde.org?
Does that read materially differently than the 4.2 announcement?
Posted Jan 29, 2009 4:04 UTC (Thu) by malor (subscriber, #2973)
They called it something other than what it was to trick people into testing it that ordinarily would not. They have explicitly admitted this. The facts of the matter are not in question.
The patch notes are irrelevant. Shipping .0 software and then putting in the README "oh, by the way, this isn't .0 software" doesn't change the implied promise, or the deception involved in releasing it that way.
Arguing against this is a stupid waste of time. They've admitted guilt. They just haven't admitted that it was wrong and abusive of their user base. And until they do, I'll be pointing people at GNOME, because that's a dev team that, at least so far, can be trusted to do what's right for users. I don't like their over-simplification of things, but at least if I install a stable release, I know they thought it was actually stable when it went out the door.
Posted Jan 29, 2009 6:01 UTC (Thu) by jspaleta (subscriber, #50639)
"Hiding behind the 'this is a mostly a release for developers' excuse is not good enough for me. Gnome is around for years, and the GNU project was not able to deliver an outstanding version yet. Is this the best that the GNU project can offer to the Joe User who wants to switch away from the commercial option of OSX or Windows? Well, nice try
I usually start my reviews with the positive points of a product and then continue with whatever I found as 'bad'. In this case, I just can't hide my dissapointment about the new version of Gnome. As a user, I expected more, and I want more."
Ah yes, that takes me back. Good times...installing Ximian's GNOME 2.0 desktop over RHL 7.3 and watching it bleed.
The only thing that has changed since 2002 and the GNOME 2.0 release..is that there are more users now to complain. I don't think the KDE developers handled communicating 4.0 release significantly better or worse than the GNOME developers did their 2.0 release in 2002...they just happened to do it 6 years later.
Posted Jan 29, 2009 6:29 UTC (Thu) by dkite (guest, #4577)
Something like the tides of the sea.
At least there is a choice.
Sometimes things have to be burnt to make it better. Xorg is doing
something like that right now with similar responses. Usually things get
much better over time, and we forget the turmoil.
Posted Jan 30, 2009 14:42 UTC (Fri) by nlucas (subscriber, #33793)
Posted Jan 29, 2009 13:54 UTC (Thu) by roblucid (subscriber, #48964)
You know, I just don't believe you!
The Developers weren't in the best spot, to take impartial decisions about the software. That was the distro's job,
There were Live CD's to try out, and it was clear many features weren't implemented. Those of whose who repsonded to the pre-release call for wider testing, have said, it was clear to us within an hour or two, that it'd not be ready for general use for a long time.
If KDE had been Fedora's main desktop, they'd have sorted the problems, so someone like Linus could have chosen to install the stable desktop, and play around with the new one when he felt like it.
Posted Jan 29, 2009 16:38 UTC (Thu) by malor (subscriber, #2973)
They've admitted this. It's not in question. They called it .0 to trick people into testing a product that wasn't finished yet. They're even proud that it worked.
THAT is what I object to. In some states in the US, they have statutes against 'theft by deception', and that's exactly what the KDE team is guilty of -- stealing time from people who trusted them, by misrepresenting the product.
Painful .0 releases are nothing new, but lying to get beta testing is. Above all other things, for free software to differentiate itself, it has to work. If free software is claimed to be stable, it needs to be stable. The KDE team's lies hurt everyone in the community. Most high-profile free software teams do fantastic work, and as a result, open source as a whole prospers. When big projects like KDE deliberately ship garbage to get more testing, they don't just damage their own reputations, they damage everyone's. And what if other teams see their evinced pleasure at all the free testing they got, and decide to start emulating them? The more that idea spreads, the more the quality of free software suffers.
This was a horrible idea, selfish and manipulative, and it should be condemned in the strongest of terms.
Posted Jan 30, 2009 9:08 UTC (Fri) by hppnq (guest, #14462)
Like others have remarked, as a Free Software user you should be aware that you receive a product WITHOUT ANY warranty. Read the license, always a good idea. At the same time, it must be annoying to do a distribution upgrade and see your desktop stop functioning.
Posted Jan 30, 2009 14:18 UTC (Fri) by roblucid (subscriber, #48964)
They didn't lie, they were very upfront about it. They didn't claim the release was "Stable" far from it, they announced a rapid development plan with frequent releases. They wanted the distro's to make KDE4 available, to encourage participation, and application porting.
Aaron Seigo, said clearly at the time, that making a 4.0 release was necessary to get this to happen, it was a very clear implication that quality was going to suffer. If you go back to Jan 2007 on this site, you'll see discussion on the very point. The PR train wreck was all too predictable, I think they could have handled it better, but it isn't their fault, the apparent blind enthusiasm with which the distro's pushed out the packages, without the stable KDE3 fallback, which received 2 updates in the past year.
Any body with any experience, knows re-writes of large projects tend to either never ship, or ship initially unstable and feature incomplete. This is one of the big reasons to prefer evolutionary, not revolutionary development.
There is a problem with version numbering. With a new project, releasing version 0.x.y signifies pre-production status. 1.0 is a milestone.
How does a widely used project redevelop? May be the project could be re-labelled, KDE4 or KDEng and then a 0.0 version number is a big warning sign to you.
Posted Jan 30, 2009 15:51 UTC (Fri) by malor (subscriber, #2973)
No, they weren't. They called it 4.0 explicitly to manipulate people into testing it that otherwise wouldn't. It wasn't feature complete, it wasn't done, but they gave it a stable release number purposely, and admittedly, to get more testers.
All they had to do was call it 4.0-alpha, and all would be well. No fuss, no muss, nobody would be upset. Once they got to feature complete, they could have called it beta.
But they didn't. They've directly copped to this: they tricked people into testing for them. They loudly insist that they weren't deceptive, but of course they were. They've stated that they knew that calling it 4.0 would get them more testers. They knew what 4.0 meant. Putting that release number has a very strong implication of stability and feature-completeness. Then, they put in the fine print, "Oh, gee, this isn't actually any damn good yet, and you shouldn't use it."
They did this to get people to test that otherwise would not. And, as you can see in the quote in the original article, they're proud that their lie worked.
If everyone started doing that, free software would be very badly damaged.
Posted Jan 30, 2009 17:14 UTC (Fri) by kragil (guest, #34373)
OK 4.0 should have been called "4.0 Developer Release/Foundations" or something, but other than that it was OK to release. That would have made the "KDE 4.0 is not KDE4" excuse more understandable.
But it did work and the libs were done. You cannot wait for 10 years to be totally perfect (just look at E17 ... that does obviously not work)
It had a lot of missing features and very few ported apps but people were able to use it productively AND IT WAS A GREAT STARTING POINT! (just look at 4.2 now)
1. Don't just call a release that mostly/only sets new foundations "$X+1.0" or people will whine for YEARS (like you do)
2. If you don't want distros to ship your stuff without fall back solutions make that very clear (mostly to Fedora devs)
3. Provide the OGG links for your blip.tv videos because then nobody can complain about a nearly perfect release [anouncement] (4.2)
Did I miss something?
I hope people will read this post before KDE 5.0 based on Qt5 is released ;)
Yes, there are something else...
Posted Jan 30, 2009 20:10 UTC (Fri) by khim (subscriber, #9252)
Did I miss something?
2a. If you preach that proper way is to install two versions of stuff at
the same time - at least make it possible.
All distributions which shipped KDE3 and KDE4 in the same box were
forced to heavily patch the thing. You can not just
"./configure ; make ; make install" KDE3 and KDE4 on the same system
(unlike GNOME 1.x/GNOME 2.x) - this makes all such talks hypocrisy.
Even this "feature" was unfinished (granted - it was in the same state as
everything else: mostly, but not 100% complete).
Posted Jan 31, 2009 15:58 UTC (Sat) by roblucid (subscriber, #48964)
There was an appeal for wide scale testing, before 4.0, giving
the reasons that the library redevelopment was quite good and stable now
and that they needed the applications to port, so they could finish
developing the new DE and get back to release discipline.
There was widely available info about the re-write, also the new Plasma
desktop, and everyone knew it wasn't finished.
The clear implication was that there'd been insufficient testing
The problem for a project like KDE, is it's hard to download the source,
just spend a few minutes doing ./configure; make; make install. They
needed a release infrastructure to make it feasible for most to try it out,
and submit bug reports. In practice that requires support by the distro's.
What KDE were, was naive about the process. Now I'm sure they see that
the approach to migration taken by developers of ext2/ext3/ext4 has a lot
to be said for it, to avoid PR problems. Especially as Desktops are an
area that hotheads find much more interesting than boring filesystems.
Posted Jan 31, 2009 20:59 UTC (Sat) by jospoortvliet (subscriber, #33164)
We didn't call KDE 4.0 that because it was stable. We did call it 4.0
because we did major surgery, and now our libraries were stable again.
We've been doing this since 10 years, and many other FOSS projects have
done the same. Like the kernel, gnome, Amarok and many more.
The fact YOU think 4.0 was, in any way, telling you something about the
USER, is your mistake. FOSS is about developers first, users next.
Distributions are for users, source code on some developer site is not.
Distributions therefor have to ensure what they ship is ready for the users
Fedora targets bleeding edge users - they considered 4.0 good enough, I
suppose. Mandriva did not, neither did Kubuntu and OpenSuse. Their choice.
Posted Feb 1, 2009 0:21 UTC (Sun) by malor (subscriber, #2973)
You're right. You called it 4.0 to get more testers.
We've been doing this since 10 years
I've been there the whole time, and I don't remember you guys ever before calling something 'done' that wasn't. Your .0 releases haven't always been that great, but to my memory, they've always been feature-complete. This time around, tou lied to us to get us to test your software before we normally would. That's new. And people are still pissed, a year later. This isn't coincidence.
FOSS is about developers first, users next.
The arrogance in this simple statement is breathtaking on two fronts. One is the fundamental belief that users are inferior.
On Linux, do you know what a user actually is? Almost always, a user is a developer of another project. And your particular software is very central to the use of their computer, if they chose your flavor of desktop, and if you screw it up, you damage the progress of other projects. They're dependent on you to get it right. Time they have to spend fixing your problems is time they can't spend fixing their own.
Further, it's worth pointing out that you lost Linus Torvalds, one of the most famous developers in the world, and yet here you're dismissively handwaving him away, lumping him in with the proletariat, the developers that aren't working on your project. Mere users. Scum.
Secondly, if you hadn't noticed, you're writing a desktop. If your focus isn't first, foremost, and always about users, then you picked the wrong hobby. Go write webservers or something. Every day you write code without thinking about users, users, users, is a day that GNOME eats a little more of your lunch.
They have come from essentially nowhere to gradually eclipsing you on the desktop. Eight years ago, only the diehard used GNOME. Today, you're in a substantial minority. This should be telling you something. And with 4.0, your focus on the needs of your own team, instead of the needs of your users, further accelerated your slide into irrelevance.
Distributions are for users, source code on some developer site is not.
This has never really been true. Remember: "users" are the people writing the kernel, too.
Your entire comment is damage control, apologia for an enormous mistake. Just admit the damn mistake, apologize, and move on. And stop lying to us. Maybe you'll start regaining some of the ground you've lost.
Posted Feb 1, 2009 11:53 UTC (Sun) by jospoortvliet (subscriber, #33164)
Well, ok, some parts of KDE needed testers, some parts were ready for that, and benefited from testing. In many other area's (plasma) we knew what the issues were, but didn't want to let all of KDE wait and suffer because it wasn't finished yet.
Again, it was a long-term decision which did hurt users in the short term but benefit all in the long term. You focus on a 1 year period (or even shorter). Linus went away for now - and so did many. Is that bad? Well, it sucks, but when you factor in the increase in speed of development, it is reasonable to expect them to be back. After all, the free desktop has what, 1% of the whole desktop market? So we did hurt a portion of that 1% to be able to get to the point where we could aim for the other 99%.
We're simply more ambitious than you think, I guess.
We can say sorry to those users we've hurt (even though I still think the distributions are to blame as well). But shouldn't the users say thank you now KDE 42 has proven us to be right?
Posted Feb 1, 2009 22:36 UTC (Sun) by malor (subscriber, #2973)
but didn't want to let all of KDE wait and suffer because it wasn't finished yet.
KDE is code. It can't suffer. Only your user community can experience pain. You inflicted a great deal of it on them, to benefit some abstract code. From other comments here, it sounds like your users got shut out of bugfixes and maintenance on 3.5 while you guys focused on 4.0. (I switched away when 4.0 shipped, so I haven't been watching that at all.) If that's true, you not only didn't provide a replacement, but also stopped improving the old stuff as well. Your focus shifted so completely to the project that you abandoned the actual users OF the project -- presumably, the original reason you started developing KDE at all.
So we did hurt a portion of that 1% to be able to get to the point where we could aim for the other 99%.
That is true, but it strikes me as shortsighted. You're punishing the people who trusted you, to go after the people who haven't. Your existing user base is your best advertising tool; their evangelism matters. When you screw them, you get people pissed off -- some of whom are annoyed enough to post screeds to LWN.
And if you're willing to screw that 1%, and it works, will you be willing to screw your 5% six or seven years from now? Why would people adopt your desktop when you're focused on your project, and don't care about their benefit?
I think you might want to collectively ask yourselves, "Why are we doing this project at all?" If that answer, and your ultimate focus, isn't on making the lives of your users better, every day, then you're probably in the wrong area for development. Users and developers would be well-served by avoiding dependency on your desktop and libraries.
You're going up against an entrenched monolith, whose user-abusing mistakes are legion. But they can get away with it, because they're a monopoly. It's those user abuses that, in many ways, prompted the entire Free Software movement. But you're not a monopoly. If you abuse users for the benefit of your project, you ultimately harm it more than you help it. You and GNOME both are tiny players, goldfish among sharks. If you're not obsessively focused on user benefit, then the other teams who ARE will take them away from you.
I'm sure you want your project to move faster, but no matter how good your program is, people won't take it up based on technical merit alone. Just look at Sony's decisions with Beta -- chasing off users they didn't like, porn-mongers, ended up being a huge blow to the format, eventually driving it off the market. If you continue doing this sort of thing, you'll end up with the best desktop that nobody uses.
Growth rates are always off how many users you have already. If you double your annual growth rate from 12.5% to 25% by abusive development practices, but cut your community in half in so doing, it'll take three years just to get back to where you started, and it'll take seven years to get back to parity with where your project would have been at the 12.5% growth rate.
The real numbers won't be that large, but carefully, carefully consider anything that makes a user switch away. Each and every one is a seed that can grow into more users -- and, if you get lucky, more developers.
Failing to water seeds you already have because you want to hike to what looks a bigger field on yonder mesa is a good way to starve.
Posted Jan 30, 2009 14:36 UTC (Fri) by boudewijn (subscriber, #14185)
Posted Jan 30, 2009 15:57 UTC (Fri) by malor (subscriber, #2973)
This is worth being upset about. It's a very bad precedent.
Posted Jan 30, 2009 18:10 UTC (Fri) by cry_regarder (subscriber, #50545)
Posted Jan 30, 2009 20:13 UTC (Fri) by khim (subscriber, #9252)
Posted Jan 31, 2009 21:01 UTC (Sat) by jospoortvliet (subscriber, #33164)
Posted Feb 2, 2009 19:49 UTC (Mon) by cry_regarder (subscriber, #50545)
1. The prev poster was implying that free software will get some sort of reputation because of things like KDE4. As if non-free software doesn't already have that reputation.
2. When I did customer support back in late 80s, the corporate policy was that if a customer for our product called too often and was belligerent to very carefully and politely introduce him to the competitor's product. Sometimes even with phone numbers for their sales department. The reasoning was that a troublesome customer is going to be troublesome for everybody...so he might as well be troublesome for your competition.
Posted Feb 2, 2009 20:48 UTC (Mon) by malor (subscriber, #2973)
Posted Jan 30, 2009 21:36 UTC (Fri) by dkite (guest, #4577)
Some problems are bad, some problems are good.
KDE had(s) the problem where people want to use their stuff. They like the ideas, the applications, the way things work.
I suspect the distributions wouldn't have paid any attention to 'improved communication' with the 4.0 release. There was palpable demand for the offering. The moment anyone of them advertised that the new KDE 4.0 release was going to be on offer, all of them had to do the same. If I remember correctly, the developers had to make an effort to tone down the enthusiasm.
All in all, the only thing it has convinced me of is that releases are the spawn of the devil.
Posted Jan 31, 2009 20:55 UTC (Sat) by jospoortvliet (subscriber, #33164)
Copyright © 2013, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds