I'll have to start by invoking the Evil before trying to prove that nowadays there is a severe lack of direction in whatever concerns the open-source desktop.

Here's Gartner explaining why Windows is broken. I will only quote a very limited amount of text, because it's enough by far: «Microsoft's operating system (OS) development times are too long and they deliver limited innovation; their OSs provide an inconsistent experience between platforms, with significant compatibility issues; and other vendors are out-innovating Microsoft. That gives enterprises unpredictable releases with limited value, management costs that are too high, and new releases that break too many apps and take too long to test and adopt.»

This is why you should never trust the analysts! Yes, I agree that Windows is delivering "limited innovation", but what is what a customer needs: a product that fulfills some specific needs, or "innovation"? Are you normally going to buy "a sofa with innovation", "a car with innovation", "a computer with innovation", or a product that does its job the way it was supposed?

In most cases, what people need is better application software, being it about an office suite, an image or multimedia editing program, or a database server. Of course, they also need a stable, secure and dependable OS, but these requirements are not necessarily related to innovation — most of the time it's rather the opposite: pushing new features too early would result in unreliable systems.

I personally don't like Windows. I don't like some of the decisions taken by Microsoft in the past, I don't like their unfair marketing policy, I don't like a lot of things with them. But it's not about innovation: take a look at Solaris, at AIX, at the BSDs — all of them are terribly conservative, much more conservative than Windows! Yet we can't say they're "broken" — because they're not.

Then, if what we blame Microsoft for is the lack of innovation, how come that some open source people have found .NET such a marvelous innovation that they reimplemented it under the name of Mono? Mono is the most prominent Microsoft technology that has been adopted by Linux — and my faithful readers should already know that I deeply hate Mono!

Now it's time to switch to the open-source operating systems: are they chosen mostly for the innovative part of the picture? Once again, I don't believe this is the case.

Linux has indeed a fast-moving code base. As a kernel, there are many people to believe that starting with the 2.6 series, the quality is not what it used to be, each and every kernel being "unstable" by the old principles. However, the overall quality of the Linux kernel is still decent.

The most volatile changes happen in the may GNU/Linux distributions. While duplication and fragmentation is considered as being good by some (it offers you the choice), there are still too many Linux distributions that are not coordinated in any sensible way, with too many different ways of configuring things, which eventually leads "unpredictable releases with limited value, management costs that are too high, and new releases that break too many apps and take too long to test and adopt". Sorry, I quoted a text meant to be used with regards to Windows... mea culpa!

I won't comment on any of the BSD flavors right now. They're surprisingly conservative, and this should be a plus from an enterprise standpoint; this is only bad for people who might want to use *BSD for a home system.

So Linux is used for many other reasons, not necessarily because of the innovation pace. In my case, I am using it... despite its current trend to constantly replacing anything that just got stable enough with something new and unreliable, for the sake of innovation — or just for the sake of changing things.

I have been using Linux for about 13 years now, and I have to say that the time when I enjoyed it the most was back in 1995-1996. It was immature (maybe), it was too young (possibly) and buggy, yet it was still in its "romantic age", when the "religious" flamewars were not a real issue.

From that era, a friend took the idea that I was a fan of FreeBSD, and he was teasing me for years with that one. What actually happened was that I considered FreeBSD as a better choice for servers, and a more interesting operating system at some point. Even back in 1996-1997, I developed a certain feeling for NetBSD — and I still value that project. Overall though, I never was that "BSD guy" my friend took me for; it's just I wasn't (at that time) involved in the "Linux vs. BSD" flamewars. I was however more sympathetic about the BSD license than about the GPL.

There was a point in the history of Linux where things started to go astray. The KDE vs. GNOME, DEB vs. RPM, and all kind of "religious" wars started to be very noisy. Besides of that, things that I used to love were declared obsolete and not actually used anymore in Linux — think of OpenWindows and olwm (OPEN LOOK Window Manager), think of the CDE/Motif (Common Desktop Environment), and so on.

Up to a point, GNOME was not really usable (not before 2.0, or actually 2.4), and I was not charmed by KDE. I was still liking FVWM and WindowMaker.

Red Hat turned to a different business model, including RHEL and Fedora Core — and this is when I thought Red Hat will become a second Microsoft. Quite sad... but Slackware was there, and it was good enough.

Then, all of a sudden, Ubuntu 4.10 converted me to GNOME! Also unexpectedly, Slackware dropped GNOME one year later. Very disturbing in my case...

At some point in 2005, I started to blog, so I will skip the sinuous way across the distros. I will only note that I was at some point almost pleased with KDE3, but since KDE3 has no future — the future is KDE4 —, I can't see how it's a good choice to invest in.

As I said, I have used a lot of distros in all this time. A lot. I have been enchanted and deceived, I have witnessed various unexpected changes of policy with many of these distros. I almost felt the frustration of using Windows, when the developers decide to go one way, and nobody cares what the users would actually prefer. It's only that the most disruptive changes were introduced by Microsoft only twice as far as the OS is concerned: with Windows 95 and with Vista. With Linux, important changes happen all the time and several times a year.

Nevertheless, the same way most people are "captive users" of Windows, I am also a "captive user" of Linux. I wasn't modifying and rebuilding a package very often, but having the power and the freedom to do that (because of the availability of the sources) is worth the pain.

Sure, it's a love-hate relationship sometimes. It's also a matter of reacting to the changes: once I have found that Ubuntu 5.10 is buggier than 5.04 and 4.10, and that it induced regressions on my hardware, I switched to SuSE/openSUSE completely (at some point I was running both Ubuntu and SuSE at home). Once they managed to lose the focus and to experiment with multiple package managers (all broken at some point with), I ditched them. Since they are the one behind Mono and since the deal with Microsoft, there is no chance I will ever return to them.

Several disturbing issues happened in the meantime. I used for different periods of time all kind of Linux distros, and I also experienced some bugs. The logical thing to do (when you're not actually fixing the bug too) is to report the bugs — to the distro's bug tracking system or to the upstream project. For some distros, the only way to discuss bugs is by the means of a forum — fine too, as long as it works.

The really shocking issue is the way the developers react to a bug report.

Most of the time, and for most of the bugs I reported in various places, a neutral attitude seemed to be the rule. The developers or the maintainers were not angry, nor offended. They noticed the bug and they just did something — or nothing at all.

In some cases (and I don't want to give names), my reports were actually... enjoyed. This usually happened with "small distros" (i.e. having a very small number of maintainers) or small projects. Those people were almost glad to be notified of a problem they were not aware of, and in many cases the fix was trivial or easy to figure out.

For God's sake, here are the facts: my bug reports were always correctly formulated, in a very normal language. The occasional outbursts of rage were only triggered by a very offending reply... from the other side!

Yes, I am using a strong language at times. It's usually very strong in some of the very few posts written in Romanian, it's mildly aggressive in the regular English-language posts, and it can be very strong again in the comments — but only as a reply to another comment who is insulting me!

This being said, I urge you to show me a bug report that was started by me with inappropriate language.

Two days ago I reported a bug in GNOME's Bugzilla, for the last time: Bug 527148 – Users don't know what the fourth, fifth and sixth levels are! While it was eventually accepted by someone and moved through cloning into Bug 527298 – Document fourth, fifth and sixth levels of keyboard layout, please note the way the discussion went.

Some GNOME developers are suffering of a disease I would call "the FreeBSD superiority". The symptoms of such a disease, when it occurs in a developer or a maintainer?

 The acute feeling that everybody should be grateful for benefiting of such a great software for free, so how do they dare to file a bug for such a petty issue?!

 Of course, all the issues are petty as long as the developers is not himself annoyed by the same bug.

 Users are stupid by definition. They don't even know what the want, what they need, what they run, what they see, what they experience. Should they want something better, why don't they wrote all the patches by themselves or fork the project?

 I don't care about usability and ergonomics, I am a developer, I am always right! (So no, your bug is not a bug. NOTGNOME.)

The irony makes that GNOME is a project that has been accused of considering the user stupid because it arguably doesn't allow for the same degree of configurability (or customizability) as KDE does — or at least not from the GUI. If GNOME is "for dummies", how can they say there is no need to explain what are some options meaning, as long as there is so few people knowing what they mean! After two decades of dealing with computers, I don't know what a 4th, 5th or 6th level is, and I couldn't find any official public documentation. If Linux is about learning (from a header file, from a source file, from a configuration file, from a man page, from a how-to, from an info page, etc.), and there is no such documentation available in a particular case, how can you say "I don't care, I just ship the software with the strings I've got from somebody else"?

They can do that, and they do that. In the process, they blocked my Bugzilla account, just to prove their lack of common sense. Being a member of the GNOME Foundation gives you power, right?

When dealing with personal opinions of the developers, your life can't be dull; here's the delightful comment #8: "What a real moron that guy is, pardon my French... This is actually NOTGNOME."

I guess a short lecture on common sense is needed for the idiotic developers, please excuse my Greek. This issue is very much like in the following imaginary situation: suppose you have a high-tech TV set, and you discover in the on-screen menus some entries written in such a way that you're simply unable to understand what they refer to. Strangely, all the other entries are perfectly clear, as the makers of those appliances are known for only providing straightforward menus, settings, buttons, whatever. Given that there is no mentioning of the new menu entries in the manual, what do you do? You call the tech support. And what do you hear? "This is actually NOTOURPROBLEM. We simply inherited these strings from the maker of the new I/O controller: we took them from their catalog, and it's not our responsibility to make them clear to the user. Yes, we know you're using our TV set, you have bought a TV set and you're expecting it to be usable, but blame it on the I/O controller and its lack of documentation, not on us."

In the real life you won't get this kind of answer from a manufacturer's hotline. You won't even get such an answer not even from the wrongdoers at Microsoft. But you will get it from GNOME developers!

As a collateral damage, I would point you to a (non-)trivial speed bug in Brasero 0.7.x, reported upstream as GNOME Bug 525501 – Brasero displays odd burning speeds (1-3-5-etc) instead of even values (2-4-8-etc) by yours truly.

I will obviously be unable to comment or report anything on the matter... with a disabled account!

This bug may be simple or not that simple. Facts are that between Brasero 0.6.1 and 0.7.0, some portions of code were rewritten from scratch, and this includes the new mechanism of guessing the maximum burning speed. With my LG units (two different models), Brasero 0.6.1 was just fine, but Brasero 0.7.x is only allowing for odd speeds, which are actually failing.

I couldn't find complaints about the speeds in 0.6.x — except for not honoring speeds that are not supported, and this is only normal regardless of the version, as long as Brasero (and GnomeBaker) doesn't (don't) check for the actual speeds: the algorithm is to check for the maximum speed, then step down with a 2x step, and let the user choose the desired speed.

The new way of guessing is uneven and may lead to odd speeds: in some cases, for CDs, in some other cases, with DVDs. This new bug is definitely a shameful one, as there are several ways of patching or fixing it:

 My own dirty patch that works for me (but it is not the right one): changing the parity of the speeds.

 The trivial way: check if the speed to be displayed is odd; if so, increment or decrement it by one, so it would always be even. This should take 1 to 3 lines of code, depending of what extra checks you make.

 The academic way: re-implement the speed checks so that they only include the actually working speeds, possibly by using the principles that work in the case of K3b. Another inspiration could come from a much older code: X-CD-Roast (used in RHEL).

It's simple to provide with a fix, but it's already too late: Ubuntu 8.04 and Fedora 9 will ship with a defective Brasero, as it will fail on some media burners, most likely on all the units manufactured by LG.

I might find some time to play with it and to re-implement the speed detection, even if I only own LG units. Will I be able to send the patch upstream? No. Would they accept it for review and possible inclusion? No. Is this open source? Yes. Is it about collaboration? It depends on who's asking. Should I fork the code for such a small change? No, I don't think it's worth the pain.

A second shock was something I have just experiences, following the article Xfburn and the unbelievable mess in XFCE.

The core issue is that Xfburn is overlooked by most if not all the XFCE-based distros. Surprisingly, it works for me, but I could only find binary packages in Mandriva and Ubuntu (from the distros I would use, that is).

Even more worrying: the Ubuntu maintainers have took the sources for Xfburn from the developer's site back in 2006, and the sources are not there anymore. As the Xfce.org site had some bad URLs with regards to Xfburn, I filed some bug reports: Bug 4000 - Xfburn page points to the wrong URLs; Bug 4001 - Xfburn inconsistently part of Goodies (also: Bug 4002 - xfce4-artwork missing / highly incomplete, on a different topic).

I assumed that most people would discard an application when they see the application's developers and owners don't care about it, don't provide working links, etc. Why should anyone use something that is not valued by the original author?

The problem is that the reactions were atrociously violent and negative.

At first, a famous and respected Slackware guy was offended to see his repository mentioned in the list of the 3rd party Slackware repositories. While I have listed all the well-known repos to prove that nobody has ever built an Xfburn package, someone took it as an offense. Weird guys... are these the guys we're so proud of, for being the open source guys?

Then, some other bad replies followed, I won't narrate everything.

I will touch however the second XFCE issue, as it's equally important: as it turns out from the comments posted by some XFCE developers to the three bugs — and also from the unfortunate comment by Stephan Arts at the original article, an injurious comment that asked for an injurious reply —, there is a major problem with XFCE, a problem I wasn't aware of.

Yes, they don't care about their project!

They do list Xfburn in the Projects page (one click away from the main page), yet they declare it "non-existent, because it was never released!" How funny, some distros and some people are using since 2006 something that has never been released!

If it's about the version (0.2.0 and stalled), or about the fact that it wasn't maintained until it was recently "restored", then those people need to be told that:

 PackageKit is only at version 0.1.11, it's young, buggy, moody and incomplete, yet it's used in Fedora 9.

 X-CD-Roast is still at version 0.98alpha15 (since ages), it's practically unmaintained (although there is some recent website activity), it's ugly, it's infringing all the HIGs, yet it's used by Red Hat Enterprise Linux as the one and only non-KDE media burner!

I have also found that (according to the XFCE guys), "there is no official XFCE media burner, there is no official XFCE archive manager (although they have even two), there is are official XFCE Goodies, they're only some projects you can use or not", and the like.

You should then go one more click away from the main page and read the Xfce Features page: "Finally Xfce provides end-user applications, always designed lightweight, which are: a simple and fast text editor with printing feature, an easy-to-use file-manager, a media player based on Xine, a CD's and DVD's burner, and an archive manager supporting multiple compression formats."

Obviously, they don't exist, do they?

I was extremely naive when I wrote the not-so-structured (but positive) What is XFCE still needing to become the #1 choice. I thought that they do care about their project. I thought I was summing some suggestions. I thought this would give some food for thought to some. I thought I was promoting XFCE.

I also thought Google Summer of Code, XFCE and the lack of a vision is raising an alarm, so that maybe someone will realize the difference between what things are versus what they can be.

I also was terribly naive to promote XFCE by the means of some posts including but not limited to: Choosing an XFCE-based distro; Mandriva's XFCE "One": rather sooner than later?; Fedora 9 XFCE LiveCD: a serious task; XFCE and Usability: Fedora vs. Xubuntu; Fedora XFCE: 8->9; Links of the day (somehow XFCE-focussed); and more, also counting the posts about Wolvix, GoblinX, Vector and Zenwalk — even from the old blog, in times when I was positive about Zenwalk.

There is a saying in my country, and in free translation it gives the following: "The well-doing is motherf*ing." The actual moral? You shouldn't care about anything, especially when that thing is neglected by the person who should care the most about it!

In the case of XFCE, the obvious lack of a strong leadership, the chaotic governance, the lack of a direction, the missing PR, the way they say "oh, this is not ready and it might never be, why are you interested in using it?" is simply stunning.

Instead of appreciating when someone tries to find a meaning there where it lacks, when someone tries to suggest some ways to enhance the visibility, to improve the coherence, and to progress in terms of market penetration, what do they do?

They bother to report a headache. Very funny.

This is not the place to insist on the fact that I very much dislike the contamination of GNOME with Mono. I am very unhappy about that, it is absolutely abnormal to add to a C/C++/Python-based system a completely new and so "un-UNIX-ish" layer just for the sake of a few applets or small applications — I have written about that in a private mail posted with permission as The Overlooked Issue of Development with .NET in GNU/Linux, and also in posts such as Mono is FUD, and many others.

The reason for using GNOME in my case is related to the relative cleanness and simplicity as compared to KDE, then it has to do with the default layout that (still) comprises two panels (unless some distro decides to revert it to the pre-2.0 times and to make it a single panel, just to be on pair with KDE and with Windows), and it also has to do with the KDE tradition of using toyish themes and icons, huge visual elements (even worse in KDE4), and other specific differences I won't detail here.

As long as I will still be able to use GNOME without having Mono as a mandatory dependency, there are good chances that I will continue to use it. Even so, it's annoying to have to remove Mono, Tomboy and F-Spot from some popular distros who come with them installed by default (the LiveCD editions are prone to that).

I would then say there is a certain way some projects are moving forward that I will name as "pissing in your own soup", practically having three main consequences: (1) you sabotage your previous work by following a certain "evolution" path; (2) you will make unhappy a lot of users who were never consulted about the upcoming changes; (3) you will make some users even more frustrated, as long as they have trusted you and they're "captive users" of your product.

I will turn one more time to the Evil to come with the first example. When Microsoft has upgraded Visual Studio from 2.x to 4.x to 5.0 to 6.0, the changes were not that important as far as usability was concerned; they were less disruptive than the switch from 1.5 to 2.x. But then Visual Studio 2002 was released and, regardless of the new kid on the block (.NET) that eclipsed C++, and despite the claims that "compared to previous versions of Visual Studio, it has a cleaner interface and greater cohesiveness, it is also more customizable with tool windows that automatically hide when not in use", I was personally horrified by the disruptive changes in the GUI's behavior. Simply put, I have found it totally unusable, and the feeling persisted for more than one year!

Instead of being of help, the self-hiding dockable windows were never hiding fast enough when you needed that to happen, they never show up fast enough when required, they could also refuse to dock back to the position they once had, the GUI had some usability regressions with regards to editing of the resources, and so on. For an experienced user of Visual Studio 6.0, the drop of productivity due to the upgrade to Visual Studio 2002 was in a specific case astronomical: some tasks required ~300% more time to complete!

As a general rule, this regression was more prominent for people who prefer to have the code as text, not to dig for it behind the controls, so it's not about VB programmers (it's about C++ guys).

We will reach the open-sources similar examples right away.

Thankfully, when open source rhymes with "UNIX-like", this usually offers some kind of protection against sudden changes that would have a momentous impact on users. Oh, sorry, the correct phrasing should be: "this offered" some kind of protection.

KDE3 didn't have the best look for a "business desktop", it suggested a playful environment, not a productive one. Furthermore, the 3.5 line was not offering any major "innovation", because KDE4 was under design. As Qt4 was radically different from Qt3, the best way to use it was to rewrite a new KDE from scratch, this way also eliminating some initial design flaws.

The problem was that KDE was originally designed in times of more sobriety and much more common sense. Nowadays, people were already influenced by the SuperKaramba way of seeing things, and the general trend of the society is to prefer the big shiny stuff to things that would have real meaning. Also, instead of trying to design an environment that would be "great by the definition of the term", they designed it to be "greater as in better than Vista".

Fact is that KDE4 is ugly by the standards of a man who has been interacting with all kind of computers for more than 20 years. When something is big in KDE4, then it is HUGE (try the clock). Usability-wise, it seems to have been designed for the casual user (or for the huge screen resolutions), because there isn't very much space on the taskbar — and no, I never want to group the windows by application! Even supposing it will reach a miraculous stability overnight, KDE4 is utterly unusable for me.

Vista is a failure, but KDE4 is uglier than Vista. Go figure. What a pity... because Qt4 happens to be the best toolkit in the known Universe...

And this is how the KDE team managed to pissed in their soup: people who won't be tempted by KDE4 might be reluctant to choose KDE3 too, because KDE3 is a "dead branch". It will only see security fixes, some bug fixing and minor enhancements... for some time, it's not clear for how long. Following the glorious tradition of Microsoft, the company who bears an acceptable OS (Windows XP) for a failure (Vista), KDE is bearing KDE3 for a brainless KDE4.

What is supposed to do a user who finds KDE3 acceptably good (yet far from perfect) and KDE4 totally unacceptable? (By the way, what will do the XP users when Vista will be the only available choice? Switch to Mac?)

To be honest, I am very critical with GNOME too (see GNOME: 10 years of self-congratulation and Mono? Mono!), however GNOME is still a usable project, despite the lack of a good official media burner (Brasero is broken), of a UML modeler, of a decent office suite (GNOME Office is a joke, and I still remember how many years they were not interested in having such a thing), of an official IDE for years — Anjuta is part of GNOME only since GNOME 2.22...

GNOME lacks a coherent vision (where are they going? is anybody who could give an answer to that?), but for the time being it also comes with useful improvements, not only with Mono...

Pissing in the very own soup is something that happened to Pardus Linux. The Pardus-related section of my old blog includes some 39 posts, so you can't say I wasn't charmed by it (and its package manager). But here's the reason for dropping Pardus: Security Model Quiz: going wheel. It's pretty straightforward. And a sad story.

The next project that lost the credibility in the eyes of a responsible person (that's me, surprisingly)... drums... Zenwalk: the last goodbye: «By pure chance, I happened to read 2 threads on Zenwalk's forums: Netpkg crashing, reported on Feb. 18.; Problem with xnetpkg, reported on Feb. 19. Invariably, JP ("hyperion") replied: "Please upgrade to netpkg 4.0, netpkg 3.7.5 is no longer maintained." "Please upgrade to Netpkg 4.0 from snapshot. You are using a deprecated version." Wow. Let me rephrase it: The version of the official package manager from the "stable" brach (current) is considered by its developer (which happens to be the benevolent dictator in Zenwalk land) as deprecated, so that users should switch to the one in the "testing" branch, snapshot! The bugs for the version of the official package manager from the "stable" brach (current) are not going to be addressed! People should instead switch to the unfinished, new version, supposed stable but unreleased yet, being it only for it lacks at least one feature present in the previous version! I can put it even simpler: JP has made a cake. A few years later, the cake is bigger and bigger, and there are more and more people to prefer it. Now, JP — because he's the chief cook, right? — is simply pissing in the cake he made.»

Have the project leader give the message that you won't maintain anymore the stable version of a key component of the system (the package manager), and tell users that they should use the unstable, unreleased (and rewritten from scratch) version: a "winning" solution.

Moving on: last month I gave a try to Frugalware (after a very long pause: Frugalware 0.5pre2: promising a good 0.5 release; Skipping Frugalware 0.6 in 5 minutes), and the result was an honest post: A Frugal experience – Three days with Frugalware 0.8.

You can find thousands of disparaging, moronic, unfair reviews for almost any distro that is available. I judge my "review" (it's not a review, but the story of my experiences) very fair and correct. But then the founder of that distro was off-limits, and I noted: «You can never get rid of the superior attitude of the developers or fans who only want people to praise their beloved distro. On Frugalware's IRC, <vmiklos> (sadly, Miklós Vajna, Frugalware's founder) is so dumb to say: "so a distro is wrong because the default dpi setting is not what he wanted." Yet another guy who is pissing in his own work (just like Zenwalk's founder): no, dear stupid, it is not what I want, it's what it should have been! If you're f-ing unable to determine the "correct" DPI, then f-ing let it be the traditional GNOME (pre-2.20) and Windows (pre-Vista) value of 96 DPI, it will always do. Screwing it to 72 DPI and than complaining of my complaint... say, if I would have filed a bug, what would have you done with it?»

I will stop here with the listing of the Linux distros that purposely fail to stick to the proper QA standards and fail to have the proper attitude that would suggest to the users that they should trust them to be reliable enough for the years to come. Windows is closed-source and it's free to fail, but why should the self-shooting in the foot happen to Linux distros and to desktop environments?

FreeBSD deserves a special mention here, as no matter what you say about it (I suppose you shouldn't criticize anything, no matter what!), you're a moron in the eyes of some of its developers — there is at least a FreeBSD committer who specifically wrote "Béranger is an idiot" while I wasn't saying anything wrong: The missing steps for a BSD world domination. Updating Firefox 2 and FreeBSD 6.2 was also decent, reasonable, rational and legitimate report, but not for the BSD Übermensch.

They will never increase their market penetration with this kind of attitudes. (Yes, I know, they don't even want to: just like OpenBSD, they prefer not to be used by "stupid users".)

I got a hint through e-mail that the way the developers of a certain project are seeing my intervention is of this kind (the text is translated very roughly): «Maybe it's not in your intention, but the style of the posts is always negative and not constructive. Attention, I don't say you shouldn't criticize, but you shouldn't only do that, plus you should also accept critical replies. My feeling is that you complain, but you don't do anything to change the situation. You expect everybody to understand your rationale and to agree with you, which is obviously not the case. When your blog was mentioned in our discussions, I have to say that the tone of your posts made that they was practically ignored altogether. Some of your ideas are interesting, but unfortunately they're quickly flooded with a discourse where you put your vision of the things above all, practically insulting ours. I suppose you won't like these words and you might write a post about how arrogant and morons we are for not listening to your advices and not following your enlightened guidance. Or maybe you would change your opinion and your approach.»​​​​​

Interesting. Sensitive people. Suppose I was one of their developers. And suppose I had the same ideas about where the project should go, buy my opinions were only mine. How would this be different than the current situation? I am merely a netizen who expresses his views and doesn't bother to fill bugs that are actually a WISHLIST, not "real" bugs.

Then, let me list a few items from a virtual wishlist, items that are actually simple to solve, but nobody would actually do this work. Furthermore, should I be able to implement these requests by myself, do you believe the projects would just accept my changes? You must be kidding.

Here's the list of the tiniest usability improvements that comes into my mind in this very moment:

 In KDE3: Konqueror lacks a keyboard shortcut to toggle the view of the hidden files (CTRL+H in Nautilus). It also lacks keyboard shortcuts to toggle between views (CRL+1, CTRL+2 in Nautilus), or at least those shortcuts are not listed in the menus. This diminishes my productivity with Konqueror.

 In GNOME: Nautilus doesn't have tooltips when the mouse is over a file, not necessarily the way Konqueror has (with previews), but at least the way Windows Explorer has, with at least the file's relevant information. This is annoying especially for the files on Desktop.

 In XFCE: this seems to be the only desktop environment under the sun that prints the desktop icon labels opaquely, and this is so à la Windows 95! Sure thing, there is a way to make them transparent, but then the result is ugly. The real solution would be to draw the text like in the other desktop environments, id est with a drop shadow that would make it more legible.

As I said: even as a developer I might have ideas that would not be accepted. What should I do then? Fork the project?

Back to XFCE, I can add that the dialog to manage the keyboard layouts is a hack I only saw in Xubuntu 8.04 and nowhere else, and I am also adding that the XFCE team refused to consider the Ubuntu/Mandriva/Frugalware "hacks" that add hibernate/suspend to the logout dialog... because it's not OS-independent! (Hint: try conditional compilation.)

Open source, but how about the minds?

It just occurred to me: the open source developers and maintainers are not accepting a blog to use the polemic journalism style. How about that?

If open source is not about love, then what is it all about? Compromise? Pragmatism? Should a user "just use it and shut up", or rather think that the only freedom gained is to have the access to the source code (something that's impossible in Windows), and the only way to change something is to change it for yourself, something that practically equates with a forking?

Please, don't answer. I have that feeling that the times when the open source developers were not so easy to offend are long gone, and that the only projects that still have "that classical scent" are those who are actually not in active development! Say... WindowMaker, with the latest stable version is from mid-2005, and FVWM: the latest stable version is from end-2006.

From now on, the reasonable choice for me is to use what works, no particular love being involved.

I can't use KDE4 because it's unpractical. I would however use Umbrello from KDE3 if it's the easiest available modeling tool.

I will use GNOME as long as it doesn't become dependent of Mono. I don't know yet what to say about XFCE — there are many interesting XFCE-based distros, but it's so discouraging to see how they don't care that much about the way the project is perceived...

After all, using an OS eventually goes down to using the applications you need. Let's make an exercise, even if you can't understand French. Here's Linux Pratique Essentiel 1, and here's a preview of all its pages. You should be able to understand the titles, and the magazine looks sympa and useful to promote some of the applications available under Linux (I might look for it at the newsstands in Brussels, where I will spend the second half of the next week).

Now scroll down to see the pages 26 to 30. The green block at page 27 is dedicated to... Xfburn, the "unreleased" project! (Boy, I like this.) The green area at page 29 presents Ristretto, another "not ready yet" XFCE project. At page 30 there is Squeeze, another "unofficial" Goodie. Nice stuff, eh?

Application-wise, I will have to live with this paradox: while Qt4 is the best toolkit, there aren't many Qt applications that are only Qt-dependent; most of the Qt3 applications require KDE, and most of the Qt4 applications would require KDE4 (QDevelop is one of the exceptions).

In contrast, one can find a lot of food for thought on GnomeFiles, as many of the applications are not actually requiring GNOME, but only GTK+. Random choices: Pyrenamer — homepage; Thunar Shares; BabyChess.

Distro-wise, what are we left with once we replace the "need for love" with the need for something that works? I am short of time and I haven't say a word about the difficult task of choosing a distro to trust for the next 1-2 years (at least). There are so many distros, and still...

Pragmatism has limits in my case. I won't use any of the distros where I had major conflicts with the main developers, or where they have simply sabotaged the trust one can have in those distros — this would mean "bad karma", you know.

As a special case, I couldn't possibly use openSUSE wholeheartedly, for a few strong reasons: the deal with Microsoft, the pushing of Mono into GNOME (and in Linux at large), the authorship of the "new" Kicker and of the "new" GNOME menu and control center, the overall feeling that I am using Microsoft Linux — the Linux that runs you, and not the other way around. This is indeed a special case of lack of vision: whatever was (or still is) good in openSUSE (and in SLED/SLES) is now poisoned with the fine touch of Microsoft and the Microsoft-like decisions, those corporate decisions where what the end-user actually wants is not important.

Some feeling of moral wellness is still required even when love is not an option.

It is however much easier to choose a Linux distro for a server usage than a distro for your desktop or laptop. For a server, anything like RHEL (or a clone of it), Debian or Slackware would fit, and many other options are good choices too. For a desktop/laptop usage, the requirements are higher for many people, including myself.

I would anyway avoid any distro I don't trust to provide with security patches in a timely manner, or a distro whose main purpose is to please those whose major excitement is that Compiz is Cool – and Why That Matters. After all, I still stand to this statement written in a moment of high frustration. An older and less passionate Requirements For a Distro of Choice is partially applicable too. It's somehow worrying to notice that more than one year ago, in times where I was deciding I could use KDE (KDE4 was not released yet and it didn't show up yet as a threat to both the common sense and to the future of KDE3), the choice was not so easy for picky people: Development roundup: 500+ distros, and none of them...

Today, I am using newer hardware, and all of a sudden I discovered that my laptop's microphone is not supported (and in some cases I can't use an external one either) under the latest releases of many distros! I have not updated that page (e.g. it doesn't say that the mike works under Fedora 9), but this still means I have no microphone under Debian (even testing), RHEL5, and Slackware derivatives (maybe -current/12.1 will change that, I'll have to test it). Quite a shock, to be forced to skip some major distros for the sake of a bloody audio controller!

Should the hardware compatibility be no problem, I could have used Mandriva 2008.1 too, but unfortunately it freezes randomly with my Intel 945GM-based video adapter. Frustration and regressions are the words of the day.

As I haven't tested "exotic" concoctions such as Draco GNU/Linux, I am currently left with a very short list of the most convenient possible choices:

 Ubuntu 8.04, despite of the current regressions, because it will be a LTS release above all, and it has plenty of available packages (not all of them broken).

 Fedora 9, because I still have a tendency towards Red Hat, despite yum & friends being much slower than apt & friends, and despite of its shorter supported life cycle.

Fedora 9 is likely to be the major scoop of the next months. I am not sure if it will prove to have a vision or to lack it, I couldn't tell if it's about hard work or about rushed new features, I don't know if it's a mere imprudence or the key to success, but I am certain that the many major improvements make Fedora 9 the topmost innovative Linux distribution of the moment, outperforming by far any competitor, especially Ubuntu!

Fedora 9 is disruptively new under the hood, and this is frightening for people who prefer things to "just work". Will its users be guinea pigs, or satisfied beneficiaries of Software Libre?

As it's not released while I am writing this, I can only rely on my experiences with alpha, beta and post-beta rawhide experiences. Given the very fast-paced rhythm of the changes, I have had some not so happy experiences with it, summarized in Mostly (but not limited to) Fedora 9, Fedora 9: Should I wait or fear the release?, No, F9 won't be a great release.

Fedora 9 might actually be a great release after all, but I have some personal worries, so I need to share them with my readers.

Users coming from RHEL5 risk to be shocked when confronted to Fedora 9 as system administrators. Changes for the best are good, but too many changes at the time can kill you. Voltaire is quoted to have said in 1772: "Le mieux est l'ennemi du bien." ("The best is the enemy of the good.") Maybe the open source developers should try to understand why Voltaire had to come with the first version of the modern saying "It it ain't broken, don't fix it".

I very much appreciate the "new" system-config-firewall in Fedora 9, but its over-complex GUI is confusing. I very much appreciate the "new" system-config-services, but its over-complex GUI is slow to populate!

Then, what was wrong with PAM and consolehelper? Was it absolutely necessary to rely that much on PolicyKit? (For the "traditional" style, see Consolehelper Quick HowTo or How to run a program from GNOME menu with root privileges. And this was already one step ahead of kdesu/gksu/gksudo/gnomesudo.)

Here's how I discovered what was that strange feeling I developed at one point, and it was all about... PolicyKit!

I was all of a sudden aware that I was able to install everything without being asked for a password, not even the next day: it was that bloody PolicyKit... used with defaults that made me feel somewhere between "like in Pardus" and "like in Windows"!

Simply put, instead of the classical GNOME Keyring behavior to keep your credentials for a limited period of time (I guess it defaults to something like 10 minutes) and to make it visible to you through a systray applet that allows you to revoke the credentials at any moment, the new behavior in Fedora 9 is like this: when you have to enter the root password for a given operation, say "Install package(s)", the pop-up dialog that asks you for the password defaults to an infinite period of time, unless you specifically check the "For this session only" checkbox!

When Joe Sixpack is installing, or removing, or whatever-doing for the first time since the installation of the system, he's trusting the defaults, and just enters the password and hits ENTER. As a result, NOBODY WILL EVER ASK HIM AGAIN FOR A PASSWORD FOR PERFORMING THE SAME KIND OF OPERATION!

This is a security issue. I don't have kids, and I might have various reasons for letting family of friends to use my account. However, to protect the system from dangerous operations, there is now one more thing I would have to do with Fedora 9: to check with polkit-gnome-authorization which are the administrative operations for which the system is never asking a password!

Is Fedora 9 trying to become less secure than Vista? I certainly hope this is not the case, but we'll have to see what happens.

As far as the "traditional" pre-PolicyKit mechanisms are used, I very much value Mandriva's approach: for updating the already installed packages, the notifier is asking you for the user password, while for adding new packages or for removing packages you'll have to know the root password. Clever and security conscious, eh?

I guess I was not able to prove that there is a lack of vision as far as the open-source desktop is concerned, but I couldn't see any clear direction nonetheless. KDE4 is trying to mimic Vista (currently with success in terms of crashes), GNOME is self-sufficient and Mono-ized, XFCE is weak-willed...

The Linux distros are themselves many, merry and continuously improving, but they're regressions-ridden (both with regards to hardware support and to end-user applications), which is not so funny. Back in 1996, all my hardware was fully supported by Slackware 3.0 and by Red Hat 3.0.3, but I can't say the same for today.

If you missed the message, then this editorial was not for your eyes.

Sincerely,