[sheng]

I know what you mean and the amusing thing is that it used to be the other way around. Back then in the older days of Gnome2 and KDE3 it was possible to use them with other window managers like Beryl and Compiz. These again also provided a 3D desktop experience and in their later development stage they were well optimized and stable. They were the one who blew windows vista out of the water when it came to smoothness, resource consumption and hardware support.

[kasabali]

I agree and I'll leverage your comment to rant a bit so please forgive me :)

It's the old CADT [0] monster still at work. Windows desktop is the same old explorer shell which has been there since 1995? Only addition is dwm, which provides the modern graphical facilities. And Windows team has been constantly improving this same codebase for 20 years now.

While on the FOSS side, desktop developers are busy with the sisyphean task of rewriting or porting desktop shells and base applications again and again. OH, they've released a new version of graphics toolkit and the old one is deprecated. We must port the desktop now. Oh we also have to port all the applications now. Some applications were so hard to port that we've abondoned them and rewrite them instead. Actually, why don't we also rewrite the desktop shell while we're at it? Promise it'll worth it. We ported old apps but they're not consistent with the new ux paradigms of the new desktop so we'll rewrite them also... And just few years later when things start to calm down, there's a new release of graphic toolkit which will eventually deprecate the old version, ... oh not again.

Notice that I'm not talking about fragmentation or lack of resources. I'm glad we have more than one major DE, and I don't think more resources or less fragmentation would change anything regarding the situation above. Even with lack of resources, linux desktops had long long time to create really high quality products but they can't, because the ground below them is always shaking. Windows team on the other hand has been vetting, fixing, improving the same codebase for 20 years. They don't change the underlying toolkit every 5 year. They don't rewrite basic applications or port them to new libraries over and over (except new metro things, but they're seperate so my point still stands). Imagine what we would have now, if Gnome devs continued to polish and improve Gnome 2 and KDE devs continued to polish and improve KDE 3, just by using the same resources they've poured into Gnome 3 and KDE 4/Plasma 5 and their applications, respectively.

Again I'll emphasize, I'm grateful for the efforts that went into linux DE's, and I don't blame their developers for shit they have to deal with (although some are partly responsible for it)

That was a long rant and thanks for listening.

[0] http://www.jwz.org/doc/cadt.html

[sillygoose]

>> I don't think more resources or less fragmentation would change anything regarding the situation above

Well, as far as I can tell, lots of Linux people are hipstery special little snowflakes and fork things because some small thing isn't exactly as they want it, and that's where most of the fragmentation comes from.

Assuming that's largely accurate, why would it not be a problem?

What if we had, say, three major Linux distros instead of dozens of half-assed ones? How many tinkerer-distros does the world need? -As many as it takes to prevent Linux from ever conquering the desktop, it seems.

But seriously.. Please stop.

[kasabali]

I totally agree with you with respect to distribution fragmentation. Fragmentation there is much worse and definitely harmful at this point.

My above statement was specifically about the desktop environments, and DEs actually fit into your scenario. There are only 2 major DEs which have most mindshare, followed by handful of smaller ones. So there is not a bad fragmentation in that area (at least historically, there is now more diversity compared to 5 years ago), and the major ones have been with us for many years, AFAIK since late 90's? That's not only plenty of time to turn both of them into kick ass desktops, but also create wonderful ecosystems around them with high quality applications. Yet all we see is rewriting or porting everything every few years.

[keithpeter]

Not clear it is the number of distributions that is the problem as they tend to package the same libraries, desktop environments and applications. I agree with grandparent post that it is the churn in underlying libraries and the downstream re-work for no increase in functionality that causes that is the problem.

PS: Linux is widely used everywhere except traditional 'desktops'. Existing mainstream DEs are pretty good (KDE5/Gnome 3.16/MATE 1.8) and provide a range of choice.

[coldtea]

>They were the one who blew windows vista out of the water when it came to smoothness, resource consumption and hardware support.

Beryl and Compiz? I beg to differ. Besides a bitch to get running, and get running well, they were not at all comparable to Vista with regards to "resource consumption" and "hardware support".

[captainmuon]

This is just an anecdote, but I share the parent's experience, too. I bought a low-end HP laptop around when Vista came out (2006/07?). It ran with Aero Glass, but not very smoothly. But first compiz-quinn and then beryl worked perfectly, also with the same blurred transparency effect that was slow on Windows. Even when I enabled the crazier effects like wobbly windows, it remained smooth, and used not noticably more resources. The only place a slowness was noticable was when quickly scrolling (webpages, for instance). However, this is still a problem on any non-Windows compositing DE in my experience.

At some point, "they" started to rewrite compiz, and it went downhill, until it was so slow I could not use compositing anymore.

[dmethvin]

On the Windows side, there was a lot of tense discussion between Intel and Microsoft regarding Intel's popular integrated-graphics chipset being shipped at the time of Vista's release. The chipset didn't have hardware DirectX support and shouldn't have been certified, but Intel had promised their customers (like HP) the chipsets were fine and they were not anywhere near the end of the product lifecycles for either the chipset or the PCs. So Microsoft fudged the rules and a bunch of consumer-grade hardware ran like crap on Vista.