Cinnamon 2.0 released

The quick emergence of cinnamon has shown that the underlying technology is sound, and that the differences between the desktops were more a matter of politics than of technology, and only appeared at the top interface layer. Now it appears that Cinnamon developers are shouldering an increasing burden of maintaining infrastructure as well, which I don't see as a good thing.

I had always been hopeful that the best parts of Cinnamon, Gnome Classic, and Gnome 3 could be merged into a single desktop environment. If not by upstream, then at least by distributions, with the same reasoning that distributions tend to use system libraries, and sometimes a broader perspective might be required to get the respective upstreams out of their NIH syndrome.

I am not someone who knows very much about the internals of these projects, but I cannot help wondering about a few things:

• Why does it take a full-fledged fork of nautilus? Apparently, the nautilus developers removed some features which I never knew existed, but were apparently useful for some people. On the other hand, great features like spatial mode or rabbitsvn in the file browser stopped working, or stopped working properly, and now there are two file browsers which appear both as "Files" in my Fedora menues, are similarly deficient relative to Gnome 2 nautilus, and are started in a seemingly random fashion (automouting in Fedora appears to start nautilus, while the cinnamon taskbar starts nemo) and it turns out they just look the same to me, both distinctly inferior to the old nautilus. So if one spends effort in fixing nautilus, why not fix things like rabbitsvn or spatial mode rather than maintaining minor cosmetic differences?
• Why can't Gnome classic build on the Cinnamon taskbar which has working right-click (that's a killer feature of Cinnamon relative to Gnome classic), why does it need to cook its own inferior version?
• Why can't Gnome classic learn from Cinnamon regarding multi-screen support which Cinnamon get more or less right and Gnome still hasn't (last I checked after installation of Fedora 19 this summer).
• Why does vertical maximize appear only in Cinnamon (and hidden there in the options) when it is truly a killer feature (which already existed in fvwm in the early 90s, btw) for work on modern wide-screen monitors where a full maximize is neither pretty nor convenient?

One other thing which worries me about Cinnamon is that it seems to be adding "features" at an enormous rate. But it is really more important to establish good defaults (which Cinnamon does rather well from my perspective) and a limited number of options for items where there is a rational need for being able to change them. This is not just a Cinnamon issue, but requires a discussion across desktop environments to come to a small number of modes of operation which work across different environments and even operating systems: for a lot of things it does not matter which way they are done. In those cases, they should work the same independent of desktop environment. In other cases, a working paradigm should be identified and a clean switch implemented, where each paradigm strives for a best-practice no-surprises implementation.

To put the problem into different words: I believe the Linux desktop (as an ecosystem) suffers from attention at the wrong level. People fiddle with the window management layer when the real issues are at the application level on the one hand, and on the system management level on the other hand (which should be independent of the desktop environment, but here the various desktop environments all suffer from mission creep). Cinnamon, just to keep my rant close to the original article, is suddenly doing its own user management dialog for no user-visible benefit. Another example which irks me for a long time is lack of a good default document reader on any of the Gnome platforms: evince, whenever I tried it, has had stability problems and also lacks "trim margins" which provides such a usability boost that I remember after seeing it once, there was simply no going back to evince. The solutions is probably that desktop environments should get out of the business of rolling their own applications whenever they cannot provide a best-of-breed solution for lack of resources (in the realm of free document readers, that's clearly okular - it uses QT, but so what?).

Maybe distributions need to steer against this trend of desktop environments trying to do everything, badly. As users, for the better or worse, we are working in heterogeneous environments anyway, so consistency (or "branding") is only relevant relative to external, best-of-breed applications, not so much within the desktop environment as that is really orthogonal to the day-to-day user experience.