I didn't ask though how can I power off my computer but where are the common buttons which do the trick...
I KNOW it's the power button! For gods shake!
For my father wasn't so obvious though and he asked me. I don't know if he is idiot or not but for sure he is not the only one around.
Your UI it's ultra cool and fantastic. Very professional maiden and practical but some essential small things like this very specific one, makes bad to the overall great experience.
I can understand out of a UI-veiw this move, but I know a lot of people who see a boundary between the standard buttons on the keyboard, the short cut buttons, the buttons on the screen (i.e. the software), and all other buttons (I think it was first last year my mother learnt how the "e-mail" shortcut button worked on her laptop she had used for 4 years straight and still double click links in browsers). And for those users the power button on the computer is for starting (even seen the button referred to as the ON-button in manuals/docs aimed at that group of computer users). Nothing else.
So from a UI-pov, good move, from a user-pov very questionable.
On the other hand this operating system maybe is aimed just for people who KNOWS computer, and just not uses them to get the work done, or for people who have never got enough computer experience to have those kinds of preconceptions about how stuff works.
For example, what happens if you do two test runs which happen to produce almost the same average FPS? The PTS just blindly accepts that result as long as it is below the standard deviation threshold, despite the fact that the confidence in that result is totally unknown.
ANOVA would also allow calculation of confidence intervals which would give meaning to your results. For example, in the Blowfish benchmark, performance varies by only ~1.8% between the top performer MeeGo, and worst performer Moblin. I strongly suspect that this is not a statistically significant result however you "call" it for MeeGo.
My personal experience is that PTS runs most tests at least three to five times and THEN looks at std dev. Also, from using some of the PTS tests in our regular QA runs, I can say that the tests we use at least are very repeatable, with one big caveat.
This caveat is that if your OS is having a lot of "background noise", the results will be less stable than when you have a "clean background OS".
In MeeGo we've mostly succeeded in getting a clean background OS (this is more important for power behavior, but the stable performance is a nice side benefit), and the results are very repeatable. I can imagine that OSes that haven't paid attention to this and/or are very noisy, will see more variability in results...
It's an interesting and well done review.
"... Also in the testing mix are Ubuntu Netbook Remix 10.04 LTS, Moblin 2.1, and Fedora 13."
Regarding Linux distributions on netbooks, you may have not noticed that there is a new "kid in the city".
I'm referring to Peppermint OS One, witch had it's first official release at 27/05/2010, as per Distrowatch announcement.
This is a lightweight LXDE distro,featuring the Openbox WM, based on Debian, Ubuntu and Mint, with many web-based applications, targeting the Netbox users.
Next opportunity, it would be interesting to see how Peppermint OS compare to the Distros included in the present review.