Last edited by 0xBADCODE; 10-29-2012 at 10:08 AM.
There are a few points here:
1) Upgrades versus clean install:
Clean Installs are always very messy. If you want to have a good, stable system don't mess around with upgrades. Seriously it's biased.
2) Microsoft has very much resources but with every release the hardware vendors don't get decent drivers with Windows out the door. Same has happened with Vista. Things changed, hardware vendors didn't update drivers fast enough. Things didn't work out of the box. Even an organization like Microsoft is quite powerless against this phenomenon.
3) Although the article mentions a few cases this is still a very limited and anecdotal data. This may very well be very not what typically happens.
Rolling releases don't count as upgrades? Arch upgrades fine... usually =).
Originally Posted by garegin
The only complaint I have heard is from not technical users finding Win8 hard to navigate. It maybe hard to remember but not everyone new what a a:/ or c:/ drive was way back before they were trained.
Now it takes a little adjusting to Mint or Ubuntu from a Windows user as well but as long as it's a gnome 2xx or Mate (Cinnamon as well) environment they take to it quick enough. I remember first trying gnome shell and laughing at how simple it was to use. I still fscking hated it but I understood it. Win 8 Metro isn't too bad just clunky and obtrusive. I hate the homescreen sooo much and my computer is not a phone! my phone running JB is =).
my own criteria
I judge an operating system by counting the number of mouse clicks and key presses it takes to swap the control and caps-lock keys.
I don't know about you, but I actually USE the control key and I refuse to suffer RS injuries due to the lazyness of developers. "Your software literally causes me pain"
So far RHEL and OSX are the winners. OSX is the REAL winner here, it understands keyboards better than any other OS. Fedora used to be up there and now it's falling behind. Ubuntu is a LOSER. Windows doesn't even show for this competition, if you count the googling that must be done to figure out how to do it.
Huh? There isn't even a comparable "repository concept" existing on MS' OS. Instead you get all sorts of "auto updaters" or have to take care of all the updates manually. And the "(un)install software" dialog in the system settings is an abomination at best (at least on Win 7 and earlier).
Originally Posted by ruinairas
"Finnally! A sensible review of Windows 8". Suprise, suprise: This article turned out to be about Ubuntu, not Windows 8.
I support your propaganda, but I would appreceate some kind of warning. I mean, common! This was not even remotely interesting. I dont need windows, so Im not going to test it myself. But it would be usefull to read a serious review from someone that is familiar with whats going on in the open source world.
Most reviews I have seen are made by people that seem to only have used the previous windows releases, and maybe just seen a mac. Its like those horses with patches on the side of their eyes. Patches so that they can only look straight forward and dont get distracted. You dont want that horse to give you perspective on stuff.
You want the whimsy crazy horse to do that.
I think you mean upgrade installs are always very messy.
Originally Posted by plonoma
JFYI, WinNT also uses ACLs. It could be debatable which implementation is best but at the end of day you usually could set up required permissions in any system, one way or another. In NT it's more flexible but also...
Originally Posted by plonoma
1) It's not like if you can efficiently and easily edit NT ACLs from batch files, etc. No, seriously, *nix style rights far more simple for automation and batch operations. Windows never targeted that. So it comes at cost. OTOH there is no problem in *nix-like to do anything in batch manner easily enough from scripts or commandline.
2) NT ACLs are overcomplicated for most system administrators. In fact it's quite rare case when administrator could fully understand NT ACLs operation and properly evaluate effective rights taking all things NT ACLs could use into account. This often leads to either dozens of issues with non-operable programs or ton of security holes. OTOH basic *nix rights system is simple and yet covers most usage patterns. And you can understand rights in effect. Most notably, MS and their devs still miserably fail to run many programs under user (non-administrative) accounts. They even gone as far as to reinvent sudo in most awkward way I ever seen - UAC. In this weird implementation not just administrator account is borked and can get "access denied" but also programs should be explicitly aware of this misfeature. Or else they will be unable to enter some directories at all. Really amazing crapwreck. They "fixed" what was not broken in incredibly twisted ways just to avoid dealing with root cause.
3) To make matters more interesting, recent windows redirects some directories for some apps. So it's not uncommon to have different filesystem view in different programs. Even if it's not what you want, it could happen. Amazing wreckage again. No, really, they got that it's "bad idea to run all programs as root" but failed to implement it without moron system hack-wrecks. While NT ACLs allowed it from the very begin, MS just failed to enforce this policy on devs. It's just win95 legacy that prevented devs from taking NT rights as something mandatory. It could be sensed even today. You see, *nix programs do not assume user is root for day-to-day OS usage. Not a case with windows apps. Half of them just fail to run as unprivileged user. So MS has been forced to invent some weird semi-limited administrator. Really moron granted that ACLs allowed to do this in usual ways. Just enforce policy on devs and rock-n-roll.
4) In linux it's up to me to decide how far I can go. I can use just basic rights. Or more advanced ACLs. Or even SELinux/APPArmor/... advanced security subsystems, should I ever need that power. Or I can in-kernel KVM or containers to split the worlds. And every approach is simple on it's own. So I can easily evaluate the results.
5) NT ACLs were well-designed. However years of marketing crap and attempts to fix what's not broken in futile attempt to solve all problems via strange hacks for compatibility reasons led to something strange and seriously wrecked. To make matters even worse, Windows filesystem also not case sensitive and could access same file in more than one way. Say in 8.3 "dos paths" notation. As the result, real-world deployments, applications and somesuch often suffer from either horrible bugs due to all these weird "hacks" from MS or system security goes well below desired levels. For example it's very common that program serving remote requests (say HTTP or FTP server) and forcing some kind of their own ACLs would EPIC FAIL to deal properly with the fact that same file can be accessed via different paths/names (i.e. long names vs dos 8.3 names). This requires specific attention and only happens in Windows. Most notably, recently well-known nginx server has been found to fail to evaluate permissions when running under windows due to such tricks.
So does NT ACLs lead to better security? Nah, they aren't. And everything else is secondary.
Last edited by 0xBADCODE; 10-29-2012 at 12:23 PM.
They reinvent repositories;)
You see, these morons are trying to reinvent software repositories. But still years behind from having anyhow comparable tools for system administrators, etc. And they got it in usual way. Aka "shop" and "monopoly". Which makes this idea far less useful for anyone but MS themselves. Not to mention it completely ruins traditional business models around PCs.
Originally Posted by Tuxee