“Back in 2003, Microsoft assembled a team of engineers to rethink the lowest levels of Windows, so that the OS could be more easily slimmed down and secured to run in servers and embedded applications. That project, called ‘MinWin’, has now started to bear fruit.”
Its too bad that it was not a MS employee who wrote this, but nonetheless, isn’t this a blatant confession that Windows sucks for all these years?
And when in eternity are we going to have UTC real time clocks?
As said before, if MinWin is supposed to only have stuff that modern OSes need, why is it so freaking huge? Surely if the rest of the computing landscape does not need so much code just to handle memory, MinWin should not need it too.
Apparently, MinWin is “minimalist” Windows in the sense that a aircraft carrier is a “minimalist” spaceship. Not a very good thought for us truck drivers in Linux land (openBSD would be a tank, cars would represent possibly QNX and those abandoned tryouts or things like freeDOS would be bicycles / motocycles)
What makes you think it’s a significant challenge to thave the RTC in UTC time? It’s possible to configure your system that way already under Windows, but it seems hardly a major technical flaw in MinWin, full Windows, or anything else.
Is it really asking too much for people to have a basic understanding of OS kernel design before posting such random garbage? Seriously…
Edited 2009-11-19 10:06 UTC
As a reply to both tomcat and PlatformAgnostic above,
RTC in UTC is not a resolved matter in Windows. In Vista and newer, setting the registry key to set RTC to UTC will only cause Windows to not update RTC. It used to be the case, pre-win2k, that WinNT could run on ARM. In those days, setting RTC to UTC actually works and the registry key was obeyed (and Windows Time would update RTC). Post-win2k, however, the monstrosity that is Windows actually tramples on the registry setting no matter what you did. And in Vista, all they could do was not update RTC.
I mean, what rubbish is this? Granted, in server situations you do not regularly dual-boot, but if you had a situation where you have 2 server images that are designed to be run alternatively (one image running and serving pages, the other being checked for integrity and maintained) on the same machine, having DST problems is a huge pain. However, if RTC is not updated, then 1) scripts that require timestamps from clients may receive “future timestamps” error (UTC moves forward 1 second every so often) or 2) in the case of a bank, transaction timestamps can go haywire if some clients update and some don’t.
Now, tomcat, can you elaborate on what was “a basic understanding of OS kernel design” I am missing here? Admittedly I am bringing up random stuff, but I didn’t even say that this is a kernel design fault. This was and is an implementation fault (registry keys not observed), not a design fault. All I could imply was that the kernel design was so huge as to be capable of hiding bugs (in this case a lousy assumption) like this, and it in itself is bad (although not catastrophic).
Secondly, it is true that MinWin is pretty damn huge for what it essentially does (no graphics subsystem, no unrequired rubbish, just plain basic hardware code and memory, IPC, security and little else. Heck, due to it being “hybrid modular-monolithic”, there are some other stuff that are not in there too. So why is it about a order of magnitude (in 10s, not 2s) bigger than, say WinCE6 kernel? (CE6 kernel ~350kiB) Is this acceptable given that it is supposed to do so little?
Notice that I am not saying that MinWin is bad. I shall refine it — I mean to have said that MinWin represents a decidedly good slimming down of Windows. But, it does seem to have a lot more to go. I would have said that Mac OS X 10.1 was a good idea, but it _was_ slow, bloated and so on such that every iteration of OS X could speed it up.
So tomcat, please elaborate on why you think my comment lacked an understanding of basic kernel design.
It’s tragicomic how Microsoft goes or tends to go slim and f\oss world becomes more and more bloated.
I welcome any improvements of Windows. Because it boosts competition and also because I ocasionally am required to interface with them.
And, yes, I’m writing this from my linux powered netbook.
~% uname -a
Linux archlinux 2.6.31-ARCH #1 SMP PREEMPT Fri Oct 23 11:12:58 CEST 2009 i686 Intel(R) Atom(TM) CPU N270 @ 1.60GHz GenuineIntel GNU/Linux
Thanks for clearing that up. I would never have believed you actually ran linux on OCT 23
I am running from the top secret Google Android OS.
~% uname -a
Google Android 4.1.32.Beta-ARCH #2 SMP PREEMPT Fri NOV 19 13:12:52 EST 2009 i686 Transmetta Nexus Processor @ 10.2 GHZ Genuine Transmetta Google
Actually that is the date the kernel was built, not current system time upon uname(1) invocation
Cheers
Touch~A(c)