Slashdot Mirror
Microsoft Warms Up to Linux
prostoalex writes "InfoWorld reports that despite warming to the OS, Microsoft won't be releasing its own distribution of Linux any time soon. From the article: "Hilf acknowledged that Microsoft's commitment to Windows does not preclude the company from continuing a strategy he has led in his 19 months at the software vendor: To see how Microsoft's proprietary technologies can better interoperate with Linux and a host of other open-source software. In fact, that is exactly what will be the focus of a discussion the long-time open-source proponent will lead at this year's upcoming Linuxworld Conference & Expo next month in San Francisco. In a session entitled, 'Managing Linux in a Mixed Environment ... at Microsoft?' Hilf, who polished his open-source evangelism skills working on Linux deployments at IBM Corp., will talk about how he and the team at the Linux/Open Source lab run open source technologies in "the most Microsoft-centric IT environment on the planet." "
Yes, I occasionally "work the video card hard" doing some of my own OpenGL work, plus a little Enemy Territory now and then.
Since you claimed "every desktop" and "every video card", your argument is thus refuted.
Ryan T. Sammartino
"Ancora imparo"
Usually it's impossible to update just the video card drivers in Windows, period. So many of these vendors now, bundle their drivers in a installation exe, which without giving me a choice install all kinds of auto-update crap , and utility *cough*spyware*cough* crap. Do I really need a s/w which phones home every 1/2 hour to check for a update for a display dirver.
for the last time people, I am "frodo from middle eaRTH", not "middle eaST".
The fun part is that I asked a Microsoft rep about the Kerberos problem and he lied to my face.
Are you sure he wasn't just plain ignorant (representatives tend to be)?
The problem with folks like you is that you have your own experience, and then extrapolate it to others'.
If every desktop crashed on every video card if you tried to do anything that works it hard makes me wonder if my days of playing Quake3 under KDE on FreeBSD AND Linux were just some magical hallucination. (GeForce2 at the time.)
Ya got modded as a troll, but I think you're more likely simply a little misguided and/or hurt that you experienced difficulties?
I mean, I've had games that couldn't run under Windows with a good graphics card, but that doens't cause me to make the laughable claim that Windows just crashes under intensive graphical goodness.
They're computers running software; they're failable, and sometimes you just get stuck with a lemon. Not that things are easy to setup in Linux, but to claim you cant create a stable desktop and gaming environment is flat out wrong.
"Old man yells at systemd"
Interestingly, doing what MS did in the way they did introduces a weakness in Kerberos.
The MSKDC populates the authorization-data in the ticket-granting ticket (TGT). This is copied into the TGS-REQ when a service ticket is requested, and then is copied from the request into the service ticket. Services make authorization decisions based on the group data in the service ticket.
According to Microsoft, this is an optimization issue. Enumerating group membership is relatively expensive, especially with nested groups, so MS chose to do it only once per login session, i.e. when the TGT is requested.
But what this means is if a user's group membership is changed while during the lifetime of a TGT (10 hours by default), the changes don't take effect until the user gets a new TGT.
Now, in an MS-only environment, you can mitigate this by using forced logoff. Basically, the administrator tells the workstation to discard the user's TGT, and the user is forced to get a new one, with new his new group enumeration.
But you can't do this to any other Kerberos implementation--like MIT Kerberos on Linux or Mac OS X. So if a mole logs in to his Linux box and gets a TGT from your domain at 0800 and starts using his privileges to wreak havoc, there's nothing you can do (other than physically disconnect him) until his TGT lifetime runs out 10 hours later.
Sucks to be you that day, doesn't it?
Admittedly this isn't a very likely scenario, but it does illustrate the point that mucking with security protocols at random like this can have non-intuitive effects.
-- Cerebus
Compare Microsoft's extensions with GCC's. Quiz: which of Microsoft and GNU do you think are introducing more portability problems by embracing and extending the C and C++ languages?
Who uses compiler specific language extensions when they're trying to write portable code? Nobody. So long as it compiles standards compliant code, it doesn't really matter how many extensions are available. Nobody's forcing you to use them. Most of the extensions for both compilers you mention are useful and valid when you are trying to optimize code for a particular platform.
And if this was the approach Microsoft took to "embracing and extending" that would be fine. But it's not. Traditionally they have pretended to adopt a standard, added extensions without telling anyone in the standards committee, actively promoted their use without indicating that they are non-standard (like calling it "managed C++") and that in most cases things could actually be accomplished in a standards compliant manner. Since they have a dominant position in the market place, use of these proprietary extensions becomes commonplace and displaces alternative products who can't obtain enough information about microsoft's new "features" to support them.
Completely and utterly different from C/C++ language extensions.
:wq