They didn't "gratuitously change" their driver model, they moved a lot of it into user space for security and stability. Video driver crash on vista/7? No worries, screen goes black, then screen comes back a few seconds later. Sound driver crash? Same thing.
You want to run 5 year old hardware? Run the OS it is certified for. Complain to your hardware vendor if there is no driver. That's the linux excuse, isn't it?
Any operating system administered by dickweeds will get owned sooner or later. I've fixed Windows boxes, I've fixed Macs, and I've also fixed r00ted linux boxes.
I run a Windows / FreeBSD network without issue, because I do my job properly.
The benchmarks he chose were irrelevant for desktop interactivity. That's the point. Con went through many months of similar b.s. last time around, said "don't contact me in future" and instead gets email from Ingo comparing his scheduler on a set of irrelevant data. Hence con's response of "do you know what desktop hardware is, and what people do with it" (paraphrased).
Out of context, assuming it was a big "fuck you" would be paranoid, yes. But after several months of well publicized animosity between key kernel developers and Con last time around, I'm quite sure he's well past caring.
Either way, several people have responded on list stating improvements on the tasks it was designed to improve, and have perhaps highlighted some of the problems with the current scheduler. Anyone who has used Linux back to back with BSD, OS X or even Windows can see that there are glaring problems with Linux for desktop use right now, and I'd certainly agree with the assertion that things were better for interactive use back in the 2.2 and even 2.0 days.
There's only so much passive-aggressive shit one can take. If you read why Con quit in the first place, Ingo/rest of list's behavior this time around is pretty much identical. Despite Con making it quite clear previously that he is NOT TO BE CONTACTED by those same people regarding kernel stuff.
Its EXACTLY the same old shit con quit over before. LKML dudes say "throughput sucks on my high end box on abstract test X". No shit. Try running the tasks it was designed to optimize (i.e., real world desktop applications) and then verify yay/nay for whether it does as claimed. Beating the CFS on abstract benchmarks on high end hardware is explicitly NOT what it is designed for.
Ingo's post was written to appear civil, but it was a big "fuck you con" if you read between the lines.
Even if you have total control over all physical access points to your LAN, and total trust in your user base, there is still a chance that internal people can try to do nasty things - and in some ways they may have more motivation to do so.
And even if they don't have motivation, there is still a decent chance they'll pick something up via usb stick, wifi or whatever when out of the office, and then bring it back in behind the perimeter firewall. Relying on perimeter firewalls pretty foolhardy these days. I agree with your assertion that the internal trusted network is going to shrink - it doesn't exist for us already. Perimeter firewalls are all well and good if you block unfiltered email, glue usb ports shut, take out optical drives and remove all the development tools from your machines, disable the wifi, and nail them to the desks. That isn't practical though.
Both Windows XP and Windows 2000 are operating systems people should be migrating off. They're both 7 and 9 years old, respectively. If you're still running either one (and haven't at least hardened it appropriately), sorry you deserve whatever ownage you get.
Um. If you run a Windows network (and I do), you DO EXPECT attacks from inside your LAN. To not expect them and not bother with vulnerabilities like this is the difference between having one owned machine due to Fred downloading a trojan/worm, and an entire botnet.
The days of the perimeter firewall are over. WANs/LANs are too big, software is too complex. You have to treat the LAN like the internet, if its a decent size. Least privilege applies.
You may say "oh, i run linux, i don't care". Yes, you too. Linux, OS X. whatever- they all have remote exploits from time to time, and to bury your head in the sane going "lalalala i'm immune" is to invite disaster.
Scheduling is actually a major difference i noticed between FreeBSD and Linux on the desktop actually. FreeBSD seems FAR more tolerant of heavy load not fucking things up and causing jerkiness, slow UI response, etc. Haven't used either OS on the desktop for some time now (a year or so), but "responsiveness under load" was a major reason I starting getting more interested in FreeBSD in the first place, way back in 1999...
For desktop use, I doubt many users *care* whether or not they drop some percentage of throughput on interactive apps, if it means that processes actually run "properly" (eg, video playback, gaming, audio processing, etc).
ingo benchmarking some abstract processes that no desktop user would actually run day to day merely reinforces con's point.
Yes, con may have come off as a bit of an arse, but given his previous "do not contact me regarding kernel matters" posting to LKML, only to be e-mailed with benchmarks on non-desktop hardware, performing non-desktop tasks that shows CFS to be "superior", I'm not surprised.
I'm guessing he's pretty much "over" banging his head against the wall, trying to get people to "see the light" (or understand that the point is improving interactivity, rather than benchmark numbers).
This was my point. If the loss of one person kills your business, your business design/contingency planning is completely fucked and you deserve to fail.
This. Also, from a corporate perspective - the SOE / imaging process in Windows 7 is way better than XP. App-v, applocker and bitlocker are all wins for the corporate desktop as well. Not to mention branchcache, better search, previous versions, etc.
For a business, there are plenty of benefits. And home users will eventually end up running what they run at work. Its how the PC came to dominance in the first place... the other flavors of machine (amiga, etc) died because they were IBM PC compatible, and IBM PC was what was in the workplace.
Are they including configure time in that? If the configure script detects "OS X" it probably has to do a shitload less tests for available libraries...
Same here. The limiting app I am waiting for at the moment is a 64 bit Riverbed Steelhead mobile client. I want to roll 64 bit at the same time to say 90% of our users, and thats the only commonly used app amongst our mobile power users that I'm waiting on...:)
The deployment tools are light years ahead of what was available for XP.
Seconding this. I have documents on the network that I turn into electronic templates for our faxing package that were done in office 97, and they still format fine in 2007.
As a user of both, i have to say that I am happy with Windows 7 performance. its not "more of the same junk". Installed Snow leopard last night, and the improvement is nice as well.
Missiles don't fire by lining up crosshairs, but if you are facing the wrong direction, the missile must burn huge amounts of its fuel to get facing the right direction.
Being able to have it launch in the right direction in the first place, or close to it (by rotating the nose) can mean the difference between a missile that has enough energy left upon reaching the target to deal with any evasive measures the target takes.
Missiles can turn, yes - but the less you make them turn, they're far more likely to hit, due to being able to deal with the target moving.
But, more to the point, designing a fighter that can't deal with close in combat is a stupid move. If dogfighting is so not necessary any more, why does the USAF spend so much on red flag exercises, and simulated guns combat? If dogfighting ability is not useful or necessary, then why build fighters at all?
30 bucks for an OS upgrade is more than fair, and most people will spend more than that on an evening at the cinema. Its not an expense most people not living in their parent's basement would even think about.
200 bucks on the other hand is a little more out of the weekly paycheck...
You're still stuck running Windows or an open-source OS. Some people are willing to pay the difference (which over 3 years works out to be about a dollar per day, and is a tax deduction for many people anyway) to not have to deal with that shit.
Slashdot Mirror
So you're a mac user, who's experienced both Leopard and Snow Leopard and are thus qualified to comment, yes?
You want to run 5 year old hardware? Run the OS it is certified for. Complain to your hardware vendor if there is no driver. That's the linux excuse, isn't it?
I run a Windows / FreeBSD network without issue, because I do my job properly.
Out of context, assuming it was a big "fuck you" would be paranoid, yes. But after several months of well publicized animosity between key kernel developers and Con last time around, I'm quite sure he's well past caring.
Either way, several people have responded on list stating improvements on the tasks it was designed to improve, and have perhaps highlighted some of the problems with the current scheduler. Anyone who has used Linux back to back with BSD, OS X or even Windows can see that there are glaring problems with Linux for desktop use right now, and I'd certainly agree with the assertion that things were better for interactive use back in the 2.2 and even 2.0 days.
Its EXACTLY the same old shit con quit over before. LKML dudes say "throughput sucks on my high end box on abstract test X". No shit. Try running the tasks it was designed to optimize (i.e., real world desktop applications) and then verify yay/nay for whether it does as claimed. Beating the CFS on abstract benchmarks on high end hardware is explicitly NOT what it is designed for.
Ingo's post was written to appear civil, but it was a big "fuck you con" if you read between the lines.
And even if they don't have motivation, there is still a decent chance they'll pick something up via usb stick, wifi or whatever when out of the office, and then bring it back in behind the perimeter firewall. Relying on perimeter firewalls pretty foolhardy these days. I agree with your assertion that the internal trusted network is going to shrink - it doesn't exist for us already. Perimeter firewalls are all well and good if you block unfiltered email, glue usb ports shut, take out optical drives and remove all the development tools from your machines, disable the wifi, and nail them to the desks. That isn't practical though.
Both Windows XP and Windows 2000 are operating systems people should be migrating off. They're both 7 and 9 years old, respectively. If you're still running either one (and haven't at least hardened it appropriately), sorry you deserve whatever ownage you get.
The days of the perimeter firewall are over. WANs/LANs are too big, software is too complex. You have to treat the LAN like the internet, if its a decent size. Least privilege applies.
You may say "oh, i run linux, i don't care". Yes, you too. Linux, OS X. whatever- they all have remote exploits from time to time, and to bury your head in the sane going "lalalala i'm immune" is to invite disaster.
SBM2 is designed to work better over WANs. hence, different physical network, vpn, whatever - if you pass SBMv2, you're owned.
Thats why linux is beating windows NT/2k/XP everywhere (which is a microkernel).
oh yay, the slash id pissing contest. do i win a prize? by the way, you post like an asshole.
Scheduling is actually a major difference i noticed between FreeBSD and Linux on the desktop actually. FreeBSD seems FAR more tolerant of heavy load not fucking things up and causing jerkiness, slow UI response, etc. Haven't used either OS on the desktop for some time now (a year or so), but "responsiveness under load" was a major reason I starting getting more interested in FreeBSD in the first place, way back in 1999...
For desktop use, I doubt many users *care* whether or not they drop some percentage of throughput on interactive apps, if it means that processes actually run "properly" (eg, video playback, gaming, audio processing, etc).
ingo benchmarking some abstract processes that no desktop user would actually run day to day merely reinforces con's point.
Yes, con may have come off as a bit of an arse, but given his previous "do not contact me regarding kernel matters" posting to LKML, only to be e-mailed with benchmarks on non-desktop hardware, performing non-desktop tasks that shows CFS to be "superior", I'm not surprised.
I'm guessing he's pretty much "over" banging his head against the wall, trying to get people to "see the light" (or understand that the point is improving interactivity, rather than benchmark numbers).
Why? Why is that a better solution than the plague that is humanity disappearing?
This was my point. If the loss of one person kills your business, your business design/contingency planning is completely fucked and you deserve to fail.
If your business can be taken out of action by an administrator leaving, then you have serious business process problems.
For a business, there are plenty of benefits. And home users will eventually end up running what they run at work. Its how the PC came to dominance in the first place... the other flavors of machine (amiga, etc) died because they were IBM PC compatible, and IBM PC was what was in the workplace.
Are they including configure time in that? If the configure script detects "OS X" it probably has to do a shitload less tests for available libraries...
The deployment tools are light years ahead of what was available for XP.
Seconding this. I have documents on the network that I turn into electronic templates for our faxing package that were done in office 97, and they still format fine in 2007.
As a user of both, i have to say that I am happy with Windows 7 performance. its not "more of the same junk". Installed Snow leopard last night, and the improvement is nice as well.
Being able to have it launch in the right direction in the first place, or close to it (by rotating the nose) can mean the difference between a missile that has enough energy left upon reaching the target to deal with any evasive measures the target takes.
Missiles can turn, yes - but the less you make them turn, they're far more likely to hit, due to being able to deal with the target moving.
But, more to the point, designing a fighter that can't deal with close in combat is a stupid move. If dogfighting is so not necessary any more, why does the USAF spend so much on red flag exercises, and simulated guns combat? If dogfighting ability is not useful or necessary, then why build fighters at all?
200 bucks on the other hand is a little more out of the weekly paycheck...
You're still stuck running Windows or an open-source OS. Some people are willing to pay the difference (which over 3 years works out to be about a dollar per day, and is a tax deduction for many people anyway) to not have to deal with that shit.