Slashdot Mirror
Torvalds on Where Linux is Headed in 2008
Stony Stevenson writes "In an interview at the ITNews site, Linus Torvalds lays out his current excitement about the future of Linux. Torvalds is looking forward to hardware elements like solid-state drives, expects progress in graphics and wireless networking, and says the operating system is strong in virtualisation despite his personal lack of interest in the area. 'When you buy an OS from Microsoft, not only you can't fix it, but it has had years of being skewed by one single entity's sense of the market. It doesn't matter how competent Microsoft — or any individual company — is, it's going to reflect that fact. In contrast, look at where Linux is used. Everything from cellphones and other small embedded computers that people wouldn't even think of as computers, to the bulk of the biggest machines on the supercomputer Top-500 list. That is flexibility.'"
So 2008 is finally the year of Linux on the desktop?
Not really much to the interview.. It can be summed up with 1 Q&A
Interviewer: Where is Linux going.
Linus: Its going where it wants to.
If games made for Windows worked 1% faster in Linux, we'd have a generation of kids who would only know windows as the OS used in businesses.
The day I see in a game forum "Use Linux, n00b." as the usual reply to "OMG! Low fps! Getting pwned! HALP!" will set the ten year count to Linux victory over Windows.
Can someone summarize Linus' earlier claims on Linux? He must have been asked where he saw Linux in 2005, 2006 and 2007. While there must be some "right on" predictions, I am sure there are some predictions that could be seen as way off course. I slashdotter is eager to know.
Linux wireless support is often better than windows (packet injection, rfmon sniffing etc)... You just need to shop around and buy decent cards if you want the best performance.
All the cards I use are Atheros based, and work perfectly with Linux... I used to use Prism2 (802.11b only) based cards which also worked well.
I've also found Intel's cards work very well.
If you run some rare type of wireless card you may find that the windows drivers aren't too great for it either, and might stop receiving any updates rather quickly. You're also more likely to have other issues, like drivers breaking when you update windows (how many older types of card don't work at all with vista? and how many of these are no longer supported by their manufacturers and so will never work?).
And don't get me started on manufacturers who sell the same model of card with different chipsets, that's wholly irresponsible. They should change the model number if they change the core chipset, as it effectively becomes a whole different card.
http://spamdecoy.net - free throwaway anonymous email - avoid spam!
This is not about politics, and this story has absolutely nothing to do with licensing, so let's not drag that dead horse up again. Sure, it's a valid debate, but there's a place and time for it, and this isn't it.
Basilisk Digital
2008 is seeing the birth of laptop computers below $300: XO, Asus EEE, and I guess some new will appear soon.
Vista alone is almost more expensive than the hardware !
Microsoft was a good alternative when computers did cost $1500, but now the price is just too heavy.
But they really can't win when the hardware is cheap.
If they keep remaining in the high performance market (which seems their belief, see DirectX 10), they'll lose their market share in 2 years, along with Dell !
There is a new 80211 stack in Linux with better structure that allows easier creation of device drivers. This makes it easier for manufactures to create drivers, like the one who designed your card. For those manufacturers that do not bother, like the one who made your card, it also makes it a tiny bit easier for enthusiasts to step in.
I hope that makes it clear for you.
:. Ultimate Control Dedicated/VM Servers
I misread "One of the things I personally am really interested in is the move over to SSD" as "to BSD " and nearly lost my coffee all over my laptop....
Oh arse
Well, maybe once you get old enough you realize that the test of any theory is practice. And maybe Linus is old enough to realize that the test of how useful Linux happens to be is how it is used.
Any guest worker system is indistinguishable from indentured servitude.
...it comes as a business platform, not an operating system. The difference is: the OS has to do its job flawlessly in the best possible way in order to minimize the amount of work (read: time, money) required, while the business platform is something that resembles an OS but also comes with a load of business services built around it in order to generate a flow of money.
The problem with the business platform is that it was built for the sole purpose of selling services, therefore when it eventually works and there's less demand for services (data recovery, repairs, etc.) it must be tagged as obsolete and replaced by something newer and shinier but still defective in order to generate again a strong demand for services.
This is the exact reason why Microsoft stopped developing XP the moment it started being a decent OS, pushing instead the adoption of that Vista crap, and also explains why anybody who cares for his/her data or systems should consider Linux, BSD and other operating systems built to work with no strings attached.
but I didn't see any momentum at any place
/. readers are pro-standards, the lack of a 'hard' standard, or small set of standard configurations is a hindrance to more widespread *desktop* adoption.
I take it you don't shop at Wal-Mart?
I didn't see anyone in my office switched to Linux.. or any of my clients.
And you probably won't, as most office PCs fall under the jurisdiction of IT overlords who dislike users replacing OSes.
Sure.. they have nothing else to do other than wrestling with Linux.
I'll take that as sarcasm, and agree with you. The biggest stumbling block to widespread Linux adoption on the desktop is that it usually does take some 'wrestling' to get it to work, whereas Windows generally 'just works'. Yet that's not a fault of Linux, it's a fault of hardware makers who decide to release a driver for Windows and NOT for Linux.
I was going to mention the lack of GUI tools for some tasks, requiring users to manualy edit init files, but then I remembered how many times I've had to open regedit and manually change registry entries. In that sense I've had to wrestle with Windows as much as Linux.
See.. how many distros ??
Actually, a good point. There are a significant fraction of Windows users who don't know which version they're running, and in order to support them you need to know that. Same with the various distros, as they all are different enough so that you need to know which you're dealing with. I was recently at an acquaintance's house and saw their computer. "Hey, you run Linux" I said.... "No, it's Ubuntu" they said. They could have just as easily said "No, it's KDE". Sadly, as much as most
how many kernal updates every week ???
Less than the number of Patch Tuesdays in a month, apparently.
Linux sure got some momentum on academia. Well... to be frank.. its not because they really like. Only because they want to escapre from paying volume-licenses.
Actually, it *is* because 'they like'. $300 is nothing when you've got research grants in the million$. Academia likes it because they can whittle away and tweak Linux until it does *only* what they need it to do, and do it efficiently and fast. Faster than Windows. And when you only need half the computers to get the same speed, or can get twice the speed with what you've got, you use Linux.
But if you really want to argue cost, then don't forget the electricity bill. The $300 spent on a license costs more when you need to buy and power more computers to get the same results in the same time.
Furthremore, there are linux idiots who worship linux OS, who monopolize linux-OS in their domain.
There are Apple fanboys too. And yes, sometimes Windows actually *is* a better choice, although thankfully those special cases are becoming fewer and fewer as time goes on.
Linux community should give up their efforts and must try to learn some lessons from M$ and either help Windows to be better OR do something like Windows for FREE.
I think they *did* learn some lessons... lessons in what NOT to do. In fact, looking at Vista, I think MS has a few lessons that *they* need to learn from the Linux community.
As for 'doing something like Windows....for free', isn't that *exactly* what Linux is?
Afterall.. true power of linux can not be executed without being a linux-geek.. who knows all the command line commands and some degree of linux kernal modding... that's pathetic.
And the true power of Windows can not be executed... FULL STOP. Can't streamline the kernel, must know all the registry tweaks which may or may not be published anywhere. THAT is pathetic.
I am running VMWare 5.5. I tried to get it running on the latest Ubuntu distro, and the one before that. What happens is that it asks if I have a compiler handy in install.pl file. Then when it attempts to compile and one of the headers buggers up.
The problem is as follows:
http://www.debuntu.org/how-to-vmware-server-workstation-under-ubuntu-feisty
I tried using the prepared binary patches with Ubuntu, but they did not seem to work for me. The only thing that worked was to go back to an old Ubuntu version and then be done with it. AND not upgrade the Linux kernel.
I am tired of this. I am tired of needing a compiler installed. Tired of doing an installation of an installation. I just want it to be installed and running.
Now talking about getting VMWare to run on some random Linux distro. Actually I can expect that. I can install VMWare workstation on Windows XP, Windows 2000, and Windows Vista, Windows 2000 Server, and Windows 2003 server without any hassles whatsoever! I can't say that of Linux.
"You can't make a race horse of a pig"
"No," said Samuel, "but you can make very fast pig"
'When you buy an OS from Microsoft, not only you can't fix it, but it has had years of being skewed by one single entity's sense of the market. It doesn't matter how competent Microsoft -- or any individual company -- is, it's going to reflect that fact. In contrast, look at where Linux is used. Everything from cellphones and other small embedded computers that people wouldn't even think of as computers, to the bulk of the biggest machines on the supercomputer Top-500 list. That is flexibility.'
The above has been in use since 1999. It needs to be retired. "We're not Microsoft," alone isn't going to cut it for much longer. If Linux advocates keep trying to use that line to the exclusion of all else, they'll eventually find that it isn't Microsoft they'll be competing with...it's Apple. That is one battle that they can't hope to win. OSX is both UNIX based, and with close-to-mainstream user friendliness. Next to that, people have no incentive to use Linux at all.
One thing I find my computer quite often busy doing is swapping. With only 512MB of RAM, and many bloated programs running, it can't hold everything in RAM all at once. But worse, I find, is when a program is doing a lot of I/O output, which gets buffered in RAM more than it should. If the data being copied is a 40GB HD video file, the assumption that I might be reading the file back in soon (so it should be cached in RAM) just doesn't cut it. An SSD dedicated just for swapping might be faster (eliminates the seeks, but still uses I/O bus bandwidth). And it won't prevent existing pages from being swapped out, requiring them to be swapped back in again (usually a lot sooner than I would be reading those large files back in, which obviously cannot be read in whole).
But is SSD the answer for this (swapping)? If it were significantly cheaper than regular RAM, I might think so. For other uses (live copies of /usr, and such) it certainly could help. What I think is the answer for my case is to go overboard on RAM. My current estimate of normal RAM usage I need for my next computer build (in progress ... 1/3 of the parts already purchased) is 2GB. But what I plan to do in this case, however, is go with 8GB of RAM ... and not enable any swap space at all. Normally, the amount of swap space I would allocate is the lesser of 1: 2x the RAM ... and 2: the amount of data that can be transferred in one direction in 30 seconds. I'm switching to SATA so the latter figure will be larger. Still, the 8GB figure well exceeds the 2GB I expect to need for a while.
Suppose with that 2GB of RAM I deploy 6GB of swap space. That gives me a total of 8GB of space for dirty pages (not counting I/O output buffers which have a destination elsewhere). But during the course of normal use, dirty pages often get forced out to swap because of things like I/O output buffering, which also in turn slows down that I/O (more so if it's in the same disk as the swap space, due to head seek times). Now compare that to 8GB of RAM with no swap space at all. The capacity for keeping dirty pages is the same. But when heavy I/O starts to get pushy, there's no where else for those dirty pages to go (to make room to needlessly overbuffer the I/O). The end result should simply be that the I/O can do nothing more than be written where it belongs as fast as it can (and it can be faster since swapping isn't using up any I/O bus bandwidth nor tying up the disk heads into other locations in the case of non-SSD).
So what else is SSD good for? Maybe for /usr if the price is right. But if SSD is just RAM, bottled up through a SATA/SCSI/IDE/etc, how is that any better than RAM? Is 16GB (high end of what /usr needs for nearly everyone) of SSD cheaper than 16GB of RAM by enough to make it worthwhile? I suspect not, unless the SSD is just using cheap RAM.
now we need to go OSS in diesel cars
You can run VMware on RHEL 3, 4 and 5 without any hassle whatsoever. If you want to use proprietary software, use a stable platform like RHEL or SLES or Ubuntu LTS. The reason Ubuntu and Fedora are able to release frequently is that they do not put much effort into binary compatibility.
What you don't seem to understand is that there is no such thing as a "Linux" desktop. There are Fedora desktops and Mandriva desktops and Debian deskstops and they are all different.
> I am tired of this. I am tired of needing a compiler installed. Tired of doing an installation
> of an installation. I just want it to be installed and running.
What's to be tired of? It's Ubuntu/Debian. There's a meta package for this. Just install the meta package.
If vmware weren't more lame, they could do this as part of their installer.
This is strictly a packaging and engineering issue. Vmware insists on
making software that needs to engage in kernel level shenanigans and
won't bother to take the extra packaging effort that entails.
A Pirate and a Puritan look the same on a balance sheet.