Linux 3.13 Kernel To Bring Major Feature Improvements

An anonymous reader writes "There's many improvements due in the Linux 3.13 kernel that just entered development. On the matter of new hardware support, there's open-source driver support for Intel Broadwell and AMD Radeon R9 290 'Hawaii' graphics. NFTables will eventually replace IPTables; the multi-queue block layer is supposed to make disk access much faster on Linux; HDMI audio has improved; Stereo/3D HDMI support is found for Intel hardware; file-system improvements are on the way, along with support for limiting the power consumption of individual PC components."

Comments

SteamOS Will 'Really Help' Linux On the Desktop

[Anonymous Coward]

"Linus Torvalds has welcomed the arrival of Valve's Linux-based platform, SteamOS, and said it could boost Linux on desktops. The Linux creator praised Valve's 'vision' and suggested its momentum would force other manufacturers to take Linux seriously — especially if game developers start to ditch Windows. Should SteamOS gain traction among gamers and developers, that could force more hardware manufacturers to extend driver support beyond Windows. That's a sore point for Torvalds, who slammed Nvidia last year for failing to support open-source driver development for its graphics chips. Now that SteamOS is on the way, Nvidia has opened up to the Linux community, something Torvalds predicts is a sign of things to come. 'I'm not just saying it'll help us get traction with the graphics guys,' he said. 'It'll also force different distributors to realize if this is how Steam is going, they need to do the same thing because they can't afford to be different in this respect. They want people to play games on their platform too.'"

Re:SteamOS Will 'Really Help' Linux On the Desktop

[Anonymous Coward]

Steam?

Sorry, linux is a steaming turd OS on the desktop. You know what Windows and OSX provide? A sane target for programmers. Linux fragmentation quite frankly sucks.

That's why the ones who take Linux and turn it into something (like Google to Android) have quite control of the whole chain of which the kernel is only a piece.

Year of Linux on the desktop is utterly meaningless because of that. Any Linux you'll eventually get will be quite unrecognizable from the Linux distro you download and install and muck around with.

Re:SteamOS Will 'Really Help' Linux On the Desktop

[Anonymous Coward]

It will help it as far as hardware support is concerned, it will not attract more developers (except game developers).

Re:SteamOS Will 'Really Help' Linux On the Desktop

[Anonymous Coward]

Hopefully it will not be using Gnome3.....

So many improvements

[Tough Love]

So many improvements! Which proves that right now Linux must really suck. It's a good thing then, that Windows, FreeBSD, AIX, Solaris, etc etc can be counted on to suck far worse.

Re:So many improvements

[Anonymous Coward]

I don't even count Windows as an operating system. It's a toy. A worthless tool to be able to run a browser or games. But it will go away soon.

Re:So many improvements

[Runaway1956]

Mehhh. I used to update my kernel as quickly as possible when they promised major improvements. It always turns out that a "major improvement" is actually an "incremental improvement". I lost the excitement over kernel upgrades some time ago. I still upgrade from time to time, but my attention is more focused on security than any supposed "improvements". I don't want to be the odd guy who is caught with some vulnerability that was fixed eight versions ago. Two versions, maybe - but eight? Nope, no way! That would be just to embarrassing.

Re:So many improvements

[TheGratefulNet]

when I used normal video hardware, I could upgrade kernels all I wanted.

with stupid fucking nvidia (binary blobs) I am stuck at their mercy. 3.11 was a hard one for nvidia (nothing worked out of the box and needed work-arounds) and of course NV was slow as hell to do their own update.

its a laptop so I can't swap out the video card. I hate nvidia and their closed source driver. nouveau is not working for me as I need multiple displays (external dvi pairs) and so I'm stuck with the nv binary driver.

when that is 'ready', THEN I'll be able to run the matching kernel for it.

sigh...

(I had a nice shiny new 802.11ac intel card I wanted to use, but it needed 3.11 to run and my nv card would not run with 3.11).

Re:So many improvements

[Anonymous Coward]

So many improvements! Which proves that right now Linux must really suck. It's a good thing then, that Windows, FreeBSD, AIX, Solaris, etc etc can be counted on to suck far worse.

You forgot to list OS X and let's face it no /. troll is complete with out taking a shot at the Anti Christ.

Re:So many improvements

[Anonymous Coward]

That is what 99.9% of the population wants and needs. Slashdot is the 0.01%.

Re:So many improvements

[Anonymous Coward]

You're obviously new to Slashdot. Slashdot may, at first glance, appear to be the intelligent and knowledgeable 0.01%, but it is really the idiotic 99.999999999999%.

Re:So many improvements

[TCM]

Sounds awesome - and you're not even running any applications yet. Some people just don't have that amount of time to piss away, though.

Re:So many improvements

[Anonymous Coward]

The idiotic 99.999999999999% beg the government to take away their freedoms in exchange for security theater, so barring a few people who use Slashdot (*cough*cold fjord*cough*), Slashdot doesn't really qualify.

Re:So many improvements

[swillden]

Slashdot may, at first glance, appear to be the intelligent and knowledgeable 0.01%, but it is really the idiotic 99.999999999999%

So, you're saying slashdot is everyone?

(You'd need 10^14 people -- more than 100,000 times the population of the planet -- in order for the 0.000000000001% to equal one person.)

Re:So many improvements

[hcs_$reboot]

Unfortunately it won't any time soon. The reason? MS Office, and its billions of unmaintainable macros that keep most companies from being able to switch to something else. Sad, but true.

Re:So many improvements

[HussamAl-Tayeb]

I don't even count Windows as an operating system. It's a toy. A worthless tool to be able to run a browser or games. But it will go away soon.

also almost all structural, architectural, and engineering applications on the planet. all of those require windows.

Re:So many improvements

[Anonymous Coward]

You're wrong, MS Office really isn't needed anymore. There are tons of Office Suits out there now, choose one and go on with your life.

Re:So many improvements

[Anonymous Coward]

Piss away? I use to spend 10-20 minutes running through driver install "Wizards" in Windows that still needed to be tweaked to be usable. All the add-on devices I have, I simply plug it in to Linux and it just works. So who again is pissing away time?.

Example: Almost 20 minutes of installing and tweaking to install my Logitech Wingman Cordless in Windows. 3 Seconds and I'm running in Linux. It's not the 90's anymore people, wake up from all the misinformation you've been fed by Ballmer and the MS Fanboys.

That's a cold, hard fact. Any Linux user, real Linux user can tell you how easy it is; not the "I use Windows 99% of the time, but I have a Linux partition" people.

Re:So many improvements

[F.Ultra]

all of those require windows.

At the moment yes, nothing is set in stone here, it's not like it's impossible for these software's to be ported if the demand comes.

Re:So many improvements

[Anonymous Coward]

No they don't. There's more or less a complete set of engineering applications running on linux.
Example: http://www.ansys.com/staticassets/ANSYS/staticassets/support/platform-support-14.5-detailed-summary.pdf

Re:So many improvements

[gmack]

I had the same problem with the Nvidia drivers but 3.12 was such a large difference on my other systems that I spent the time and patched the Nvidia drivers myself using patches I found using a Google search "Nvidia Linux 3.1.2".

You should be able to do the same but if you get stuck, feel free to contact me for help.

Re:So many improvements

[KiloByte]

As you build your own kernels instead of using distro-provided binaries, what's the reason to not skip a release you don't like? 3.11 breaks VirtualBox, it works fine both with 3.10 and 3.12-rc (and now with 3.12.0). A regression that lasts is bad, one that has been fixed in a later version means just "please upgrade".

Re:So many improvements

[Anonymous Coward]

Not sure why is this "funny". It should be "Obvious if you bother to take math in highschool"

Re:So many improvements

[Eravnrekaree]

That is some of the most absurd thinking I have ever heard. The fact that things improve and get better does not mean that things are bad as they are now. These new features are additions to an already great product. Linux as it is is already better than Windows, so the new features will make it better still. Notice as well, that each Windows realese always comes with a long list of improvements from the last version. The whole purpose of having new versions is to improve things and add new capability. Had this not been done you would still be at Linux 0.01 with the Minix FS, no graphics, really not much of anything people expect of a modern OS. Why dont you just go back to using MS-DOS if you dont want any modern features, and leave us alone. For most that would be miserable and most people want a powerful capable OS and do not want to use MS-DOS, and like to be able to have things like graphics, web browsers, 3D games, improved, more powerful facilities that make it easier to develop new applications, servers, etc.

Re:So many improvements

[UnknownSoldier]

Note: My 2nd desktop is Linux 3.5 + GTX Titan + 16 GB @ 2560x1440 and love it for development.

You are missing one thing:

*BSD's pf focuses on getting it right instead of Linux always re-inventing some half-assed packet filtering every few years. I find *BSD to focus on stability and Linux to focus on flavor-of-the-month re-implementation of features. i.e. Leading Edge vs Bleeding Edge.

My Win7 box has been reduced to just gaming (aka Steam) but I prefer OSX or Linux for development and CUDA research.

Re:So many improvements

[Grishnakh]

There are, but they aren't compatible with all the wacky macros that companies have built on top of MS Office.

Re:So many improvements

[Anonymous Coward]

The answer is funny. The fact that the parent had no intelligence is as sad as you don't having any perception of humour. Must be a sad life.

Re:So many improvements

[higuita]

those companies also dropped excel macros for android apps, so that problem is being solved!

Re:So many improvements

[Bill, Shooter of Bul]

That's just for binary crap drivers. Most people don't need them, and just use the open source ones. 0 time spent dealing with them with any upgrade.

Re:So many improvements

[jellomizer]

Exactly, Linux seems to get the green light whenever it comes out, any attempt to point out any disadvantage especially with such systems such as windows would mark you as a troll. However you can bash windows all day, and keep getting mod points.

I Like Linux, it is a good OS. But it isn't perfect, and if they spent less time on evangelicalism and more on being rational and seeing the issues and improving them, then we could get a really good OS.

I have had and reported a lot problems with Linux in the past that I didn't have with Other OS's and my reports get trashed back saying, why would you want to do that!, or Fix it yourself, or why didn't you do your research to realize that while XYZ hardware v.2.0 works fine, but there is an issue with XYZ hardware 2.001 (That they pushed out without letting users know, in the same box) doesn't work.

Re:So many improvements

[Boawk]

Unfortunately it won't any time soon. The reason? MS Office, and its billions of unmaintainable macros that keep most companies from being able to switch to something else. Sad, but true.

I think this extends to many Microsoft technologies. I predict that some years from now programmers/maintainers of the Microsoft development ecosystem will be the COBOL programmers of the 21st century.

Re:So many improvements

[Kwyj1b0]

At the moment yes, nothing is set in stone here, it's not like it's impossible for these software's to be ported if the demand comes.

It's a chicken and egg problem - developers won't port unless the demand is there, and the demand won't rise till a lot of people move to Linux, which they won't because they don't have their software for Linux.

I think the whole debate is meaningless - maybe developers will start releasing their work in the form of virtual machines, or users might run virtual machines for their software. A linux base, with a Windows virtual machine (licensing issues with Macs OS on non-mac hardware) for Adobe Photoshop. PCs are good enough that they can take the performance hit of virtualization (except games in some cases). There might be also a few domain specific problems that can be addressed (like color settings for graphics work).

But on the whole, why would developers want to develop for more than one platform, if a single platform can (without too much effort) be used by all possible users?

Re:So many improvements

[F.Ultra]

I can only answer for myself and the reason that we develop for more than one platform is so that when/if the change happens then we will not be left behind (and fly by the competition when they have to stall all development for a few years in order to create a port). It was common sense to only support IE4 back in the day also, not so fun for those guys when people started to switch to Firefox et al.

Re:So many improvements

[Kwyj1b0]

I can only answer for myself and the reason that we develop for more than one platform is so that when/if the change happens then we will not be left behind (and fly by the competition when they have to stall all development for a few years in order to create a port). It was common sense to only support IE4 back in the day also, not so fun for those guys when people started to switch to Firefox et al.

Fair enough, and the question is then how confident is your organization in diversification (or more importantly in projecting future trends), and the cost of being wrong. There are countless stories of companies that spent a lot of money developing for multiple platforms, and then being burned when it became clear that diversification was pointless. If, for example, Google released Chrome a short time after firefox started becoming popular, would your strategy have worked.

On the whole, I wonder whether companies in the coming decade will bother to target multiple OS, as hardware/multicore improvements allow efficient virtualization. Either start with an OS-agnostic framework (if it can support your needs), or target one system. Hell, target Linux, because you can give free linux VMs to run under Windows (instead of expecting your Linux clients to pay for a Windows license). Similar to a JVM, but much better because you aren't stuck behind Oracle.

Re:So many improvements

[jedidiah]

...or you just wait awhile.

It's not Windows. You don't have to be running the very latest version just to avoid being owned and infected. The same goes for any operating system that's not from Microsoft.

Re:So many improvements

[jedidiah]

> also almost all structural, architectural, and engineering applications on the planet. all of those require windows.

Like you've actually used any of them ever? You could probably count the relevant user population here on your fingers.

So that brings into question the qualifications for anyone to even comment on this issue as well as the general relevance.

It's like Photoshop but you don't even know what names to drop.

Plus this is a far more geeky than average population of users. Yet you seem unable to even name these mythical apps. These apps will seem even more obscure to the population at large.

PCs may have seemed like they were "the only platform" last year but it's not last year anymore.

Re:So many improvements

[siride]

I keep seeing these quotes from Linux fanbois about those difficult and counter-intuitive wizards on Windows, but I've never encountered any issue of significance when installing drivers on Windows, and most stuff really does work out of the box (also generally true on modern Ubuntu and halfway modern hardware, but wasn't true even a few years ago). I mean, I'm sure there are some pieces of hardware that misbehave or create issues on any OS, but I really have a tough time believing clicking next through wizards for a few minutes or maybe downloading a special patch or version is in any way harder than grabbing Git versions of drivers and compiling them yourself, or having to remember the Magic SysRq key sequences to reboot when that last modprobe fucked up your video card.

Re:So many improvements

[F.Ultra]

The cost for us has been marginal, in fact we switched to completely develop everything on Linux and cross compile with GCC when building the WIndows version even though the majority of our customers are WIndows users.

To further your idea with the VMs, the increased usage of the "cloud" have increased our users requests for Linux versions of our software so I think that you are on the correct track!

Re:So many improvements

[leaen]

You're obviously new to Slashdot. Slashdot may, at first glance, appear to be the intelligent and knowledgeable 0.01%, but it is really the idiotic 99.999999999999%

Note as there are 7 bilion people this means that everybody is on slashdot now.

Re:So many improvements

[Blaskowicz]

I liked MS-DOS. It did not lack 3D games, in fact unless maybe you were in the US and could afford to use a modem, a home MS-DOS machine was used only for two things, playing 2D games and playing 3D games.

There *are* many improvements

[larry bagina]

Posted by timothy on Fri Nov 15, '13 10:04 PM

Go figure. I would accuse Dice of firing the editors and replacing them with ESL wage slaves but that would probably be an improvement.

Re:There *are* many improvements

[NoMaster]

Posted by timothy on Fri Nov 15, '13 10:04 PM

Go figure. I would accuse Dice of firing the editors and replacing them with ESL wage slaves but that would probably be an improvement.

It was posted 9:34AM Saturday, Indian time.

Just sayin'...

Nice

[Anonymous Coward]

I always wanted a Linux growing up, but I wasn't allowed to have a pet. Seriously though Linux is far better than most of the shit these days being masqueraded like they're pure gold to a bunch of simpletons lacking in the technology dept.

That's only going to work for so long until the next generation hits adulthood.

Re:Nice

[Runaway1956]

"That's only going to work for so long until the next generation hits adulthood."

Do you really think so? The "next generation" is in high school right now, taking "computer science" courses, that consist of teaching Microsoft Office. What the next generation actually needs, is to learn to question authority. If they can learn that lesson, then maybe they can break the corporate lock-in tradition that the current generation seems so happy with.

Time will tell, of course. If Microsoft, Apple, and the major manufacturers start losing big money, then you'll be proven right.

Re:Nice

[Mitchell314]

What the next generation actually needs, is to learn to question authority

Apparently somebody's never met a teenager before. :P

Re:Nice

[PopeRatzo]

I'm not sure what's going on in the school district you're in, but I've been in some Chicago Public Schools recently and they're teaching the kids a lot more than just "Microsoft Office".

I brought home one of the puzzles that some of the kids made on a 3D printer using some CAD program. It had nothing to do with Office. There were Linux machines scattered around the room and lots of the kids had iPhones or Android phones (not part of school, but don't think they're not learning how to work those platforms, make apps, etc). Now this happened to be a "selective enrollment" school (where you have to test in) but I don't think they're going all that far afield from what's happening in other schools.

I was pretty surprised at how well they seemed to be doing in terms of avoiding the obvious. Now, you'll still see Office in some of the career prep courses, but not anything like what I saw in schools' computer labs back in the early 2000s.

Re:Nice

[cykros]

Teenagers don't question authority, by and large. They yell, throw tantrums, stomp their feet, and make a lot of noise, and then once that angst is out of their system, they promptly tend to get to doing whatever it is that the authorities have told them they should do to "get ahead".

In any case, it's not about authority here...the real issue is that to most teenagers, or most people in general, a computer is merely an entertainment device, rather than a powerful tool that can be tailored to one's own needs. It doesn't matter how easy the latest user-friendly scripting language gets, "programming" remains something they envision as involving binary and machine code, purely there for autistic folks and aliens.

What we really need is to integrate programming of SOME kind into the general curriculum of our schoolchildren. And for Christ's sake, leave enough holes open on the local school network for kids to have fun learning to poke holes in the restrictive environment you've set them up in. The classes teach them HOW to do things, and the rebelliousness of getting around the restrictions gets them interested in doing them (and then the combination of heavy handed laws and bug bounty programs bring them back into societal correctness once they enter adulthood...hopefully).

The absolute LAST thing kids need is a user friendly interface. Save those for grandma, give the kid a raspberry pi, a book on Python, and then put them up behind a firewall that blocks most anything their friends will be wasting their time with. Not because you want to keep the kid OFF of such sites, but to make them at least learn a thing or two from time to time in their attempts to waste time in an otherwise purely wasteful manner.

Re:Nice

[Anonymous Coward]

I have. Most of them (like most adults) are idiots who ask the most trivial questions and don't question authority where it counts. What they usually do is mindlessly oppose everything, and that leads to them growing up and forgetting all about it.

Re:Nice

[Anonymous Coward]

The last thing we need is yet another generation of people who think writing SEO-web shit or yet another fart app is programming. We need to at least have a generation that is capable of maintaining infrastructure, be it the Linux kernel, core apps, or at least knowing how the fuck to access a disk on a sector basis.

At one job, I was looking for programmers who knew what they were doing about cryptography. Met lots who could do cutesy API stuff and very well animated lock icons, but someone who could actually look at example source code for coding some actual security infrastructure were impossible to find.

At least give the kids something to fight for. Put a firewall in place, and let them figure out how to VPN out at the minimum.

We need fewer point and drool programmers... we need more people who actually can figure out lower level infrastructure and not just tied to the highest level in the computing stack.

Re:Nice

[Anonymous Coward]

Nah just need to remove the absurd penalty involved if you get caught hacking the school network, at least as long as you don't use your exploits for any sort of personal gain. I had a lot of fun playing with the school network. Despite never once taking advantage to having full reign over the system I was thrown out of school for a year when I got caught. The effect it had on my life after school was enough to turn me away from hacking. In the 10 years since that was added to my record I have had a total of three jobs, I gave up even trying to find a normal job because I'm sick and tired of explaining it and getting turned away because of it.

What really added insult to injury was that the kids that picked fights on a regular bases almost always got detention for that, a couple got kicked out for a week, but none of them kicked out for a whole school year. It pisses me off so much that I'm shaking in anger just thinking about it.

Re:Nice

[Runaway1956]

My youngest son graduated high school three years ago, in southwest Arkansas. I taught the boy more about computers than the school system did - then the little smart ass learned at least five times as much as I taught him, on his own. He is putting himself through college, and informs me that the majority of the courses, majority of the teachers, and majority of the students are clueless boobs who can't do much more than the Microsoft-centric high schools taught them.

I'm sure that mileage does vary, depending on location, but Microsoft still has a lot of lock-in going for them in the school systems.

Re:Nice

[Grishnakh]

Teenagers don't question authority, by and large. They yell, throw tantrums, stomp their feet, and make a lot of noise, and then once that angst is out of their system, they promptly tend to get to doing whatever it is that the authorities have told them they should do to "get ahead".

Exactly. Teenages only rebel in small ways; they don't really think about things from a big-picture point-of-view. I remember going to middle school and high school in an upper-middle-class suburban area: all the kids at my schools were die-hard Republicans (this was back in the Reagan/GHWBush years, so this meant something slightly different back then). It wasn't until they went to college that many of them started changing their views radically.

Re:Nice

[BlackHawk-666]

Nobody wants to hire a kid who thinks he can hack and then gets caught. It shows you can't be trusted at any job where you might need to touch a computer, phone or money. Worse still, it shows you lack skills.

People want to hire the guys who performed the hacks and got away with it, not the ones who overestimated themselves and failed.

Re:Nice

[Anonymous Coward]

Wow. That read (to me) as though you get some pleasure & felt superior by deriding someone. (We all can be guilty of that sometimes. Human nature I suppose, but not one of the better bits.) Regardless, IMO you're not not giving enough consideration to a few points:

• Anyone can get caught; the real world isn't a hollywood movie & there are no hacking "gods" (just different degrees of skill)
• Given that the GP (not me!) states "Despite never once taking advantage to having full reign over the system" I'd agree that the punishment was totally disproportionate - probably because the school was embarrassed to be caught with their pants down.
• He was a kid. What kid has never done something stupid?

Re:Nice

[jedidiah]

In the local school district, they use some office application I have never heard of before. This doesn't seem to harm the kids in any way. They aren't unable to cope with other brands of spreadsheet or word processor afterwards (like the Lemmings will claim).

This is kind of the way it should be for school in general. Teaching concepts and whatnot.

If you can't handle different brands of word processor, you're going to be totally f*cked when your brand of choice makes a major UI change in the future.

Re:Nice

[PopeRatzo]

If you can't handle different brands of word processor, you're going to be totally f*cked when your brand of choice makes a major UI change in the future.

That's for sure. I learned on Nota Bene. Now nothing throws me.

Re:Nice

[Anonymous Coward]

Teenagers don't question authority, by and large. They yell, throw tantrums, stomp their feet, and make a lot of noise...

Is this from your extensive experience coaxing young girls into your pedo-van?

Re:Nice

[Nivag064]

My youngest son started become a teenager when he was 6, he is almost 15 now and usually more mature than a lot of people 20+.

Re:Nice

[Anonymous Coward]

Exactly. He just did something he knew was wrong using a computer. That means it shouldn't count, he was using a computer and therefore everything he does makes him a hero. See also Aaron Swartz

Re:Nice

[Anonymous Coward]

you should have kicked the shit out of the principal, then they would have let you back in. i mean, come on!

Re:Nice

[Anonymous Coward]

Im 14 I got a rasbperry pi and I went through those holes on the school network that you speak of.
I got yelled at for that. a lot!

But thats besides the point i put my brother on arch on the rasbperry pi.
And I totally agree about the programming in school curriculum.
I brought it up with my principal and she said it would cost too much money!!!!
They freakin waste so much money as it is on pointless projects that never get done and field trips for stupid stuff.

When will they wake up????!?

Re:Nice

[Anonymous Coward]

The freaking government will hire them all day! :)

Re:Nice

[Anonymous Coward]

OOOOOOOOOOOH the microsoft office crap at my school was an idiotic move.
Guess what. They called it Information technology.
Man if IT was so easy that you could type words into libreoffice all day then anyone would do it.

Freaking ridiculous.

too late...

[Anonymous Coward]

already switched to osx years ago.

Re:too late...

[jedidiah]

> already switched to osx years ago.

Bought some Macs. Was not really impressed.

Glad I can buy a much wider range of PCs for far less money if I am not married to a single highly proprietary operating system.

The worst part about running MacOS is being stuck with Apple hardare.

Re:too late...

[Billly Gates]

Linux always breaks, apps are not tested, things crash, guis are horrendous and 10 years behind MacOSX and Windows, etc.

PC is great for running Windows which is more solid than Linux (dont give me an argument how BSOD happen every 15 minutes because back when you ran Win ME 15 years ago the same situation must exist now).

Linux only runs great on servers from tested releases like Redhat sadly.

If you hate Windows MacOSX is your only option for desktop work. Raedon drivers break every apt-get update rendering X useless.

Intel support is stellar this time.

[deviated_prevert]

Considering that the 14NM Broadwell chips are not scheduled to ship till the second quarter of 2014. With support for power saving per component coming along it looks like using the Linux kernel on laptops will also be much more inviting. It is all well and good that the advances in the kernel hardware support are keeping pace with what Microsoft is doing. I am still eagerly awaiting a great high end powerhouse Linux laptop. As it is the old IBM T42 non-pae clunker that I am writing this on is still very usable but if a company ever finally does ship an OS agnostic laptop with high specs I will jump at the chance.

The temperatures in hell are dropping but I am not going to hold my breath as Windows still holds the retailers and manufacturers by the balls to say the least. However with both Intel and AMD actively supporting the Linux kernel this quickly for their most important product lines perhaps a manufacturer like Samsung or Lenovo might actually try to market a real full blown Linux based device for a change instead of just dabbling in Android consumer craptronic devices.

Re:Intel support is stellar this time.

[Nerdfest]

I'm not sure what you require for high specs, but System76 has some decent 'gaming class' laptops that might fit the bill. Their Bonobo 17" maxed out a pretty decent machine, though quite heavy for a laptop.

Re:Intel support is stellar this time.

[WaroDaBeast]

[...] the old IBM T42 non-pae clunker that I am writing this on is still very usable

Out of curiosity, which distro do you run on that machine? I'm asking this question because distros that do not have PAE as a requirements are rare birds, so to speak.

Re:Intel support is stellar this time.

[deviated_prevert]

[...] the old IBM T42 non-pae clunker that I am writing this on is still very usable

Out of curiosity, which distro do you run on that machine? I'm asking this question because distros that do not have PAE as a requirements are rare birds, so to speak.

Mint but not current, but Slackware current can run it and so can Knoppix. You just have to chose the right kernel options.

Re:Intel support is stellar this time.

[deviated_prevert]

I'm not sure what you require for high specs, but System76 has some decent 'gaming class' laptops that might fit the bill. Their Bonobo 17" maxed out a pretty decent machine, though quite heavy for a laptop.

I am hoping that the newer chips next year will will sip power in laptop form and get battery times up over 10 hours like a tablet can. Unfortunately the gaming type of laptops being made right now, though powerful enough to do everything I do are still power hungry and heavy as a brick. The Broadwell specs are incredible however especially for laptops and with a good 6-9 cell battery should easily out distance anything on the market today for performance and battery times. I currently get 5 hours out of an aftermarket 9 cell add on on this old T42 which is still better than most high end laptops with i5s and i7s.

Re:Intel support is stellar this time.

[Anonymous Coward]

Just select a 486 kernel from your distro: works very nicely for my pentium-M fanless laptop (the cpu does not even support NX) currently running Debian Wheezy.

Re:Intel support is stellar this time.

[Anonymous Coward]

The biggest issue for a Lenovo or Samsung isn't Linux hardware support, it's the nightmare of customer support on Gnome3, and the coming X-to-wayland transition clusterfuck that drops next year.

Re:Intel support is stellar this time.

[Duncan J Murray]

My thoughts exactly, as I type this on a slightly more modern T60p, after my I dropped my T40 and sheared off the hinge last year.

To be honest, I'd be happy if Lenovo would simply sell a laptop without an OS if that's easier for them.

D

Re:Intel support is stellar this time.

[WaroDaBeast]

Choosing the right kernel options? Sounds like you build your own kernels. I asked that question because I also happen to own a machine that doesn't support PAE, and I was curious as to which distros (in its latest iteration) still offer non-PAE kernels.

I'm using Manjaro partially because of that, by the way.

Re:Intel support is stellar this time.

[deviated_prevert]

The biggest issue for a Lenovo or Samsung isn't Linux hardware support, it's the nightmare of customer support on Gnome3, and the coming X-to-wayland transition clusterfuck that drops next year.

On the contrary the apps that work on Gnome will all work on XFCE or for that matter KDE or LDXE or Enlightenment or whatever ....YOU DO NOT have to install the full Gnome DE just the primary libs. This is why Linux just work because the truth is you can do what you want with the desktop environments.

This mythological bullshit that is spread by morons that Linux is not ready for the desktop is always written by those who do not have a frigging clue about how it works and why it works. FOR instance I can run a music notation program called rosegarden without a hickup in any desktop environment I frigging well chose. AND ROSEGARDEN is a QT based program. If I have the correct GTK libs installed I can run all sorts of other stuff that runs on GNOME without a hickup.

Stop all the fud and bullshit once and for all you guys it is about to be exposed for what it really is BULLSHIT. Just try a full install of Knoppix and see what is possible, hell I don't give a rats ass if it does not have Firefox installed directly, I can just download the latest build and run it directly from my home directory and not even install it to USR in the first place. AND THIS IS why Linux will catch on eventually because more and more people are starting to look past the FUD AND BULLSHIT being spread around about Linux software and the very advanced and complete desktops that are available.

Re:Intel support is stellar this time.

[lsatenstein]

Considering that the 14NM Broadwell chips are not scheduled to ship till the second quarter of 2014. With support for power saving per component coming along it looks like using the Linux kernel on laptops will also be much more inviting. It is all well and good that the advances in the kernel hardware support are keeping pace with what Microsoft is doing. I am still eagerly awaiting a great high end powerhouse Linux laptop. As it is the old IBM T42 non-pae clunker that I am writing this on is still very usable but if a company ever finally does ship an OS agnostic laptop with high specs I will jump at the chance.

The temperatures in hell are dropping but I am not going to hold my breath as Windows still holds the retailers and manufacturers by the balls to say the least. However with both Intel and AMD actively supporting the Linux kernel this quickly for their most important product lines perhaps a manufacturer like Samsung or Lenovo might actually try to market a real full blown Linux based device for a change instead of just dabbling in Android consumer craptronic devices.

Do you want manufacturers and vendors onside? What does it take? It takes some free code and the right to add $100.00US per system. The manufacturer and the retailer need that to cover returns, repairs, some Geek staff and profits.

Would I pay $100.00 for a fully pre-installed system that works (graphics, sound, network, security, and user software selection), as an off-the shelf package? I guess I would! When I look at the time I spend on a downloads, on testing patches, and rebuilding my test Linux, I would spend that money. Time saved will give me more time to respond to slashdot about things. Important things that I would not be able to comment on with a limit placed on my spare time .

Re:Intel support is stellar this time.

[Blaskowicz]

Yes, like dealing with five different "Save as"/"Open File" window styles is so great. (GTK2 and GTK3 count for two different ones)

Re:Intel support is stellar this time.

[Anonymous Coward]

How high-end? My Asus UX32VD is >99% supported. I expect other Asus models to be similarly good. Samsung tends to be slightly worse supported than Asus but not painfully so.

sounds like a bloated pos to me

[Anonymous Coward]

i have enough win boxes i need to tweak on a daily basis just to get them to act like a computer.
now some newb decides he's gonna turn my iptables into roadkill?
FU.

Re:sounds like a bloated pos to me

[Gravis Zero]

i have enough win boxes i need to tweak on a daily basis just to get them to act like a computer.
now some newb decides he's gonna turn my iptables into roadkill?

U MAD BRO?

Re:sounds like a bloated pos to me

[Anonymous Coward]

What are you, some grandpa who can't learn new things?!

When the old shit is so fucked up by design that it can't possibly be made to work nicely, of course you have to have something new and shiny! This isn't some stupid BSD that works for 20 years with the same packet filter that can be adapted to new needs, you fool!

This. Is. Leeeenox!

Re:sounds like a bloated pos to me

[Anonymous Coward]

bro im old enough to still be pissed when they ditched ipchains, meanwhile bsd just works.

linux is the windows of unix, a subpar off brand alternative that got big because ibm promoted it.

Re:sounds like a bloated pos to me

[sjames]

No. There will be a translation layer that looks like iptables you can use.

Re:sounds like a bloated pos to me

[BitZtream]

Yes. Its another example of change for the sake of change rather than meaningful development. Thats obnoxious and wasteful.

Re:sounds like a bloated pos to me

[Anonymous Coward]

Great, so rather than continue to use what works and people have become familiar with, let's first spend the time and resources to replace it with a somewhat more functional replacement, then let's dump additional resources on a translation layer so that everything can superficially appear the same.

Re:sounds like a bloated pos to me

[sjames]

Yes. It's great when people do the right thing!

Re:sounds like a bloated pos to me

[timeOday]

I demand backwards compatibility with ipfwadm!

Re:sounds like a bloated pos to me

[Billly Gates]

XP user die hard joke.

BAH I HATE aero, I hate Metro, I hate bloat, I hate anything new and my 512 meg pentium IV is fine and I feel threatened by MS cancelling updates after just a mere 13 years of free work.

XP is the best just because I am familiar with it and like the green hills and blue sky background and its ugly fonts on LCD screen. Until a OS mimicks this precisely I wont change and look for a reason to hate X etc.

Re:sounds like a bloated pos to me

[Billly Gates]

Or just use FreeBSD.

I do for my virtual routers in VMWare. I guess at work you might not have much of a choice but I do not linux when it comes to routing. If you have any power and just need routing only FreeBSD and even OpenBSD runs on the same hardware and is really tuned for that kind of work. Linux seems great at running java app servers and other apps. Each OS has its own advantages.

Security fix backports

[tepples]

I still upgrade from time to time, but my attention is more focused on security than any supposed "improvements". I don't want to be the odd guy who is caught with some vulnerability that was fixed eight versions ago.

Some Linux distributors, instead of providing a new kernel that may break old applications or devices, instead backport security fixes to an old kernel.

Re:Security fix backports

[Anonymous Coward]

Some Linux distributors, instead of providing a new kernel that may break old applications or devices, instead backport security fixes to an old kernel.

Kernel upgrades are not suppose to break userspace.

Not all security fixes are always marked as such and get backported.

Re:Security fix backports

[Anonymous Coward]

I still upgrade from time to time, but my attention is more focused on security than any supposed "improvements". I don't want to be the odd guy who is caught with some vulnerability that was fixed eight versions ago.

Some Linux distributors, instead of providing a new kernel that may break old applications or devices, instead backport security fixes to an old kernel.

The Ubuntu LTS+HWE model is interesting; you can basically install the kernel from Ubuntu 12.10, 13.04, or 13.10 on your 12.04 LTS system.

Re:Security fix backports

[Anonymous Coward]

Kernel upgrades are not suppose to break userspace.

All drivers are supposed to be open source and applications should not depend on undocumented (or even documented) implementation details. Both these assumptions do not survive an encounter with reality. Sometimes this is expected (binary blob drivers), in other cases it might be a bug that only appears in a new kernel (scheduler change -> race condition, order of results changed). This is not even a Linux exclusive problem, the amount of application bugs that get exposed by changes on windows isn't small either and AFAIK microsoft even had to add checks for the process name in order to support older applications that broke with the new implementations (for example SimCity 2000 got a special memory allocator to avoid segfaults).

Re:Security fix backports

[Anonymous Coward]

Some Linux distributors, instead of providing a new kernel that may break old applications or devices, instead backport security fixes to an old kernel.

Why does Linus allow kernel updates that break applications and drivers? Oh... I forgot, because he can and there is no one to chew him out with vulgarity and death threats.

Re:Security fix backports

[Eravnrekaree]

a kernel update should never, ever break the user space. period.

Re:Security fix backports

[Dan Ost]

Some Linux distributors, instead of providing a new kernel that may break old applications or devices, instead backport security fixes to an old kernel.

Why does Linus allow kernel updates that break applications and drivers?

Because he has decided that those updates improve the kernel somehow. That's his job: to improve the kernel.

If some applications get broken when the kernel is improved, it's the application developer's job to fix them.

This is as it should be. Any other model ties the hands of the kernel developers and then they can't do their job.

Re:Security fix backports

[Blaskowicz]

It's nice (you can get newer Xorg as well) but by default makes your partition fill up with hundreds megabytes of unneeded kernel updates when you run apt-get dist-upgrade, unless you have the necessary knowledge to prevent the original kernel from being updated. And lots of crap to clean up in /boot. That's an annoyance.

Right...

[Anonymous Coward]

This is just a cash grab by the Linux developers.

i'm watching a stream of some fag play Knack...

[Anonymous Coward]

on PS4 that shit looks generic as fuck...is this really what the "next gen" is all about? this looks it could be a pc game from any time in the last five years. So fucking lame! PS4 fucking sucks!

http://www.twitch.tv/pkb0t33

Re:i'm watching a stream of some fag play Knack...

[0123456]

The PS4 is just a low-end gaming PC with a Sony sticker on the front. Of course the games are going to look like something a PC could play a few years ago.

Re:i'm watching a stream of some fag play Knack...

[Anonymous Coward]

Xbox One also. The hardware of the two is almost Identical, so don't play the fanboy card.

Re:i'm watching a stream of some fag play Knack...

[Blaskowicz]

You elitist, there's no way most people's three year old PC can run that kind of game, and not the Linux PCs. You at least run into shit like having a slightly too slow CPU, unless you talk about an old high end PC or one upgraded to not get under 20 fps in fire fights. Half the people are running laptops, too.

I sadly stopped caring about PC gaming : I have absolutely no need or desire to upgrade my CPU (3GHz dual core AMD) yet games are written for even faster dual cores, or for quad cores. I'm glad my PC works at all, thanks. I would need to get a motherboard with higher current or different socket, and that's a pain. I'm even using linux anyway which makes it hard or impossible to run even a ten year old game (except blizzard and valve ones).

blast! video card improvements!

[Gravis Zero]

sure i'll tear out my video card, fork over cash and put in a better supported one but which is best supported now?! they improved the Intel, AMD and NVidia video card drivers! dammit, cant they just improve one video card driver at a time?!

will these first world problems never end?!

bcache is a HUGE improvement for some workloads

[raymorris]

Bcache, merged in 3.11, improves IO up to 100X. Not 100%, 100X, or 10,000%. It may well be worth an upgrade if you're running a distro 2.3x and have random IO on multi TB storage.

Re:bcache is a HUGE improvement for some workloads

[AliasMarlowe]

Bcache, merged in 3.11, improves IO up to 100X. Not 100%, 100X, or 10,000%. It may well be worth an upgrade if you're running a distro 2.3x and have random IO on multi TB storage.

The multi-queue block layer which is merged in kernel 3.13 gives a 3.5x to 10x increase in IOPS. This change is mostly targeted for SSDs, but gives similar improvements on HDs as well. However, it's not clear whether this improvement is relative to 3.11 or not.

Re:bcache is a HUGE improvement for some workloads

[Anonymous Coward]

Bcache (merged in 3.10, btw) has one very big drawback for me. To prevent writing to the backing partition (your 'multi TB storage') outside of bcache you have to convert your those partitions to some bcache format that writes a custom superblock. As far as I can tell this conversion is one-way only and the tool to do it in-place (as opposed to format-and-restore-from-backup) is not supported by the bcache folks, although I may be wrong here.

Semi-related ps: For non-enterprise workloads there are several fuse-based caching mechanisms that do not need a conversion of the backing store.

Re:bcache is a HUGE improvement for some workloads

[Anonymous Coward]

Hyperbole. Bcache may improve IO 10x - 100x for highly localized and frequently re-used blocks. It does NOT improve overall IO up to 100x.

Re:bcache is a HUGE improvement for some workloads

[drinkypoo]

Bcache (merged in 3.10, btw) has one very big drawback for me. To prevent writing to the backing partition (your 'multi TB storage') outside of bcache you have to convert your those partitions to some bcache format that writes a custom superblock. As far as I can tell this conversion is one-way only and the tool to do it in-place (as opposed to format-and-restore-from-backup) is not supported by the bcache folks, although I may be wrong here.

This is precisely why I have not even tried to implement it, even just to see if it's good. This requirement is a complete non-starter, and I have never heard any technical reason why it is a necessity. Indeed, if the idea is to cache blocks, then you should be able to cache any kind of blocks. If that is unworkable within this architecture, the project should be thrown away and reinitiated by someone with some standards. Forcing a new format on users that won't be back-compatible with older kernels and distributions when it's not necessary is completely unacceptable.

Sad, though, because I really want what it does. Perhaps someday dm-cache will make it into mainline. It does what we want it to do, but it's not mainlined and it only appears for some kernel versions - looks like 3.0.8 is the latest.

Re:bcache is a HUGE improvement for some workloads

[Anonymous Coward]

EnchanceIO is an alternate ssd caching solution for linux. It's a kernel module that can be setup against any existing FS and can be attached/removed even whilst the FS is mounted. Supports writeback/writethrough and a bunch of cache replacement algorithms.

Re:bcache is a HUGE improvement for some workloads

[drinkypoo]

EnchanceIO is an alternate ssd caching solution for linux. It's a kernel module that can be setup against any existing FS and can be attached/removed even whilst the FS is mounted. Supports writeback/writethrough and a bunch of cache replacement algorithms.

Well, I used teh google and found out that it's a commercial product but that it's also open source. Actually, it's Free Software, apparently GPLv2.

So what I get from this is that this is basically dm-cache as a module. Yes? Hmm, either I need to install more deps, or this isn't going to work on 3.11

Re:bcache is a HUGE improvement for some workloads

[__1200333]

Perhaps someday dm-cache will make it into mainline.

dm-cache is in mainline since 3.9. Now please test it and let me know if I should bother trying it!

https://www.kernel.org/doc/Documentation/device-mapper/cache.txt
http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=c6b4fcbad044e6fffcc75bba160e720eb8d67d17

Re:bcache is a HUGE improvement for some workloads

[drinkypoo]

dm-cache is in mainline since 3.9. Now please test it and let me know if I should bother trying it!

Well, crap. Now I have to copy my SSD's contents to my HDD... I'll get back to you, I guess. If it's even in my kernel, or if I can even get it to compile in.

Re:bcache is a HUGE improvement for some workloads

[Nivag064]

100X == 100X
100X == 10,000%
100X != 100%

I must have missed the point raymorris was trying to make...

Re:bcache is a HUGE improvement for some workloads

[drinkypoo]

Hmm, perms problem. Wacky. This is the only cachine solution I have actually figured out so far. Apparently it is really straightforward and actually can cache your root volume, but only write-through. That's fine, that's what disk buffers are for.

Re:bcache is a HUGE improvement for some workloads

[Anonymous Coward]

You did miss it, because raymorris used a comma where he meant to have an em dash or colon (or a preposition):

Originally: Not 100%, 100X, or 10,000%.

Correct: Not 100%: 100X, or 10,000%.
Also correct: Not 100%---100X, or 10,000%.
Also correct: Not 100%, but 100X, or 10,000%.

Firewire still a PITA!

[Anonymous Coward]

I've been using linux for a while and until I can hookup my firewire Edirol FA-66 straight up from an install with any distro I will
Continue duel booting with windows. If something does work and I update it.It ends up breaking something. Or the next release distro wont work at all I'm getting extremely p&$@ ed off of late. Sense about 8.04 of ubuntu. Version 10 or 12 of slackware. Around 2011 with distros thing have just gone ta crap. Config files init scripts get changed, shifted, fuggin moved around short linked to something somewhere .It gets a bit teeny weeny annoying. And by 2013 you'd think things would have been down pat by now!Cmon. Im going to grit my teeth have shell out for a MAC!. Or sell everything to get linux in a Korg OASIS. If friggin Korg can do it right someone else can.I prepare to pay maybe not the equivalent of the Korg OASIS. I just want something that works!.

Re:Firewire still a PITA!

[Anonymous Coward]

Don't hold back, tell us what you really feel :)

The FA-66 works for me (TM) on Debian with the ffado driver, if I recall correctly it worked out-of-the box (no futzing with udev/device nodes what have you). You _have_ to use jack-firewire though. If you want a gui you could try qjackctl (sp?) I guess, but have not tried it myself. As for updating your distro: Debian rarely introduces regressions when (dist)upgrading, while Ubuntu is very, very bad in this department imnsho.

Re:Firewire still a PITA!

[Anonymous Coward]

You don't need to shell out anything for a MAC - you already got one free with your NIC.

Re:Firewire still a PITA!

[HiThere]

Sorry, but you need a correction:
Debian rarely introduces regressions when (dist)upgrading the stable tree,

There are not-infrequent regressions when upgrading testing. I expect sid is worse. Serious regressions are much less common, but I used to figure that at least once during a cycle of testing the machine would get so borked that I'd need to switch to using stable for a week or so. OTOH, that didn't happen during the last cycle, and hasn't so far this time. But smaller regressions happen on nearly a weekly basis. (And I no longer EVER autoremove on a testing distribution. It's usually safe, but it's occasionally required a total reinstall.)

Stable, however, is, as it should be, stable.

Ad: Innovation Alliance

[symbolset]

Oh God that is one hilarious ad landing page. Well worth the click to earn your innovation protection badge!

tubg1rL

[Anonymous Coward]

a BSD Box (a pIII

Re: much improve

[Anonymous Coward]

much improve. so amaze. wow.

mod do3n

[Anonymous Coward]

accGounts for less rules to follow The problems United States. - Netcraft has log on Then the are having trouble the wind appeared politics openly. of reality. Keep

very nice management software

[Anonymous Coward]

Bahaquote quote features as given in its website ‘Add delete customers on the fly’, ‘Attach images’, ‘Permission Levels’, ‘Supervisory controls’. All I can say is that they stand up to the expectations.

And it still contains the binary blobs you love...

[Anonymous Coward]

https://en.wikipedia.org/wiki/Linux_kernel#Firmware_binary_blobs

iptables

[HussamAl-Tayeb]

Will there be a compatibility layer so I can still use the iptables application/syntax for a while?

Re: iptables

[behrooz0az]

Yes, There is a compatibility layer already.
I think it was mentioned on phoronix like 2 weeks ago.You may find it you go back in their archives far enough.

Re:iptables

[Anonymous Coward]

A better question is when will the documentation for iptables be finished? If NFTables is actually going to have good docs, then I say make a clean break and forget iptables ever existed.

NOUVEAU's most important updates yet.

[DMJC]

What's not mentioned in the news article is that the most important updates to Nouveau are being posted in the 3.13 kernel. There's going to be Power Management support for most current generation graphics cards. This is a huge thing. Performance wise it is going to lead to a massive jump in performance on the Nouveau drivers. The only other outstanding parts of nouveau are the OpenCL support, and the SLI support. After this update it should be possible to use the nouveau drivers for a lot more serious 3d work than they have been used for in the past.

Re:NOUVEAU's most important updates yet.

[Anonymous Coward]

"Outstanding" is one of those words can mean one thing or the opposite. In this case upon first reading I thought you claimed that OpenCL support is excellent, and on next reading realised that it was absent.

Next, fix the desktop

[jones_supa]

The kernel is now done. It has been done for years. Of course new hardware comes and needs to be supported. But everything in that department is rolling quite nicely. The kernel guys know what they are doing. The Linux kernel is stable and if a problem pops up, it gets fixed.

So these days the kernel is a nice black box which I don't have to worry about. Now, fix the desktop. That's where the interesting stuff is happening. Fix the terrible performance problems and lack of configurability of Unity. Make a rich graphical configuration tool for touchpads. Make the boot process beautiful: currently I just see the distro logo flashing in and out with some occasional scary lines printed in framebuffer console. Fix the little glitches here and there (quality assurance?!). Make DVD burning work correctly. Make it so that I have to never fight video tearing.

Re:Next, fix the desktop

[Anonymous Coward]

You should check out Elementary OS.

Re:Next, fix the desktop

[Anonymous Coward]

None of that has anything to to with Linux, which if you please remember is not an operating system. It is just a kernel.

Re:Next, fix the desktop

[Mike Frett]

Or just use Xubuntu......

Re:Next, fix the desktop

[jones_supa]

I have to agree that Xubuntu is the least broken one.

Re:Next, fix the desktop

[drinkypoo]

Not only is the kernel never done, but the GUI is worked on by completely different people. I don't know the name of your logical fallacy, but I like to call it the GNOME and KDE fallacy in nerdland. If all those people just worked on one DE, it would be great right? No, it would suck horribly. Too many cooks with their own ideas. But you're even farther off the mark because you want people who like to write kernel code to write application code. That's like asking a cabinet maker to build you a house... or asking an ordinary carpenter to build your cabinets. Either one can probably do the job, but they're specialized for a reason.

Re:Next, fix the desktop

[jones_supa]

I'm not asking the kernel guys to write the desktop software. :D I just wish the desktop stack had a similar level of quality to the kernel.

Re:Next, fix the desktop

[jones_supa]

Well, you and I need more than the kernel. By the way, obligatory xkcd.

Re:Next, fix the desktop

[Anonymous Coward]

Have you considered using KDE?

Re:Next, fix the desktop

[jones_supa]

Yeah, KDE and XFCE are best quality-wise.

Re:Next, fix the desktop

[Zontar The Mindless]

I would say "Get off that 'buntu junk and onto any real Linux distro"... but then I'd get modded Troll.

Honestly, it seems like every time there's a Linux story on /. these days, half the posts are complaints specific to Ubuntu.

If you don't want to fight your system/desktop, use a different one. I suggest CentOS with Window Maker DE for you guys. (I use OpenSUSE/KDE4 myself, just so you know.)

You use Linux because Linux means you've the freedom to choose, right?

SO QUIT BITCHING,
AND START SWITCHING!

Re:Next, fix the desktop

[Zontar The Mindless]

And Window Maker for units with older/minimal graphics hardware.

BTW, as someone who bitched incessantly about the train wreck that was the initial release of KDE4, I must say the devs turned it around very nicely. I would now happily recommend it to anyone who wants released from the fetters of the Gnome desktop or wishes to avoid further interaction with the new and improved train wreck that Unity is shaping up to be.

Personally, if I even used Ubuntu and I'd not done so already, I'd be been running away from it just as fast I could go.

Re:Next, fix the desktop

[Anonymous Coward]

hey, you no more drinky! you sleepy sleepy now!

3.13 = MAAAAAJOR FEATURES.... 4.0 = minor bugfixes

[Anonymous Coward]

What the fuck is wrong with you Linus?

Re: SteamOS Will 'Really Help' Linux On the Deskto

[Anonymous Coward]

If your programmers cannot figure out which libraries to use and/or link everything statically, might I suggest that instead if crying fragmentation they look for a different profession.

BTRFS stable when

[Eravnrekaree]

When is BTRFS finally going to be declared stable and become default on major distros? Its features were needed years ago. The Copy on Write features are killer features that have been needed on Linux for years, such as to implement a filesystem level versioning, system restore an restore point feature and improved snapshot features. Ext4 is only a stop-gap and Ext is really starting to show its age.

Re:BTRFS stable when

[Anonymous Coward]

Yeah, Copy-on-Write is just wonderful for SSDs where a precious few erase cycles can be used up willy nilly by software developers who think hardware resources are infinite.

Re:BTRFS stable when

[Anonymous Coward]

Supposedly OpenSUSE was going to set BTRFS as default during installation.

Re:BTRFS stable when

[MarsLander]

Try ZFSOnLinux. It's not merged into the kernel mainline for licensing reasons, but it is easy enough to install, is well tested, and has all the features you need (and more).

Re:BTRFS stable when

[drinkypoo]

Requiring a component that isn't just found on your distribution's LiveCD is a recipe for nightmares later if something goes wrong. Your suggestion is wholly inappropriate for anyone who might ever have to work on their computer.

Re:BTRFS stable when

[Eravnrekaree]

Thats a total non issue, you turn off COW on those devices, or anywhere its not wanted. Duh? I would fully expect that COW should be an optional feature, in fact, it probably wouldnt be used until you turn it on by creating a restore point, snapshot, ot something such as that.

Re:BTRFS stable when

[fnj]

ZoL all the way. I know longer care if the BTRFS glacier _ever_ arrives.

I was hesitant, thinking ZoL was toy status, but I bit the bullet, installed and took the learning curve. It seems fully mature to me. I had confused ZoL with the ZFS Fuse toy, but ithey are completely separate things. ZoL is a high performance, reliable and mature "real" kernel mode file system.

Creating an 18TB double parity RAID-Z2 storage pool takes only a handful of seconds and is completely ready to go. There is no traditional long "build" stage. In general all "mkfs" operations are essentially instantaneous.

For me on CentOS6 it was a simple repo addition and "yum install". It hooks into DKMS for when I do future kernel upgrades.

Re:BTRFS stable when

[Bengie]

19nm 250GB SSDs are good for about 500TB of writes. Most people aren't too concerned. They already figured out how to make SSDs have infinite writes, they just need to figure out how to mass-produce the feature. Upcoming MRAM based SSDs, which are scheduled for the next 1-2 years, will inherently have no write limitations, so no wear leveling at all. You can already buy a 750GB SSD for $500.

Re:BTRFS stable when

[Anonymous Coward]

I agree. I have been running ZFS on my Linux boxes for over a year and it has been a great experience. I don't really care if Btrfs ever becomes stable/popular, there is already a solution in place.

Re:BTRFS stable when

[higuita]

you know, everytime you write anything on a SSD, you are really doing a copy-on-write on firmware level... the wrote block is really fully read and copied with the changes to a new block, So the btrfs copy-on-write is not doing anything worst

Re:BTRFS stable when

[Anonymous Coward]

I've been wondering if/when we would see MRAM devices. Are you sue of that 1-2 year timeframe and can you cite sources? Given MRAM's characteristics I was expecting we'd see it replace DRAM first...

Re:BTRFS stable when

[KiloByte]

That "glacier" works just fine since many years ago. Don't confuse "in active development" with "unstable". When it comes to data safety, btrfs runs circles around any other filesystem.

Re:BTRFS stable when

[Bengie]

Whoops, sorry, meant ReRAM. Same difference :p No wear leveling and DRAM like speeds and non-volatile. Assumg you're curious about ReRAM and not the MRAM that i originally said http://www.xbitlabs.com/news/memory/display/20130625232112_SK_Hynix_Further_Delays_Commercial_Production_of_ReRAM.html

Looks like it got pushed back a bit further.. /cry

Last year an HP visionary predicted that smartphones and tablets would start utilizing ReRAM-based storage sometimes in 2014 - 2015. However, with current plans to start commercial manufacturing of ReRAM in late 2013 it looks like the first mass products featuring the technologyare only going to emerge in 2015 - 2016.

Re:BTRFS stable when

[Bengie]

I've heard nothing but horror stories about BTRFS, from even the past few months. People talking about anything from random data loss to full on entire FS went bad. What I've read anyway.

Re:BTRFS stable when

[fnj]

Unstable means you better not rely on it. If it could be relied on, RHEL7 would be rolling out BTRFS as a full first line option, or as default - but they are not. Unstable means RHEL6 can't read a BTRFS volume formatted under Fedora18. Come on, it has been 5 years since BTRFS 1.0, and it is still "experimental".

Stable means you can exchange volumes between any linux system over the last 5 years or so without a problem. I can do that with ext4.

Re:BTRFS stable when

[Zontar The Mindless]

Supposedly OpenSUSE was going to set BTRFS as default during installation.

I'd really interested in knowing where you think you heard this.

Re:BTRFS stable when

[KiloByte]

Btrfs features that break backwards compat are not enabled by default. Same with ext4.

Re:BTRFS stable when

[Anonymous Coward]

Not so, it suffers the same problem as all file systems when it comes to database files. Taking a snapshot of the system in the middle of writing key blocks (effectively breaking the key chain) still gives you a corrupted database file. or a database record gets half written out.

Re:BTRFS stable when

[KiloByte]

A database that can't handle power loss in the middle of a write would break on other filesystems as well, so btrfs is no worse. Except that, if you're ok with filesystem-specific code, you can use a btrfs transaction, which allows consistency without multiple fsyncs.

Re:BTRFS stable when

[lsatenstein]

When is BTRFS finally going to be declared stable and become default on major distros? Its features were needed years ago. The Copy on Write features are killer features that have been needed on Linux for years, such as to implement a filesystem level versioning, system restore an restore point feature and improved snapshot features. Ext4 is only a stop-gap and Ext is really starting to show its age.

SUSE13.1 is saying that BTRFS is suitable for home use. They see no problems in a desktop or non-banking industry environment. Ditto for Mint16, and Fedora20. Btrfs is here to enjoy. But... for most work that I do, ext4 provides faster I/O. I timed it so I know so.

Re:BTRFS stable when

[lsatenstein]

Whoops, sorry, meant ReRAM. Same difference :p No wear leveling and DRAM like speeds and non-volatile. Assumg you're curious about ReRAM and not the MRAM that i originally said http://www.xbitlabs.com/news/memory/display/20130625232112_SK_Hynix_Further_Delays_Commercial_Production_of_ReRAM.html

Looks like it got pushed back a bit further.. /cry

Last year an HP visionary predicted that smartphones and tablets would start utilizing ReRAM-based storage sometimes in 2014 - 2015. However, with current plans to start commercial manufacturing of ReRAM in late 2013 it looks like the first mass products featuring the technologyare only going to emerge in 2015 - 2016.

Two questions that I would like to have answered follows "Is reRAM more power consumptive during operation?" Are we therefore looking at RF radiation noise appearing as white noise?"

In my early IT years, we bought magnetic cores, 28 guage wire and we wired up and tested our own core memories. We needed some bit values to be remembered if there was a power failure. The environment was hostile to moving memory. (Circa 1070 is when I did this stuff). Finally, we bought magnetic latching relays, which turned out to be a less expensive approach, but required lots more cabinet space.

Re:BTRFS stable when

[Anonymous Coward]

"...When it comes to data safety, btrfs runs circles around any other filesystem...."

This is interesting. Do you have any research papers backing up this claim?

Here are at least two research papers by professors that examine ZFS ability to combat data corruption, and ZFS seems to be completely safe:
http://en.wikipedia.org/wiki/ZFS#Data_integrity

I have never seen research on BTRFS, which actually is in alpha stage, not even beta. Alpha is active development and they add new features. Beta is when you iron out the bugs and do minor stuff, trying to make the product stable just before release. Like when people are in the beta phase and report bugs, but no new major features are added. BTRFS adds new major features all the time - so it is under heavy development, hence, alpha stage. I have never seen a filesystem under active development being safe? Do you have any links, or is it just an opinion?

Re:BTRFS stable when

[Wolfrider]

--I usually like/agree with your posts, but you might want to think outside the box a little here.

--I've been testing ZFSOnLinux for the past few weeks and I think it's getting close to production-ready; they just added support for POSIX ACLs (for Samba) and it works pretty well with a minimum of configuration. My test rig is a ~2005 older dual-core 64 bit with 4GB of RAM and secondhand spare hardware running Debian Testing--64, and with a few tweaks I'm getting ~50MB-70MB/sec with a ZFS 2x320GB-disk mirror over Samba. The HDs are from 2 different manufacturers and still work well together.

--As far as recovery goes, you're not limited to CD/DVDs; you could install an entire recovery environment (including ZFS, etc modules) to bootable USB stick, with a bit of forethought and planning. (Altho if your equipment is old enough that it doesn't boot from USB, ZFS is prolly not for you as it really only works well with 64-bit and at least 2-4GB of RAM.)

--Certainly ZOL is not for everyone, but development on it is still *years* ahead of Btrfs. If you haven't done anything with it yet, I'd strongly recommend you put together a test rig out of COTS parts and see just how good it is; the docs on zfsonlinux.org are good enough to get started with. ;-)

Re:BTRFS stable when

[drinkypoo]

I've been testing ZFSOnLinux for the past few weeks and I think it's getting close to production-ready;

That does not address the issue of it not being included in distributions.

As far as recovery goes, you're not limited to CD/DVDs; you could install an entire recovery environment (including ZFS, etc modules) to bootable USB stick,

Right. So if something happens to my stick I'll have to produce another one before I can do troubleshooting. If I'm in the field with no resources, I'm just boned.

Certainly ZOL is not for everyone, but development on it is still *years* ahead of Btrfs.

what I'm really suggesting is that the filers not rin Linux.

I prefer to always suggest Linux, but not this time.

Re:BTRFS stable when

[KiloByte]

BTRFS has been used in production on many sites. Including mine, but this might be too small to count for you.

And as for data safety features: a good part are shared with ZFS, including checksums and send/receive, but some rely on BTRFS' internal format, which resembles log-structured filesystems more than classical ones. One that saved the bacon of an under-backuped disk of a customer of mine was that BTRFS never overwrites data in-place but always writes the new version of both data and metadata to a random place on the disk. The newest version of the root node and some recently written stuff was in the area that got corrupted, but older generations of the data existed elsewhere.

Re:BTRFS stable when

[J053]

(Circa 1070 is when I did this stuff)

So, were you William the Conqueror's IT guy?

Re:BTRFS stable when

[lsatenstein]

Next to the 9 of course is the 0, and therein was my error. I should have proofed my reply, but from a tablet, it is a problem of too small a font to notice the finger slip from the 9 to the 0.

Glad you noted it.

Re:BTRFS stable when

[Wolfrider]

> Right. So if something happens to my stick I'll have to produce another one before I can do troubleshooting. If I'm in the field with no resources, I'm just boned.

--That's what backups (spare sticks), redundancy and online storage are for... ;-)

--But if you're dead-set against it, I won't bug you with it - NMJ to try and change your mind. I just see a lot of value in ZOL and wanted to pass it on. :-)

Re:BTRFS stable when

[Anonymous Coward]

I have ZoL on my home server and I love it. Granted it's not some enterprise installation but I have had no errors with it at all.

Re: SteamOS Will 'Really Help' Linux On the Deskto

[Anonymous Coward]

If you think that is the only issue with fragmentation, Linux tards are even bigger morons than I gave them credit for.

Re: SteamOS Will 'Really Help' Linux On the Deskto

[knightghost]

Dynamic linking usually fails. New versions of code rarely adequately support backwards compatibility. Java is probably the worst offender. Any long term stable system is going to have 20+ versions of java or other libraries to maintain stability of its programs.

Define "much faster"

[wonkey_monkey]

the multi-queue block layer is supposed to make disk access much faster on Linux

What do you mean by "much faster"? Have we been chugging along in the slow lane all these years?

Re:Define "much faster"

[higuita]

until now, you only had one IO queue, taken care by a single CPU (even if your machine had hundred of cpus). For slow HD, it didn't matter much, as the bottleneck was usually the HD (unless you had huge array of disks). with SSD you can hit the cpu limit, trying to process all those request. with this patch, you get one queue per cpu, so you increase a lot the parallelism of IO on high speed IO devices.

For home users, it probably will not do much difference, as they don't usually have big IO setups, but for enterprise, virtualization and storage setups, it will make a big difference. Solaris had this for many years and that is why on big storage setups solaris had a very good speed.

TRIM support?

[jones_supa]

I wonder when the deficiencies in SSD TRIM support will be fixed. I don't think there's still any distro which enables discard support automatically in fstab.

HAL, for example

[tepples]

You mean like the deprecation of HAL in favor of udev, breaking user space programs that rely on HAL?

workloads like mysql with a TB disk

[raymorris]

I said "for some workloads" twice. Specifically those "localized" workloads would include a web server with a MySQL database, a mail store, or other frequently accessed files - a very, very common workload. The database and other frequently accessed files end up on SSD while the large, sequentially accessed files such as videos stream from spindles.

I also said "up to" - in some cases it might not be 100 times as fast, but "only" ten times as fast.

  For some common types of workload, bcache (or dmcache) makes a big difference.

4.0 is not for now

[higuita]

Linus said that he wants to have a 4.0 in about a year, not for now... and that he wants a clean, more bug fixed version (to see if distros use it as a long term stable version, instead of something like 2.6.32 or 3.2.x

Linux always has major improvements

[jd]

The problem is with the definition of major. There are almost never minor improvements to the kernel, using conventional ideas of major and minor. Ergo, these terms need updating or removing. They serve no useful function in the context of the Linux kernel.

Timetable

[wjcofkc]

It seems kernel releases are becoming more frequent. One thing I did not see in the article (unless I read right over it), is when this kernel is supposed to be released. Does anyone have a clue or can you point me in the right direction?

Re: SteamOS Will 'Really Help' Linux On the Deskto

[jedidiah]

...says the troll that's probably touched Direct3D never.

Re: SteamOS Will 'Really Help' Linux On the Deskto

[jedidiah]

I still play my Loki games despite every clueless git trying to make pronouncements about things they don't have any experience with (game programming or app programming in general).

FTFY...

[Burz]

The biggest issue for a Lenovo or Samsung isn't Linux hardware support, it's the nightmare of customer support on Gnome3 + Cinnamon + KDE + XFCE + LXDE + Unity...

Also, I'll say that I actually like Unity except for the one feature (which everyone else seems to hate also)... replacing the 'start' menu with a search window. People often want quick access to apps that they don't run every day and thus can't remember the names, so having a bunch of accessory, control panel and utility icons crowd into the list when I start typing what I thought was the app name really REALLY sucks. Listing apps hierarchically under a handful of categories is what works; the user should have both a menu and search available, and not be herded into primarily using search.

The window handling, menus and taskbar in Unity are an improvement although they take a little getting used to.

As for Wayland, I think it will flounder the way X11 has been doing. It is one thing to design software components by committee... its quite another to do so through a committee of committees (which is what many FOSS consortia are); it took these pretenders decades to get display configuration half-right. Mir may or may not pan out, but it seems to have a strong vision for what the graphics layer needs to be.

Re: SteamOS Will 'Really Help' Linux On the Deskto

[knightghost]

I still play my Loki games despite every clueless git trying to make pronouncements about things they don't have any experience with (game programming or app programming in general).

1 example proves infinite? Now which logic fallacy is that...

dm-cache benchmarks better, is less sexy, but ...

[raymorris]

The benchmarks I've read, which were reviewed by the kernel mailing list, indicated that dm-cache has the best performance in many cases. My gut feeling is that I'd rather use bcache, but I don't know why.

The current benchmarks have one huge failing, though. They test random IO by doing truly random IO all over the disk. Real random writes, in real workloads, is concentrated mostly in a relatively small number of blocks, such as the database and log files. That's important because the caching systems put the frequently accessed blocks in cache. True random benchmarks, with no blocks being frequently accessed, counteracts what the cache is doing. What's needed is a set of benchmarks run with random IO within four files of a few GBs each , to simulate a database, mail store, or other frequently accessed region of the disk.