Slashdot Mirror

Feel free to point out specific examples where Linux that has been ahead of the curve.

Okay, I just did what I told you to do, and I went to LWN. This week is for LWN subscribers, but last week is freely available, and since I wanted to link it, I wen there.

And I found a fascinating discussion of patching the Linux memory management system to save power by shutting down parts of the RAM that aren't needed! Which version of UNIX did they copy that from? Or will you concede that this is new?

LWN kernel page from last week

There are plenty of other examples; feel free to actually read the LWN archives. LWN is great.

Because Linux was free and good enough, and it was being adopted in droves on the backend. So IBM decided to support it, rather than be left behind.

Linux was "good enough" that people were not buying as many IBM big-iron machines. Now IBM sells its big iron as a way to run lots of Linux VMs. As I said, a disruptive technology, and I don't see why you are so determined to denigrate Linux as a mere not-quite-as-good-as-UNIX.

steveha

Linux was just a Unix clone. I can't think of any area where it stood out as driving innovation.

It started out as a UNIX clone, but has lots of innovation going on under the hood. Subscribe to Linux Weekly News and read the kernel updates every week, and you will get a better feel for the innovation going on.

Note that IBM is pushing Linux. IBM used to push their own UNIX, AIX; but now they have taken all the best features from AIX and ported them to Linux. Can you think of any reason why IBM might have done that?

Linux has been a disruptive technology, and if it were such a blah "me-too" technology as you seem to think, I don't know why it would have been so disruptive.

PostgreSQL: Just a relational database, and usually behind the heavy-hitters in terms of features. Mainly notable for at least being competitive with the big, commercial databases.

Oracle: Just a relational database, and I hear it can be a real pain to work with it. Mainly notable for being well-supported and crazy expensive. See, this sort of negative comment even works on the industry leader.

Let's face it, SQL was invented in the 70's and every SQL database system is "just a relational database" with some combination of features and price.

And let's face it, mostly people just need a relational database that they can trust with their data. There are a few companies that have very specific needs that only Oracle can handle, but most would be just fine with PostgreSQL.

For innovation, how about NoSQL? For some purposes, these work better than SQL; for others, not; but you have to admit these are not "just another relational database".

In the area of desktop environments, GNOME 3 is doing something really different. Lots of people hate it, understand, but it's definitely a new environment. And I think Enlightenment was pushing the envelope a lot in the early days.

The GNOME 3 example shows that sometimes innovation is met with resistance. Some innovations are not popular and languish in obscurity (deserved or not). I think there are a lot of innovative open source projects you have never heard about. (I have a Python project I thought was pretty innovative, but the problem it solves doesn't seem to be a problem people have, because nobody seems to care. On the other hand, ElementTree is very popular and I would call it innovative; if you disagree, what exactly do you think ElementTree is a clone of?)

steveha

If anyone here doesn't keep up with Linux Weekly News, here's a example article and discussion thread involving said Florian Mueller:

http://lwn.net/Articles/434318/

Never knew much about him, but still was startled by his descent into on-payroll patent troll.

I think it's grossly unfair to characterize the huge amount of effort that someone puts into a filesystem as merely being a case of NIH, when something genuinely interesting and better is being produced. For example, you can remove a disk from a Btrfs volume (or shrink the volume) while it's in use. There's also a tool that not only converts ext3/4 volumes to Btrfs in-place, but also does it in such a way that you can mount the volume in Btrfs, change your mind, and go back to using it in ext4, because the ext4 metadata is stored in a sparse Btrfs file, preventing the metadata's destruction until it that file is deleted, finalizing the conversion. I wish I could remember the thing I originally read, but this article I just found is also really informative.

BtrFS is basically an open ZFS, so that should tell you something.

WRONG!

Please educate yourself; knowledge is right here.

Stallman doesn't post on Slashdot - his highly efficient setup for web browsing does not support the new ajaxy comment system. ~

Personally I think it is not stupid to say it if it actually happens. It doesn't mean it isn't stupid to do it. BTW, on the FireStar patent Sun was able to get it invalidated:
http://lwn.net/Articles/289747/

The absurd claim being made is that somehow you have the legal standing to interpret whether Android is a derivative work under the GPL.

I agree, and that's why I never said I had legal standing. This is a discussion site where users comment on all sorts of things in which they have no legal standing. Whether a particular GPL developer feels that have been infringed or not, however, doesn't decide if the GPL was infringed.

The primary author, Linus Torvalds,

Linus hasn't been the primary author for many years -- these days he just accepts patches, and mainly just accepts ones from his trusted lieutenants. The Linus kernel contains millions of lines of code, and he doesn't require copyright assignment, so his legal standing is greatly diminished from the early days.

Feel free to send him an email and tell him he has to sue TiVo, and Cisco, and Google, and D-link, and AMD, and nVidia and most other audio and video device makers. I'm sure he'll appreciate your opinion.

Actually, you won't find non-GPL code from companies like nVidia in the kernel. They provide a GPLed interface that they then write proprietary blobs to. In the past few years, even firmware has been moved out of the kernel:

Moving the firmware out

Cisco and D-link have both been sued for violation of the GPL.

The GIMP guys are working towards support for 16 bits per channel. I was hoping to learn about the progress toward that, but I didn't see it discussed here; mostly people were griping about UI.

As I understand it, the GIMP core engine has 8 bits per channel pretty much hard-wired into it and it would be a pain to fix that. Instead, the GIMP guys have been working on a new engine called GEGL, and this was designed from the ground up to handle higher bit depths and to allow non-destructive editing. I believe GIMP 2.8 is the target for full GEGL integration. GIMP 2.7.x has GEGL partly integrated (used in some filters).

Meanwhile, if you need something like GIMP that supports high bit depth right now, take a look at CinePaint, a fork of GIMP hacked to support 16 bits per plane; it has been used for post-processing in movies. According to Wikipedia, it has fallen out of use because GIMP can do anything it can do; but that seems wrong to me, because GIMP doesn't have 16 bit per plane support yet.

I was really hoping that people who know about this stuff would post about it here.

P.S. It's from 2007, but here is an article about HDR photo editing on Linux. http://lwn.net/Articles/225652/

steveha

I also analyzed Google's amicus curiae brief in the Bilski case

We've seen what your so-called "analysis" is worth

Posted Apr 10, 2011 17:57 UTC (Sun) by cyd (subscriber, #4153)
Hi Mr. Mueller,

I am flattered and pleased that you have deigned to reply! Not even in my dreams have I conceived of conversing, even over the Internet, with someone worthy enough to have been credited in three separate (!) Blizzard Entertainment titles. However, I sense that you are being overly coy in stating your qualifications. No doubt there are numerous employee-of-the-month and high school yearbook awards that you have omitted to mention, due to a misplaced sense of modesty. Please, don't hold back on this account.

Posted Apr 10, 2011 18:01 UTC (Sun) by FlorianMueller (subscriber, #32048)
@cyd, you are actually right that I didn't list all of them. CNET Networks UK handed me (jointly with an activist organization) the Outstanding Contribution to Software Development Award, Silicon.com put me on the list of the 50 Silicon Agenda Setters, Linux Magazine named me runner-up to its Outstanding Contribution to Linux & Open Source Award. But the reason I didn't mention those awards is because I wanted to focus on those with the most impressive fellow laureates -- more impressive than any fellow laureates Groklaw can point to.

Groklaw shutting down in May
Posted Apr 10, 2011 20:11 UTC (Sun) by elanthis (subscriber, #6227) [Link]
I think you failed the sarcasm test, buddy. :)

I know you're responding to someone claiming you're a nobody, but just as a friendly (really) bit of advice: waving your fame-dick around in public just makes you look like you're trying to compensate for something else, and doesn't impress anybody at all.

Really, you are priceless ... but not in a good way :-p

See also Mueller's remarkable performance over at lwn.net.

... but it's worth posting here, but without the aliens.

Go on over to LWN and look at Florian's continued meltdown about how PJ isn't relevant and he is.

http://lwn.net/Articles/437650/

There's a lot said there that exposes Florian's true colors.

He heaps praise on the people who spread the most FUD about Linux. Robert Enderle, MOG, Dan "Lyin'" Lyons, and Ed Bott led the charge in the media against Linux. The only person he left out to praise was Rudy De Haas ("Paul Murphy" pseudonym). I'm sorry, but the list of above people have nothing worth listening to and his defense of them shows what side of the fence he's on.

--
BMO

His moaning and groaning on LWN is typical

The word "her" needs citation. There was an avatar named "PJ", who claimed that "PJ" meant "Pamela Jones", but there was never any verifiable track record, such as past and current employers, and "PJ" never presented "herself" in public. I just explained in another comment here that this lack of transparency wasn't reasonable.

But assuming that "PJ" is a person, I don't know what you mean by "talent". "PJ" claimed to be a paralegal, admitted not to have programming knowledge, and very apparently failed to understand the world of business. In other words, we talk about a person who apparently would have liked to become a lawyers but failed to get there, and who missed some other important perspectives that "she" would have needed to provide holistic analysis of the issues "she" covered.

What a load of crap, considering
1. not only that Mueller is not a lawyer, but has done his best to give the impression he is, and not corrected it when people fell for it
2. has always been aware that PJ is a real person - he posted his crap on SJVNs blog at zdnet, the same Steven J Vaughan-Nichols who wrote that he's met PJ.

3. when I pointed out to this piece of work on the weekend that PJ had spoken with SJVN and RMS, he didn't ask for any citation for SJVN, just RMS. He already had seen the SJVN article. He's such a bad liar.

Of course, what can you expect from someone who is trying so desperately, and failing so badly, to re-invent themselves as a mouthpiece for hire for somebody ... ANYBODY ....

He's just another amateur troll, one of those "useful fools" we occasionally hear about ...

Actually, they are following this guide: http://lwn.net/Articles/370157/

If Nokia close Qt, the community win.
If Nokia keep Qt open (but make it sucks), the community is destroyed.

WTF? Who stores passwords in a form that can feasibly be reversed?

Most code use very bad password hashing, until recently lwn.net stored passwds in clear-text, now they use bcrypt. But even if a site uses password, they will mostly use a bad hash algo.

1.0 - yes
2.6.38 - not quite, in fact he's not even in the top 20

Linux is written, mostly, by Redhat, Intel, Novell and IBM (each of these companies is at least double as productive as the most productive single developer). It is, in other words, clear proof that individuals can write free software without evil corporations. Oh wait ...

If you're just implementing a MP3 decoder (which is all Firefox would need to do to play MP3 files) then you might not have patent issues even today: http://lwn.net/Articles/166346/

It's a shame the Slashdot article didn't link to any sources.

I probably should've posted this straight away, but I would recommend two pieces of reading:
- http://lwn.net/Articles/250967/ (all parts) gives a good idea of all the sorts of effects that memory layout can have on your program's performance, and what you can do about it (to the degree that you can).
- http://www.slideshare.net/naughty_dog/multiprocessor-game-loops-lessons-from-uncharted-2-among-thieves specifically speaks about how to improve parallel processing efficiency by (amongst other things) doing what I wrote about above.

Linking > repeating :D

Exactly: Oh no, they had to COMPETE! The OHA is a cartel, they compete where they want to compete, mainly in securing the most lucrative marketing and subsidy agreements with networks, splitting the rents in the system. In the areas they don't want to compete, they use open source to crush the competition, plain and simple.

Yeah, pretty much. I have to wonder how well this complies with anti-competition laws - normal open source is safe for obvious reasons, but I have no idea if this will be. (Have been wondering this for a while. There's another older cartel called the LiMo Foundation that has the exact same business model: develop a Linux-based mobile phone OS amongst themselves using open source-like methods, but with key source code only available to members and not to end users or outsiders.)

http://lwn.net/Articles/222773/. Red Hat plays very well with others. Part of the problem is the logistics, with Git and new Kernel development you're looking at literally thousands of source code patches (which would make for a completely unwieldy SPEC file) because Red Hat back ports stuff to keep a stable Kernel in the Enterprise Linux..

Transparent Huge Pages

It doesn't matter what that is*, it's got "Buy Me!" written all over it!

OSX might have the Dock, and Windows might be up to version 7, but my Ubuntu machine has Transparent Huge Pages!


*save your breath, I actually looked it up.

Isn't this the version that 200-line patch was slated for?

I'm pretty sure that's what "automatic process grouping" is.

Yup. Some links:

• This LWN talks about the switch from TTY-based grouping to session ID-based grouping.
• Lennart Poettering's alternative solution using cgroups, which works perfectly fine as long as you don't care the changes are in user space (i.e. you have to manually set this up on each computer).
• Another alternative is using Con Kolivas' BFS, which reportedly shows similar improvements, not to mention actually pays attention to nice levels. Of course you actually have to build your own kernel, or get it from someone else, or use a distro that uses it by default like PCLinuxOS or Zenwalk.

As someone who knows bugger all about Linux, can anyone confirm if that patch will have any kind of impact on Android Devices or is it the kind of thing only a desktop user will see a difference with?

The Android kernel and the Linux kernel are pretty much irreparably forked, after the Linux people (perhaps rightly, I don't know) refused to accept the Android patches back into the trunk over the wakelock controversy. Unfortunately, the rift there never healed and there was never any real resolution.

In order for this to apply to Android, Google would have to port the changes over.

I had been willing to give them the benefit of the doubt and at least try it out before forming an opinion. Then I read the following:

User studies can be informative, but in this area, we're really interested how experienced users work with a lot of windows, so the most basic approach of paying people off the street to sit in front of of a computer for an hour to do predesigned tasks wasn't going tell us much. - GNOME Shell 2.91.90 released

Fortunately, it was well planned, not just a result of someone changing their mind while writing an email.

I'm guessing you're not a developer? What you are saying makes no sense. keeping changes in a single lump is worse for everyone: it is absolutely certain that they still have a git tree with patches in it internally, and producing the diffs would be dead simple . Talk to a RH kernel engineer, you'll find out they are very pissed off about this decision -- they just can't comment publicly.

As usual the comments on LWN.net give the full story.

Reasonable filesystem support. ext3 and ext4 are just terrible filesystem for 24x7 production systems.

Yes. I still do not have confidence in either of those filesystems.

First of all, they are fairly old designs. There is no good way to do consistent backups, checksums, or consistent incremental backup. FreeBSD, Mac, Solaris, and even Windows are way ahead here. I'm holding out hope for Btrfs, but it's not really here yet.

And if Linux is going to use an old FS design, you'd think it would at least be stable. But I have no confidence that it is. They are still working through issues like http://lwn.net/Articles/328363/ . There have been many issues that my colleagues or I have encountered with CFQ or ext3/4 ... right now, someone is dealing with a load pattern that apparently causes ext4 to freeze outright (known issue, but I don't have a link).

The Java sandbox was at the interpreter level and did not provide protection at the OS level. The google native client stuff sandboxes it at the OS level and only allows for communication via RPC calls to the parent app (e.g. drawing on a canvas), much like the seccomp approach for Linux which is a true sandbox

mounting noexec under Linux shouldn't prevent plugins from being loaded, since they are not executables but shared objects.

I thought the shared object loader respected noexec, or at least was configurable to do so. See, for example, this thread:

$ /lib/ld-linux.so.2 /tmp/ls
/tmp/ls: error while loading shared libraries: /tmp/ls: failed to map segment from shared object: Operation not permitted

Actually no; speaking as a person responsible for licensing Ubuntu for a large corporate, this guy, their COO Matt Assay started mouthing off about OpenCore. I wanted to arrange Ubuntu as a platform for our developers who mostly target RHEL but want to work well with the rest of the office. Assay's idiocy seriously affected their credibility and suggests that, unlike people like RedHat they planned to take other people's contribution OpenCore I didn't want to find out six months down the line that the developers are deliberately ignoring us (they are allowed to use whatever they want; I have to sell to them not tell them what they use). Now that he's left I'm waiting to see if there is some change in policy.

Sorry for posting AC; I know that affects credibility, but I'm not willing to put the company name up. Canonical needs to make it clear that they won't put other people's contributions into OpenCore and I'll be back like a shot. I'm not willing to have them attempting to trick their way in through another route.

Meanwhile, if anyone wants to start a Linux Distro which has a) proper back end support like RedHat has and Canonical was beginning to have and b) long term credibility that they will stay committed to F/OSS so that our developers will be happy to work with them then I'm seriously interested. I guess the best hope for this is RedHat, but there are other companies that have the capability and patents to do it (IBM; I'm looking at you). Alternatively a bunch of serious looking (yes I'm talking guys with suits) Debian consultants with a single company name, at least a 5-10 kernel hackers, 20 app developers including strong presence in OpenOffice and presence in each continent and the ability to sell support based on things that look like product licenses would also be worth considering. Think about it.

I'm afraid you are right. The difference is quite noticable. And it is too bad.

Is there a better site now?

For Linux specifically there is http://lwn.net/

I wouldn't know of a general "news for nerds" site without the slashdot / tabloid aspect.

Read http://lwn.net/ instead.

If you look into how modern distros control device permissions (e.g. on /dev/snd/pcmC0D0p ) you may find they make use of ACLs to allow lists of users access to things (without resorting to groups). However this gets complicated fast.

Additionally, a number of Linux security modules (SELinux, Apparmor, TOMOYO) alloow the use of common apps under a MAC model. Fedora really does run out of the box under SELinux after all.

However, I feel that what is needed is the ability to disclaim privileges even when running as a normal user. There is experimental user namespace work on Linux that w allow unprivileged users to create namespaces which may in the future provide such an ability.

It's always amusing to see Debian fans complaining that the end users are always going for Ubuntu instead of "hey, why not choose Debian, it's the original and it's the best!" when Debian keeps making moves like this.

What moves? This is a release announcement. Do you mean that Debian should stop releasing stable versions?

It's already bad enough to think that a new Linux user would want a browser called "IceWeasel" or would understand that it's really just Firefox renamed because of some silly branding/icon tiff with the mozilla folks.

So go complain to Mozilla.

having a bunch of useful drivers removed

There are no drivers purposefully removed, FAFAIK. Are you referring to firmware, perhaps?

a nonstandard kernel!

Nonstandard? Are you referring to the kfreebsd kernel? It is very much standard as released by FreeBSD. Or do you mean the firmware-split in the Linux kernel? That feature has been upstream for years. Or maybe you mean that it's a non-NTOS kernel? I guess you're right on that one, but most Free people would consider that a plus.

Listen, it's ok to do stuff like this if you're really into teh sooper 100% free as in freedom rms-approved purity, but don't subsequently go complaining when ordinary end users don't want it because it's unusable to anyone other than a free software hacker.

Listen, it's ok to use any Linux you like, but don't subsequently go complaining when ordinary distributors release a Free operating system that you woudn't use.

Oh, and regarding Ubuntu: please get back to me when Ubuntu releases a supported server variant that runs on my NAS.

You should be aware that Debian is not allowed to use the trademark "Firefox" and also have the ability to apply patches such as security fixes(1). It's not called "Iceweasel" out of anything but necessity. You think this is a Debian-specific issue? Well, no, it's actually a major problem for all other distributors as well(2).

1
2

So the links are 5 years old, but the issues surrounding the trademarks haven't changed or gone away. Distributions shipping "Firefox" have abrogated their ability (and responsibility) to be able to apply changes and security updates to the software without the explicit concent of Mozilla Corporation.

Not exactly free software when it comes on those terms, is it?

Regarding the kernel, I assume you're referring to the non-free firmware removal. Maybe you haven't been fully informed that the non-free firmware was actually removed from the upstream kernel sources as well. As a result, the Debian kernels are far from "non-standard", they are standard!

Regards,
Roger

[FFS Slashdot, it's 2011 and you still can't handle UTF-8!]

To be fair: how is the cross-application support for ODF? Does it really look the same in various word processors? Honest question, not trying to troll here.

Good question. See here: ODF Plugfest: Making office tools interoperable
and here: http://officeshots.org/

I usually don't reply to people this stupid. But it's a slow night.

The 'byte stream' model is not from UNIX, its just the way the hardware is laid out physically.

No, the hardware isn't laid out that way. The byte stream model is a software-implemented convenience to hide things like disk blocks and packet sizes. There's overhead associated with that, in several senses. You usually have to impose some protocol on top of the stream just to define the boundaries between items. There have been non-UNIX systems where files were record-oriented, rather than stream-oriented. UDP is transaction-oriented, although the transactions aren't reliable. QNX messaging is transaction-oriented and reliable, yet a transaction maps to one network packet if possible.

RDMA is pretty much a stable of high speed cluster computing, however its DMA that allows pretty much everything in your PC to work without slowing the processor down. Even your keyboard controller uses DMA to get the characters into somewhere useful.

Actually, DMA is part of the problem. The trouble is that most DMA is applied to real memory addresses; it doesn't pass through the MMU. This is a historical artifact of the minicomputer and PC world, which for cost reasons didn't have channel controllers like mainframes. As a result, DMA has to be managed by the OS. On IBM mainframes, the OS could give an application hardware access to a dedicated raw device, and that didn't allow the application to write outside its process, because the channel controller's memory mapping was set up by the OS.

Now that transistor counts in I/O controllers are no longer an issue, it's worth rethinking this. With the right hardware, two applications on different machines should be able to communicate safely without OS intervention.

As far as what you're calling RDMA via Infiniband, I've seen massive clusters (some of the largest in the world) using it ... safely.

Current Linux support for RDMA exists but has problems. RDMA and paging do not play well together. There's a proposal to put support for something like that into Linux, but it's really ugly. It's called "ummunotify", which is intended to notify processes when their MMU state is being changed by the kernel. This is so they can coordinate with the other machine that has RDMA access into their address space.

Personally, I think it's time to get rid of paging. Historically, paging systems at best yield the effect of having twice as much RAM, and RAM is so cheap today as a fraction of system cost that it's a nonissue. If you don't have to worry about page fault delays, performance is far more repeatable.

Yeah, who? Maybe you should have a look before asking inane questions or start spreading what looks suspiciously similar to FUD.
There's even an article for you here.

Ken's "attack" is easily broken provided you have access to an already existing C compiler without the backdoor. Of course, if you're paranoid enough you might think that perhaps the backdoor is present in every existing C compiler...

For details see this thread:

http://lwn.net/Articles/321218/

Then you'll get one kid who goes "Aww man. I totally thought this was Ms Pacman! I built it with no sense of direction whatsoever!"

Here's an interesting article you might like to read, which covers the area of sexism etc. Considering this sort of thing is topical in the community at the moment, you might like to consider whether your joke could have unintended consequences. Maybe a joke is just a joke, I'm not sure - but its food for thought. http://lwn.net/SubscriberLink/417952/306ab90f60276bf2/

If it were being broadcast on a shortwave radio band..

It was -> http://lwn.net/2002/0207/a/radio-free-linux.php3

Actually it was surprising that on the dozen or so times I saw linux.fm mentioned on IRC for the past week no one seemed to be aware of it being "old news".
Is this the span of our cultural memory?

Debian CUT == Constantly Usable Testing.

A recently started project in Debian with a similar goal of a rolling release (along with an idea of installable snapshots).

http://cut.debian.net/
http://lwn.net/Articles/406301/

For the curious, LWN covered the remote wipe capability back in September.

By that time FS-Cache (general "L2ARC" for Linux: http://lwn.net/Articles/312708/ ) will be in the kernel. And maybe btrfs will be considered stable.

Some say the b-tree variant in btrfs is smarted then the one that was used in ZFS.

I guess time will tell.

An early comment on LWN captured the technical argument best, I think, which I guess illustrates both the quality of the articles and posters on LWN. The background to this is we are discussing CPU scheduling. If you don't know what CPU scheduling is, think of it as form of mind reading. I'll illustrate.

Lets say you have asked your computer to do several things, in fact so many that if it follows the usual method of simply dividing its time equally between them it is going to annoy you. The video you watching might start flickering, or the music you are listening will drop out. So obviously the computer must now give more CPU time to playing your movie and less to whatever background task you started, such as that MP3 transcode of your 20,000 song library. Except how is the computer is supposed to know this? This is how we get to mind reading.

The hack we are discussing is essentially the discovery of a way to read the minds of one particular type of computer user - the Linux Kernel developer. The Linux Kernel developer is in the habit of starting huge background jobs called kernel compiles. These kernel compiles take a looong while, so the kernel developers, being very clever people, have invented all sorts of ways of speeding them up. One of those ways is to divide the task into lots of little bits, and then fire off separate tasks to do each. This takes maximum advantage of available CPU cores, soaking up every skerrick of available CPU time. This naturally enough leaves none left over for other important tasks like watching a movie while waiting your kernel compile. In this particular case the default CPU scheduling strategy of giving each task an equal share of CPU is woefully poor, because there might be 20 kernel compile tasks and just one movie watching task, so the movie player ends up with 1/20 of the available CPU time. This isn't enough to play a movie.

The mind reading trick discovered boils down to this: Linux Kernel developers use the linux command line interface to fire off the kernel compile. And it turns out that for years now the kernel has been able group the tasks started from a command line and give that group a single portion of CPU time, as opposed to a equal portion to each task in the group. Thus you only have to split up the CPU time into 2, one portion going to the kernel compiler group and the other going to the movie player. Naturally enough the movie player works real well with a 50% allocation of CPU, and so we have a happy kernel developer.

Now we come to the merits of the two hacks. They both do the job I just described equally well. The difference between them is that one, the kernel patch, is automagic, meaning it happens automatically without anybody having to lift a finger. But it comes at the expense of bloating linux kernel a tiny bit, even for users who won't benefit from it. The other way currently has to be done applied manually using a process the vast majority of Linux users will at best find difficult, tedious and error prone.

Seems like a simple decision eh - lets take the tiny bloat hit and not inflict our long suffering desktop users with yet another Linux user-unfriendly idiosyncrasy. But here is the rub: it doesn't help them. In fact, for some it might have a negative impact (a gstreamer pipeline started from the command line strings to mind). The people who will benefit from this are the ones that use the command line heavily and regularly. People like Linus. Which is why he liked it so much I guess. But these are precisely the people who will have no absolutely no trouble doing it the manual way.

Thanks for the nice words about LWN! Here's a special link to the LWN article on per-tty group scheduling for Slashdot folks. Hopefully a few of you will like what you see and decide to subscribe.

You've got a per-tty mouth.

There is a patch for linux to do this called bcache. See LWN article about it.

Thanks for the nice words about LWN! Here's a special link to the LWN article on per-tty group scheduling for Slashdot folks. Hopefully a few of you will like what you see and decide to subscribe.

They are not the same. The kernel patch groups processes by owning TTY. The bash shell change groups them by session. Source: http://lwn.net/Articles/414817/

The kernel has a mechanism to schedule groups of processes, and it has for years. By grouping tasks together, you can make one process (video playing) get the same cpu share as a group of processes put together (compiling code). By doing this (instead of the video processing being equal to just one of the compiling processes), everything feels more interactive, even though it's actually slightly slower.

No one uses scheduling groups because they have to be setup by root and it's not the easiest thing in the world (you have to write stuff into sysfs, I think). No distributions set them up.

The magic kernel patch just adds a simple rule to the scheduler. When a process starts, it goes into a group with the rest of the processes in that TTY (virtual terminal). This means the user doesn't have to do anything and the groups are setup automatically.

Poettering thinks this is somewhat hackish, and that things shouldn't be based on what TTY a process is started on. He made the little script to prove that this can easily be done in userspace.

Linus has rejected this, basically saying that we've had years for people to make something like this and no one did until the kernel patch came along. The patch is simple, reasonable, and doesn't require distributors to ship updated userland files to put processes in groups.

I should note that my understanding comes from LWN, which has had excellent coverage of this on their kernel page, as always. You'll be able to see their articles in two weeks if you're not a member (which is worth it if you like this kind of stuff).