Torvalds rebukes desktop critics

Specialised desktop development out of the question.

Linus Torvalds, creator and maintainer of the Linux operating system kernel, has reacted angrily to suggestions that the kernel's development process is skewed in a way that prevents improvements on the desktop.

Torvalds was responding to criticism by programmer Con Kolivas, who had developed a patch designed to improve the performance of specific Linux desktop features.

Kolivas claimed in an interview that the kernel development process privileges enterprise-centric server features, while basic desktop features remain almost unusable.

In emails to Linux developers over the weekend, Torvalds argued that kernel developers are well aware of desktop performance issues because they generally use the platform on the desktop themselves.

"I've always considered the desktop to be the most important part. And I suspect that that actually is true for most kernel developers, because quite frankly, that's what 99 percent of them end up using," he wrote. "The whole argument about how kernel developers think that the desktop isn't important is totally made-up crap."

The real issue is that a highly specialised patch such as Kolivas' is doomed to failure because it ignores the bigger picture, Torvalds said.

"Most kernel developers realise that Linux is used in different places, on different machines, and with different loads," he wrote. "You cannot make everybody happy, but you can try to do as good a job as possible. And doing 'as good a job as possible' very much includes not focusing on any particular load."

Torvalds recently moved to address some desktop performance problems by replacing the scheduler component of the Linux kernel, but chose to use the CFS scheduler rather than Kolvias' SD scheduler.

He said the decision is not set in stone.

"It's not like we've come to the end of the road: the baseline has just improved," he wrote.


What are your views on this subject? Use the form below to post a comment on this article up to 500 characters.


Characters remaining: 500

Add your commentComments

zajelo3 | Published: 02:47 GMT, 16 October 2007

If you read how everything happened on the LKML it seems that Linus' main concern was using the patchset he knew would be maintained properly by a stable maintainer. Kolivas kind of proved Linus right by quitting. His patches will be sorely missed. Him quitting didn't hurt Linus I'm sure, but it sure will hurt all of us desktop users in the long run.

dgray | Published: 00:56 GMT, 03 September 2007

I believe that Linus does understand Linux leadership, especially in regards to the GPLv3. The biggest problem in Linux desktop adoption is the GUI programmers, (KDE, Gnome, fluxbox, compiz-fusion, xfce, etc..) to be honest how different are their common goals? - not that fundamentally different. The linux desktop would be alot more complete if they weren't all busy reinventing the wheel. As far a cons patchet I loved it, I wish more kernel devs paid as much attention to the desktop as he did.

MattD | Published: 20:16 GMT, 13 August 2007

I have to agree with Billy, though his comments are a bit harsh. There are some great things happening in the Linux world. Dell, Acer and Lenovo selling Linux is one of the major Linux stories this year. What Linux needs now are leaders looking to take Linux over the next hurdle.... adoption. Torvalds is not that man. He really should keep his mouth shut. He should stop playing the role of a petulant teenager quick to blast other people in the FOSS world. He's attacks on the GPL 3, FOSS leaders, alternative kernels mimic those of trolls. His 'no comments' on competitors who are always looking for ways to destroy Linux makes me question his motivations. He does a soft-dance when Microsoft accuses him and his team of thievery, piracy and other illegal-patent infringing acts but is far too quick to attack those who are genuinely trying to move Linux forward. With leadership comes responsibility which is something the self-centered Torvalds does not understand.

snakeeyes | Published: 10:58 GMT, 04 August 2007

hey, billy, do know what your talking about?who do want to replace mr. linus? you?what do you know about kernel?haaaaa?i think all you know is to insult the person and to be famous with that attitude,it's that right?

Billy | Published: 04:11 GMT, 02 August 2007

Linus Torvalds is an idiot. His mis-management of the operating system from day one is pathetic. By definition, an OS requires management and rules for all the sub-layers. Torvalds should be ignored, along with his lame comments towards his critics. He's nothing but a kernel programmer, hiding in his cubicle. When Linux needed leadership, he went into hiding. Take Operating Systems 101 for the facts.

Wayne | Published: 21:01 GMT, 01 August 2007

The egos go both ways and stifle advancement for the desktop crowd. Sure, the money is in the server section but, people want decent desktop performance too. It's no where near as good as it is with Windows. A lot of us was hoping that by now, Linux would be a force to be reckoned with against Windows but, the egos on both sides and the greed has pushed it away from the desktop. It competes in the server section but what about the millions of desktop people out there? Con is the only one so far that seemed to be on our side and tried to help and he got blasted and people in charge thumbed their noses at him and us. Con has a lot of valid points. Because those points differed from the direction the main developers seem to be taking Linux, he got blasted. They are too concerned with the server side that the desktop side is suffering. At least Con tried to help there. More than can be said for the others.

Wayne | Published: 21:01 GMT, 01 August 2007

If Con's patches are questionable, then why haven't the other main developers not tried to take his code and make it work to where it would be right for both sides? Greed and egos and the lack of concern for the desktop people. That's why. The desktop side isn't important to them and they don't like someone trying to help.

Mallik | Published: 16:28 GMT, 01 August 2007

Con Kolivas is known for high performance patches. But the problem is that it leaves some part of the job to risk. Con's patches are good for 95% of the cases. BUUUUUT ..He leaves 5% to chance. That is risk. Introducing such a thing into something as important as the kernel would simply mean disaters. Also Remember...CFS is actually SD concept with some differenes. Con is a very good developer... The kernel people do not want to leave anything to chance in the kernel. They want the kernel to act in a more predictable way and not leave anyhting to chance.

TDDPirate | Published: 12:04 GMT, 01 August 2007

I do not understand why is it impossible to have multiple scheduler modules in the kernel. The user can choose the desired active scheduler module at boot up time or even replace it at runtime. When you want to play games, use one scheduler. When you need to edit files, use another scheduler. When you leave a long job with heavy I/O to run during the night, have the PC switch to a third scheduler. Do we need to switch to Hurd for this kind of functionality?

Frank | Published: 07:24 GMT, 01 August 2007

Benchmarks that I've seen (I think from the CFS author, Ingo Molnar) show that CFS generally does a better job than SD, although I don't think the SD tested against had CKs patch(es) in it. In any case, the CFS scheduler has numbers that make it look *very* good.

Related Operating Systems news

Microsoft, Linux rivals mock Google Chrome OS

Operating system crippled by reliance on web access

Google releases Chrome OS to waiting world

Stresses speed and security of operating system.

NSA helped with work on Windows 7 security

Privacy organisations concerned about spooks' involvement.

Microsoft previews Office 2010 beta to developers

But pirated copies of the software are already leaked online

Related Operating Systems reviews

Jolicloud OS

Moblin 2.0 review

Ubuntu Netbook Remix



Email this article to a friend or colleague:


PLEASE NOTE: Your name is used only to let the recipient know who sent the story, and in case of transmission error. Both your name and the recipient's name and address will not be used for any other purpose.

Techworld White Papers

Database security: Preventing enterprise data leaks at the source

IDC discusses the growing internal threats to business information, the impact of government regulations on the protection of data, and how enterprises must adopt database security best practices...

Download Whitepaper

Service-oriented security

SOA has become an integral part of enterprise software by providing a framework to efficiently develop software as services that is easily sharable, reusable, and integrated. No where is the need more apparent than in the Identity Management space. Welcome to the age of Service-Oriented Security (SOS).

Download Whitepaper

Data protection prospective vendor checklist

Organisations need a way to map business needs against all these challenges in procuring a technical solution. To help, SANS has developed the following Prospective Vendor Checklist.

Download Whitepaper

Unlock the power of the mainframe

This whitepaper presents the notion of CICS as an integration hub based on a component-based, service-oriented architecture supporting Web services. Highlights will review the challenges and contrasted support for Web services natively in CICS.

Download Whitepaper

Techworld UK - Technology - Business

COLT White Paper

Are all VoIP services the same?

Questions to ask your service provider to ensure you get the VoIP service you need
With careful choice of partner, your business can have all the advantages of VoIP access - reduced costs, flexibility and simplicity - without the drawbacks.
This white paper is your guide to ensure you get right the VoIP service and details the pitfalls which businesses would do well to avoid.

Download white paper
BMC

Ride the express lane in the journey to speed ITIL adoption

Explore the challenges in making the journey to ITIL and the criteria for selecting consulting services
By following ITIL practices, your IT organisation will become more closely integrated with the business. We recommend making the journey to ITIL in a sequence of six incremental steps, the phases of which are driven through execution of a strategic transformational roadmap.

Download white paper

Webcast: IT Financial Management: Cost Optimisation for Efficiency and Agility.
On Demand Webcast
Join this webcast to learn about the techniques and technologies that can help you prove the value of IT to the business by understanding the true cost of today's IT services and those that will be necessary to deliver future success.

Register Today

Site Map

IDG Network

* *