'Be careful' warning accompanies latest Linux kernel The next Linux kernel has been released with a tidy little warning from Linus Torvalds for code committers to pay more attention and be more careful. Linux Kernel 2.6.35 has been kicked out after Torvalds said he could see very little reason to offer any more release candidates. Features include changes to networking with …
People developing patches intended for mainstream inclusion will always want to get them into the next stable release because that results in more testing, more feedback and more developers helping to maintain the changes. Keeping a set of changes up to date for several months or a couple of years against such a rapidly changing target is very hard work.
The solution is probably for Linus to delay work on the next mainstream release for a cycle of say 2-3 months, so that more work can take place on getting the contents of Linux-next into acceptable shape, either by improving what stays in there or by kicking out what shouldn't be in there.
Sounds like they've finally hit the point where they have to exercise some proper control over what is introduced into the kernel. Shame, because it will take some of the life out of it, but necessary if there is this sort of concern.
From my own experience the latest kernels have broken the chain of success on my netbook - wireless is now unreliable and no-one seems to want to fix it. I have to go back a number of kernel versions or use a special to have reliable wireless.
I'm not complaining too much, mind - it is still free and effective for what I want to do.
@JustanKOS:
You obviously do not know what the term 'linux-next' refers to.
trying to figure out how to balance between developer enthusiasm and stability has been the main point of Linus's job for most of his professional life. All he does is review code. That's pretty much it. This is why 'git' was created, this is why linux-next was created, etc etc etc.
This has been going on for decades now.... and you have not noticed any of it?
I've been aware like most people of the efforts going on to develop Linux but not the detail. I have to say that I have always been impressed by the efforts of all involved and, despite my gripe about my wireless problems, by the quality and stability of the end-product.
My comment was just an observation based on my experience and my presumption that the comments the article referred to were uncommon. If Linus is always saying such things then ignore me - it's a non-story and all will be well as long as people take heed of him.
New kernel out, Linus makes a totally unrelated statement about the development branch. However, that doesn't stop us from using a misleading title to get views and - even better - some redundant flamewars over open source and operating systems in the discussion forum.
Interesting URL for the article, BTW.
Speaking as a programmer, I would agree that making a kernel modular as with Minix or Hurd should make it easier to develop, as the better understood messaging between autonomous kernel processes should make the whole program more manageable. However, making these communications occur through shared memory data structures in a monolithic program rather than sockets between multiple autonomous processes does seem to offer much higher performance. The history of Linux v. Hurd seems to bear this out.
One of Hurd's problems is getting enough momentum so that enough people want to work on it. If it continues to perform poorly in practice compared to Linux the Hurd is likely to remain a research operating system used by a few hundred computer scientists as opposed to one which is used on millions of developer desktops, on value for money servers and on most embedded systems that can do Internet Protocol.
Minix isn't intended as competition for Linux, other than in a narrow area of computing science education where it is useful for a student to be able to work with a system small enough to be fully understandable.
Gavin, kudos to you and the Reg for putting out a summary of the current state of affairs in Linux kernel development. It's too arcane for this non-kernel developer to wade through the mailing lists or try to glean information from friends in the know. Great summary, hope you keep these coming.
Google files a lot of patents for a lot of different things, including packet routing. Even though I couldn't identify at this any particular patent potentially covering their contribution, I certainly wouldn't exclude this as one may still be in the pipeline awaiting publication.
What's the Linux position regarding the IP for contributed code?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
