Page 1 of 2 12 LastLast
Results 1 to 10 of 17

Thread: Row brewing over Linux patches

  1. #1
    Mentor
    Join Date
    Jun 2001
    Posts
    1,672

    Row brewing over Linux patches

    A proposal to help Linus Torvalds keep up with patches for Linux has sparked a controversy over whether the operating system has outgrown its creator.
    http://zdnet.com.com/2100-1104-826165.html

  2. #2

    Re: Row brewing over Linux patches

    Wait a second...I read some of those comments and I'm not so sure I agree with what they're saying.

    The patches that are being rejected are probably being rejected for good reasons (the code is ugly, it breaks a lot of other things, it's too big, it's hard to read, Linus can't figure out what parts of it do, Linus doesn't think that a particular patch would be beneficial to kernel development, etc., etc.). I have a feeling that most of the people complaining are just pissed because their patches were rejected and they weren't given reasons why.

  3. #3
    Guest

    Re: Row brewing over Linux patches

    One maintainer above them all
    One maintainer to rule them all
    and in the darkness bind them !

    I like Linus Torvalds as a developer, brilliant hacker - but he REALLY sucks at maintaining a stable kernel. It would be much better to have him direct the unstable branch and then as soon as it reaches maturity hand it over to another guy, then go on with the unstable dangerous stuff, after a much needed vacation break.

    Having an eager to innovate kernel hacker maintain something stable... bad - the urge to include new stuff is to great.

    I would rather that we pop out major kernel versions more often and then only fix bugs, security and speed issues in the stable series.... NEVER including new stuff. Then have a development cycle of say 6 months for the major version kernels. This might give problems in relations to new hardware, but to hell with that - we're not producing an M$ OS here !

  4. #4

    Re: Row brewing over Linux patches


    One maintainer above them all
    One maintainer to rule them all
    and in the darkness bind them !

    I like Linus Torvalds as a developer, brilliant hacker - but he REALLY sucks at maintaining a stable kernel. It would be much better to have him direct the unstable branch and then as soon as it reaches maturity hand it over to another guy, then go on with the unstable dangerous stuff, after a much needed vacation break.

    Having an eager to innovate kernel hacker maintain something stable... bad - the urge to include new stuff is to great.

    I would rather that we pop out major kernel versions more often and then only fix bugs, security and speed issues in the stable series.... NEVER including new stuff. Then have a development cycle of say 6 months for the major version kernels. This might give problems in relations to new hardware, but to hell with that - we're not producing an M$ OS here !
    Hear Hear!!

  5. #5

    Re: Row brewing over Linux patches

    You guys should read the kernel mailing list or some of slashdot's responses to this. *

    http://slashdot.org/article.pl?sid=0...mp;mode=thread

    This was one particular comment that I like that did well to summarize Linus proposal on how to handle the kernel's code.

    This is another comment that I particularly liked that summarized Linus' proposal for code management.

  6. #6

    Re: Row brewing over Linux patches

    OK, well, in one of those articles, it's suggested that Linus have people under him to monitor the patches (clean them up, reject them, whatever), and possibly have those people have their own people underneath them to help out with things (and so on and so forth).

    My question about this whole thing is this: how do we know if these people will be up to the task? What if all this additional management ends up bogging things down, slowing progress to a crawl, or suppose that because a team leader just doesn't like a code contributor, who just happens to have an essential patch for the kernel, he rejects all of that guy's code, well, just because? How will that be any better than Linus' current system? All I've seen are a number of hypothetical situations and suggestions...I don't see all that many people volunteering to do these essential and recommended jobs.

  7. #7
    Guest

    Re: Row brewing over Linux patches

    well i've been following that discussion on the lkml since the first post about it, and i must say i agree with Linus on this.

    The biggest problem is that everybody just sends their patches directly to Linus, instead of sending them to the maintainers of the subsystem their patch is for. This is why so many patches get dropped and take so long to make it into the kernel.

    If people would just send their patches to the right person there wouldn't be so many problems with kernel development

  8. #8
    Moderator
    Good Guru
    Compunuts's Avatar
    Join Date
    May 2001
    Location
    California
    Posts
    3,935

    Re: Row brewing over Linux patches


    It would be much better to have him direct the unstable branch and then as soon as it reaches maturity hand it over to another guy, then go on with the unstable dangerous stuff, after a much needed vacation break.
    If I'm not mistaken, THIS is the current set up. Linus no longer maintain a kernel once it reaches stable status and hand it over to Alan Cox for it. Cox is the chief maintainer for most of the stable kernels except that there needs a new release of stable kernel.

  9. #9

    Re: Row brewing over Linux patches

    One of the things that has kinda irked me about kernel development is the naming scheme: the 2.4 kernel series are called 'Stable', yet they haven't really been all that stable and usable up until 2.4.16 was released, and now people are having problems with 2.4.17. Why can't they work on the 2.odd-number kernels until they have something suitable and stable enough for end users to use and then spend a year or so making that (the 2.even-numbered kernels) even more stable?

    I wouldn't mind having one Stable branch with kernels that are safe to use on just about any of my linux installs, and a Testing/Unstable branch with experimental stuff in it. So what if it takes them a year or so to put new stuff into the kernels! I just want a stable kernel that I can use right now!

    Btw, I noticed while going through the lk mailing list that there was a status report for 2.5.3xxx and it mentioned that ReiserFS and XFS support might be placed into the kernel at some point; I think it'd kick butt if the next (after Woody) release of Debian would allow you to choose between Reiser, ext3, and XFS when you installed it. (mmmm...XFS...*drool*)

    http://www.atnf.csiro.au/~rgooch/lin...newsflash.html

  10. #10
    Moderator
    Good Guru
    Compunuts's Avatar
    Join Date
    May 2001
    Location
    California
    Posts
    3,935

    Re: Row brewing over Linux patches

    Correction ( or addition really ) to my own post above is that Alan Cox maintains 2.2.x series and Dave Jones maintain 2.4.x series since there are needs for 2.2.x kernel to update such as security.


    the 2.4 kernel series are called 'Stable', yet they haven't really been all that stable and usable up until 2.4.16 was released, and now people are having problems with 2.4.17.
    The issue is really sticky. The 2.4.x series is STABLE ( at least on my machines and many others ). The majority of problems are occured on some weird and hardwares that do not have standardization on hardware vendors' part such as Nvidia not releasing info on their hardware so that developers have no way to test every single version of the hardware Nvidia produced. It is not only that prblem but just an example. Plus, Linux developers do not have all kinds of hardware at their disposal to test it around. If we wait too long, then the technology will be quickly outdated and far behind. And that's why many Linux distro ( such as Red Hat ) do not include the latest kernel with their latest distro release. I believe that it's the judgement call on the part of users. You don't really deploy the latest Microsoft's Windows just the day after they release it in your production environment, do you??

    Why can't they work on the 2.odd-number kernels until they have something suitable and stable enough for end users to use and then spend a year or so making that (the 2.even-numbered kernels) even more stable?
    The current discussion is exactly from that kind of thing. People think that Linus can't get the patches for their hardware fast enough in Kernel's releases.

    I wouldn't mind having one Stable branch with kernels that are safe to use on just about any of my linux installs, and a Testing/Unstable branch with experimental stuff in it.
    Welcome to Debian world. You seems to be using Debian so I'm not surprised that you have that analogy. But it has its own quarks as well. People tend to think that Debian's latest stable release with Kernel 2.2.19 is too out of date although it's rock solid on my machine and I've not seen any advantage for my machine to take advantage of from newer kernels.

    there was a status report for 2.5.3xxx and it mentioned that ReiserFS and XFS support might be placed into the kernel at some point
    I think it would be great if he did. It is long overdue since Journaling file systems is a must for now a days use. But then again, he needs to support all of them and not pick and choose one or two of them.

    I think it'd kick butt if the next (after Woody) release of Debian would allow you to choose between Reiser, ext3, and XFS when you installed it. (mmmm...XFS...*drool*)
    According to this, they have already have that. According to them, you can choose your flavor of choice. The first disk is for Vanilla, second is for idepci, third is for compact, fourth is for Reiserfs and fifth is for udma100-ext3. The rest three disks are for programs.

Similar Threads

  1. Critical Patches
    By bruce.ringlerII in forum Windows - General Topics
    Replies: 2
    Last Post: 07-10-2006, 08:09 PM
  2. Wicked Storm Brewing
    By Fatal Error in forum General Chat
    Replies: 18
    Last Post: 09-10-2005, 08:26 PM
  3. Linux Kernel Security Patches
    By shebang in forum Linux - Software, Applications & Programming
    Replies: 3
    Last Post: 09-08-2004, 12:21 AM
  4. Kernel Patches
    By stryder144 in forum Linux - General Topics
    Replies: 2
    Last Post: 09-30-2002, 06:56 PM
  5. Kernel Patches..
    By gmoreno in forum Linux - General Topics
    Replies: 2
    Last Post: 05-28-2002, 04:32 PM

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •