Announcement

Collapse
No announcement yet.

Linux 3.0 Kernel May Remove Some Old Cruft

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Linux 3.0 Kernel May Remove Some Old Cruft

    Phoronix: Linux 3.0 Kernel May Remove Some Old Cruft

    The discussion surrounding Linus Torvalds' proposal to end the Linux 2.6 kernel series and continue on as the Linux 3.0 kernel has continued on since it began less than 24 hours ago. The reaction has largely been positive and supportive of this proposed change. Of the few objections, some see no reason to mess around with the versioning, but now there may be a reason for this change: to drop the old cruft that's been living in the kernel...

    http://www.phoronix.com/vr.php?view=OTQ3OQ

  • #2
    Michael, any chance we can get some links to the sources without having to fumble through several other Phoronix posts to get to them? Cheers.

    P.S. It would be amazing if external links were a different colour, like blue or something!

    Comment


    • #3
      Originally posted by Kazade View Post
      Michael, any chance we can get some links to the sources without having to fumble through several other Phoronix posts to get to them? Cheers.

      P.S. It would be amazing if external links were a different colour, like blue or something!
      Agree completely. The Phoronix page-juicing, link maze is horrible.

      Comment


      • #4
        Aha, the source

        Alan Cox's email on LKML here: https://lkml.org/lkml/2011/5/24/320

        Comment


        • #5
          Originally posted by Kazade View Post
          Michael, any chance we can get some links to the sources without having to fumble through several other Phoronix posts to get to them? Cheers.

          P.S. It would be amazing if external links were a different colour, like blue or something!
          I completely agree.
          ## VGA ##
          AMD: X1950XTX, HD3870, HD5870
          Intel: GMA45, HD3000 (Core i5 2500K)

          Comment


          • #6
            Originally posted by darkbasic View Post
            I completely agree.
            What's that stuff Ralph Baechle is saying about hardware refusing to die? Let that old cruft/zombie undead junk run on the "old" 2.6 kernel. No need to propagate that junk until the end of time.

            Another interesting perspective is that Linux can lead hardware changes. Especially when it comes to servers -- if Linux won't support crufty old junk, then hardware makers won't keep including that crufty old junk. At least not in server hardware.

            Comment


            • #7
              It sure does sound like a great idea to move really old drivers to an "obsolete" area where they'll at some point be completely removed from the tree down the road as most really old systems that use such drivers like ISA, EISA or MCA for example are not quite as prevalent nowadays. "Oddball" drivers that only support one or two odd devices that aren't in much use any longer should be also be considered for removal. With such cleaning then the tree'll be leaner and quicker to build not to mention making it easier for maintenance.

              Linux 3.0 should be a new start so perhaps this is the time for kernel devs/driver maintainers to look into doing an overhaul of the kernel tree and removing the cruft that has built up over 20 years of Linux kernel development.

              Originally posted by droidhacker View Post
              What's that stuff Ralph Baechle is saying about hardware refusing to die? Let that old cruft/zombie undead junk run on the "old" 2.6 kernel. No need to propagate that junk until the end of time.

              Another interesting perspective is that Linux can lead hardware changes. Especially when it comes to servers -- if Linux won't support crufty old junk, then hardware makers won't keep including that crufty old junk. At least not in server hardware.
              Ditto...like Windows has dropped support for some old hardware its time for Linux to follow suit

              Comment


              • #8
                I don't agree with his reasoning. 3.0 shouldn't have any special features just because it's 3.0, if they did that it'd turn out to be like 2.5.

                If he wants to drop the old drivers, he should wait until 3.1 instead of rushing it through the merge window for 3.0

                Comment


                • #9
                  Originally posted by pvtcupcakes View Post
                  I don't agree with his reasoning. 3.0 shouldn't have any special features just because it's 3.0, if they did that it'd turn out to be like 2.5.

                  If he wants to drop the old drivers, he should wait until 3.1 instead of rushing it through the merge window for 3.0
                  Did you read the proposal? https://lkml.org/lkml/2011/5/24/338
                  -- 2.8 to mark the beginning of the housecleaning.
                  -- 3.0 to mark the completion.

                  2.8 would be a short lived transitional kernel.

                  Comment


                  • #10
                    Originally posted by droidhacker View Post
                    Did you read the proposal? https://lkml.org/lkml/2011/5/24/338
                    -- 2.8 to mark the beginning of the housecleaning.
                    -- 3.0 to mark the completion.

                    2.8 would be a short lived transitional kernel.
                    Either way, version numbers shouldn't be about features. This is what Linus has been saying since 2.6 was released. I don't think there should be a goal for the kernel to reach 3.0. If Linus wants to call 2.6.40 version 3.0 then that's good enough. 3.0 doesn't have to be special.

                    Comment


                    • #11
                      Originally posted by pvtcupcakes View Post
                      Either way, version numbers shouldn't be about features. This is what Linus has been saying since 2.6 was released. I don't think there should be a goal for the kernel to reach 3.0. If Linus wants to call 2.6.40 version 3.0 then that's good enough. 3.0 doesn't have to be special.
                      Its not about features, its about the evolution of features. In a kernel series, you're pretty much doing add-add-add-add, so the new features keep piling on. You can generally expect forward compatibility. This would mark a new kernel that is not 100% forward compatible from 2.6. Go with the big number jump to denote this fact.

                      Similarly, you can't always expect any 2.6 kernel to be a drop in replacement for a 2.4, 2.2, or 1.x. You *can* expect a 2.6.y kernel to be virtually a drop in replacement for 2.6.x where y>x though. See what I'm getting at?

                      Comment


                      • #12
                        I've always been frustrated that I need stuff from drivers/scsi to get my sata hard drives to work.

                        Also, as far as I know, a lot of hardware, like sensors, still uses ISA.

                        Comment


                        • #13
                          ISA still useed there too, don't remove!

                          Comment


                          • #14
                            Originally posted by pvtcupcakes View Post
                            Either way, version numbers shouldn't be about features. This is what Linus has been saying since 2.6 was released. I don't think there should be a goal for the kernel to reach 3.0. If Linus wants to call 2.6.40 version 3.0 then that's good enough. 3.0 doesn't have to be special.
                            Who Cares? I really don't get why people are so wrapped up in version numbers these days...

                            Comment


                            • #15
                              Originally posted by roland View Post
                              I've always been frustrated that I need stuff from drivers/scsi to get my sata hard drives to work.

                              Also, as far as I know, a lot of hardware, like sensors, still uses ISA.
                              That's because of the driver model that makes IDE/SATA drives appear like SCSI devices

                              Comment

                              Working...
                              X