Announcement

Collapse
No announcement yet.

New NTFS Linux Driver Spun A Ninth Time, Still Under Review

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

  • #31
    Originally posted by skeevy420 View Post
    Also, see Android. That's literally proprietary Linux.
    Its always funny to read this. Android, with its kernel, tools, framework, toolkit, Java Runtime Env and all bells and whistles is fully Open Source, should even be FLOSS (Apache License 2). Just the Google Apps manufacturers put on top of it aren't. But all the Userland Apps are fully open.

    On the other hand we celebrate Jolla with their Sailfish OS, a standard RPM distro with a closed source UI, closed source userland and based on a toolkit with a company behind that makes a lot of money from selling code that was GPL & community contributed under a greedy proprietary license. Android looks a lot better in comparison.

    Comment


    • #32
      Originally posted by ed31337 View Post
      Pure speculation: Perhaps Microsoft is secretly working towards rebasing Windows to run off of Linux?
      No.

      Just No.

      There would be no sane reason to do this.

      If anything their interest would be Linux server and azure

      Comment


      • #33
        Originally posted by oiaohm View Post
        That is a bit of a pushing it. Windows 10, Windows 8.1, Windows 7, Windows Vista, Windows XP and Windows 2000 all have one thing in common some screw up in the NTFS driver resulting in user files magically pulling the disappearing act that had to be patched. The idea that the Windows ntfs driver is reliable and tested well is in the maybe camp. So it is possible that a driver based off full complete NTFS specifications may end more more reliable.
        I know large organizations which have been running Windows since Windows NT 3.51 and haven't lost a single file on their hundreds of PCs and laptops. The Windows NTFS driver is reliable and tested, period. Unlike Linux, Microsoft extensively tests it and has a very stringent QA/QC policy at least for data retention (they do have regressions and bugs in less critical areas of Windows).

        I'm not sure specs are enough to create a fully functional implementation because I've read a lot of horror stories about Paragon products when people completely lost their data. Actually I once lost an entire drive full of data when I tried using Paragon Hard Disk Manager. Luckily there was a backup.

        Comment


        • #34
          Originally posted by Alexmitter View Post

          Its always funny to read this. Android, with its kernel, tools, framework, toolkit, Java Runtime Env and all bells and whistles is fully Open Source, should even be FLOSS (Apache License 2). Just the Google Apps manufacturers put on top of it aren't. But all the Userland Apps are fully open.

          On the other hand we celebrate Jolla with their Sailfish OS, a standard RPM distro with a closed source UI, closed source userland and based on a toolkit with a company behind that makes a lot of money from selling code that was GPL & community contributed under a greedy proprietary license. Android looks a lot better in comparison.
          Open source components on a closed platform is a real grey area. All of the AOSP is open source...but a lot of the stuff that makes Android worth using isn't. Moreover, unlocking ones bootloader to modify the open source components puts phones in a state where the closed source parts will quit working, like Android Pay.

          On the other hand we have people still trying to say that Qt is closed source. I guess that's the point you're trying to make due to all the Qt fud on Phoronix.

          FWIW, Sailfish isn't much more than a Qt based Android Rom that, like Android, has lots of open source components with closed source components sprinkled about. IMHO, not much of a difference between the two other than the UI tookit being used. They're no more open or closed than the other.

          Comment


          • #35
            Originally posted by birdie View Post

            I know large organizations which have been running Windows since Windows NT 3.51 and haven't lost a single file on their hundreds of PCs and laptops. The Windows NTFS driver is reliable and tested, period.
            First of all, there is not one driver but multiple ones as there are multiple versions of NTFS.
            Second of all, a tried fs with exponentiell file creation times is a joke.
            Third, there are a billion ways to break NTFS. I cannot count the times I endes up with a Undeleteable file because of some NTFS quirck.
            Last but not least, I had to store years of pictures on a NTFS raid yor my parents. For shits and giggles, I checked the files after 1.5y and out of the 170k files, at least 5 had been partially truncated. I have yet to see something like that happening on ext3, ext4, or btrfs.

            Comment


            • #36
              Originally posted by Alexmitter View Post
              Its always funny to read this. Android, with its kernel, tools, framework, toolkit, Java Runtime Env and all bells and whistles is fully Open Source, should even be FLOSS (Apache License 2). Just the Google Apps manufacturers put on top of it aren't. But all the Userland Apps are fully open.

              On the other hand we celebrate Jolla with their Sailfish OS, a standard RPM distro with a closed source UI, closed source userland and based on a toolkit with a company behind that makes a lot of money from selling code that was GPL & community contributed under a greedy proprietary license. Android looks a lot better in comparison.
              But then there's the matter of redistribution, which is key, but apparently easily overlooked.
              Vendors like Samsung and the like personalize the open source AOSP base, apps included, and redistribute all of it as proprietary, because they can, as the license is permissive, not libre.
              Any Android smartphone, in its factory conditions, practically runs completely on proprietary software, excluding the skeleton of a dead Linux kernel version, which then is bastardized with proprietary drivers muscles on top, making it non-updatable.
              Such is the world of ARM.
              Android is only "FLOSS" if you install something like LineageOS on a supported smartphone, and even then, only if you are referring to the userspace, as the Linux skeleton will still be relying on proprietary drivers, some of which might be reverse-engineered in the year twothousand-never.

              Does this even make sense in this thread?
              Even if MS decided to replace NT with Linux, x86 devices running Windows/Linux still wouldn't be as closed as Android smartphones.

              Comment


              • #37
                Originally posted by pininety View Post
                First of all, there is not one driver but multiple ones as there are multiple versions of NTFS.
                Second of all, a tried fs with exponentiell file creation times is a joke.
                Third, there are a billion ways to break NTFS. I cannot count the times I endes up with a Undeleteable file because of some NTFS quirck.
                Last but not least, I had to store years of pictures on a NTFS raid yor my parents. For shits and giggles, I checked the files after 1.5y and out of the 170k files, at least 5 had been partially truncated. I have yet to see something like that happening on ext3, ext4, or btrfs.
                I'm aware of different NTFS drivers but they all have been reliable unless you're talking about outages, broken hardware or people pressing the reset button. I wonder why you're continuing to pit your sample size of two against the sample size of multiple organizations and individuals in the world. Windows is used by governments, military and overall has over a billion installations in the world. If NTFS had been as unreliable as you portray it to be, people would have long abandoned the OS altogether.

                And speaking of personal experience: I've yet to lose a single file or have any FS breakage in my 25 years of using NTFS (starting with NT 4.0). I can say the same about over two dozen people that I know personally, none of whom has lost any files yet.

                I dare you to tell me a single method of breaking NTFS by using it normally vs "a billion ways to break NTFS".
                Last edited by birdie; 20 October 2020, 10:39 AM.

                Comment


                • #38
                  Originally posted by birdie View Post
                  I'm aware of different NTFS drivers but they all have been reliable unless you're talking about outages, broken hardware or people pressing the reset button. I wonder why you're continuing to pit your sample size of two against the sample size of multiple organizations and individuals in the world. Windows is used by governments, military and overall has over a billion installations in the world. If NTFS had been as unreliable as you portray it to be, people would have long abandoned the OS altogether.
                  This is bullshit. In most cases if a file is broken, people will just look for a copy of the file or assume they messed up a save. That is true for all fs, most noticable in lustre fs where we see broken untouched files on our cluster regularly yet it is still used as there are not much alternatives. People use NTFS because they need to run Windows. Thats it.

                  Originally posted by birdie View Post
                  And speaking of personal experience: I've yet to lose a single file or have any FS breakage in my 25 years of using NTFS (starting with NT 4.0). I can say the same about over two dozen people that I know personally, none of whom has lost any files yet.
                  And how have you checked that? You know how I noticed that the these files are getting corrupt? Because my mum complained about a truncated picture being displayed.
                  Luckily, I had the sha512 hashes of all files and could compare and guess what, multiple other files also turned up with different hashes. Just by using the data, you would have never noticed but with sheer luck.


                  Comment


                  • #39
                    Originally posted by pininety View Post

                    This is bullshit. In most cases if a file is broken, people will just look for a copy of the file or assume they messed up a save. That is true for all fs, most noticable in lustre fs where we see broken untouched files on our cluster regularly yet it is still used as there are not much alternatives. People use NTFS because they need to run Windows. Thats it.


                    And how have you checked that? You know how I noticed that the these files are getting corrupt? Because my mum complained about a truncated picture being displayed.
                    Luckily, I had the sha512 hashes of all files and could compare and guess what, multiple other files also turned up with different hashes. Just by using the data, you would have never noticed but with sheer luck.
                    Again you continue to defame NTFS without providing any hard evidence that anyone except you and your mom have lost files on it while over a billion of people continue to use it with zero issues. I personally have served three large organizations in my city with over a hundred employees and dozens of servers with uptimes measured in weeks and rebooted only when updates arrived. We did not lose a single file over the span of 15 years while having multiple shared folders with over hundreds of thousands of files while not a single file ever got corrupted. I'm glad you can use checksumming and we used it to, only MD5 which was faster back then (SHA256 is faster with modern CPUs since it's HW accelerated). Again, not a single corruption. I'm sorry but your sample size of two is worth nothing.

                    Also, you claimed you know a billion of legitimate ways to cause a corruption with NTFS. So far you've provided none and I will simply call you a bullshiter. I needed just one and you keep on telling me about your mom's PC. OMG. As we she had a server with RAID using enterprise grade drives, UPS and multiple measures to prevent accidental loss of power or HW resets.

                    Comment


                    • #40
                      Originally posted by birdie View Post
                      I know large organizations which have been running Windows since Windows NT 3.51 and haven't lost a single file on their hundreds of PCs and laptops. The Windows NTFS driver is reliable and tested, period.
                      Large organisations normally run WSUS as well so controlling when windows updates get applied. So its not large organisations that get hurt by Windows NTFS bugs its the small ones. The large organisations let apply smaller organisations the patches first.

                      Originally posted by birdie View Post
                      Unlike Linux, Microsoft extensively tests it and has a very stringent QA/QC policy at least for data retention (they do have regressions and bugs in less critical areas of Windows).
                      Except that is not 100 percent true. How Microsoft QA file system drivers for enterprise is horrible. Very much like Linux. Your home/small business users get the patch if it does not break things so good.


                      This is only a recent example but there are examples with every version of windows. back to NT 3.5.

                      Originally posted by birdie View Post
                      I'm not sure specs are enough to create a fully functional implementation because I've read a lot of horror stories about Paragon products when people completely lost their data. Actually I once lost an entire drive full of data when I tried using Paragon Hard Disk Manager. Luckily there was a backup.
                      This is the problem you have not considered that mainline Linux does have fairly stringent QA. The second version of the patch set for NTFS file system driver Paragon submits for mainline kenrel has 300 faults detected by sparse https://www.kernel.org/doc/html/late...ls/sparse.html fixed that all could have results in entire drive going belly up.

                      Yes that its on to the 9 ninth revision get mainline is mostly found code quality issues.

                      By the way there will have to be a 10 version.


                      The intel automated bot checking for code issues has not liked the patched because it using a value that value has not been defined still in one location. The closer a patch gets to mainline more different parties automated tools will start going over it with Linux. Turns out mainline code into Linux is not easy and items that would pass a normal enterprise level software standards will fail.

                      Remember mainline driver with Linux will have to be built to a way higher quality standard than Paragon own internal so what Paragon has a company as provided is a low quality than what the mainline ntfs driver will be. In fact the quality difference will be massively different particular when 1 tool that mainline drivers have to test forced removed over 300 faults. Intel bot and other bots that check Linux kernel mainline seeking patches the quality standard level is multi different companies idea of quality so the requirements are high than any companies individual requirements.

                      Those submitting hyper-v drivers to Linux from Microsoft have complained that the Linux quality requirements are insanely hard.

                      Reality is until the ntfs driver finally get mainline will will not know how far its quality has moved up. No question the mainline version of the Paragon NTFS driver is going to be better quality than the NTFS driver Paragon sold to their customers. Where it lands compared to Microsoft is going to be interesting.

                      Comment

                      Working...
                      X