Announcement

Collapse
No announcement yet.

F2FS Will Have Faster Case-Insensitive Lookups With Linux 5.4

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

  • F2FS Will Have Faster Case-Insensitive Lookups With Linux 5.4

    Phoronix: F2FS Will Have Faster Case-Insensitive Lookups With Linux 5.4

    The EXT4 case-insensitive lookup optimization added to the file-system with the current Linux 5.3 cycle has been ported to the Flash-Friendly File-System (F2FS) for the upcoming Linux 5.4...

    http://www.phoronix.com/scan.php?pag...Insensitive-54

  • #2
    I find it really sill what a fuzz is made here re-occurringly about such an anti-features. Who in it's right mind would want to use such a crippled DOS name lookup thing? Commenting something like that previously I learned it is also about unicode normalisation, which fair enough should be more useful. But case-insensitively alone? I certainly am not interested at all in such an antiquated implementation detail.

    Comment


    • #3
      Originally posted by rene View Post
      I find it really sill what a fuzz is made here re-occurringly about such an anti-features. Who in it's right mind would want to use such a crippled DOS name lookup thing? Commenting something like that previously I learned it is also about unicode normalisation, which fair enough should be more useful. But case-insensitively alone? I certainly am not interested at all in such an antiquated implementation detail.
      It's not like it's on by default and case-insensitivity can be considered to be a feature for non-power users and non-technical people like my Dad. It's also something that's needed for Wine and some programs running with Wine.

      I'm glad we're getting another tool in the box and that native Linux file systems and tools are getting closer to being as feature-filled and as robust as ZFS.

      Comment


      • #4
        Originally posted by rene View Post
        Who in it's right mind would want to use such a crippled DOS name lookup thing?
        It's situational, mostly for running Windows crap in Wine and file sharing with Samba (that do expect this and at the moment are relying on a more expensive workaround).

        Can be useful for any folder that stores only man-created files as most people won't 100% remember the case of all letters in the filenames.


        I don't see what there is to be so enraged about. Someone paid someone else to implement it, that's more proof that there is some need, even if niche. (Although I would not call Samba usage "niche" by any stretch of the imagination)

        Comment


        • #5
          But not on openSUSE Tumbleweed.

          Comment


          • #6
            Originally posted by rene View Post
            antiquated implementation detail.
            Funny considering case insensitive is harder to implement than case sensitive. Instead of simply comparing bytes you have to take care of sensitivity. When you have a lot of alphabets (We have Unicode) it's become quite difficult. Really compare with DOS? DOS had filenames with format 8.3, supported only ANSI and wasn't case preserving. Modern implementation (like this on ext4 and F2FS) should support Unicode, be optional and case preserving. If you compare this with DOS you can also compare modern Linux with 70's UNIX.

            Another thing: Case insensitive is more natural to typical users and humans at all. Why "Letter.txt" and "letter.txt" should be different things? Why changing character case make some name totally different name? Case sensitivity forces user to think like computer do - file names is not file names but just a string of bytes. If Linux is about freedom why not give case insensitivity to user which wants it instead of trying to make him take different approach with arguments like "Its old Windows/DOS crap don't take it"?

            Comment


            • #7
              Originally posted by xorbe View Post
              But not on openSUSE Tumbleweed.
              It's a widely known fact that people on Tumbleweed don't need this.

              (more seriously, what are you talking about?)

              Comment


              • #8
                Originally posted by rene View Post
                I find it really sill what a fuzz is made here re-occurringly about such an anti-features. Who in it's right mind would want to use such a crippled DOS name lookup thing?
                DOS isn't truly case-insensitive. It is always-uppercase.

                Comment


                • #9
                  Originally posted by dragon321 View Post

                  Funny considering case insensitive is harder to implement than case sensitive. Instead of simply comparing bytes you have to take care of sensitivity. When you have a lot of alphabets (We have Unicode) it's become quite difficult. Really compare with DOS? DOS had filenames with format 8.3, supported only ANSI and wasn't case preserving. Modern implementation (like this on ext4 and F2FS) should support Unicode, be optional and case preserving. If you compare this with DOS you can also compare modern Linux with 70's UNIX.

                  Another thing: Case insensitive is more natural to typical users and humans at all. Why "Letter.txt" and "letter.txt" should be different things? Why changing character case make some name totally different name? Case sensitivity forces user to think like computer do - file names is not file names but just a string of bytes. If Linux is about freedom why not give case insensitivity to user which wants it instead of trying to make him take different approach with arguments like "Its old Windows/DOS crap don't take it"?
                  Because all this added complexity open a mostly unnecessary pandora can of bugs. Very few people ever asked for this, and the Linux kernel already has enough random bugs. Make software simpler not over complex at each and every corner.

                  Comment


                  • #10
                    Originally posted by rene View Post
                    Because all this added complexity open a mostly unnecessary pandora can of bugs. Very few people ever asked for this, and the Linux kernel already has enough random bugs. Make software simpler not over complex at each and every corner.
                    This is code that does not affect most other parts of the kernel. At worst, this specific feature is bugged.

                    Comment

                    Working...
                    X