Announcement

Collapse
No announcement yet.

Fedora 21 Aims To Have LBZIP2 Replace BZIP2

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

  • Fedora 21 Aims To Have LBZIP2 Replace BZIP2

    Phoronix: Fedora 21 Aims To Have LBZIP2 Replace PBZIP2

    Fedora developers are eyeing the independent lbzip2 compression implementation to become the default bzip2 on the next release of this Red Hat sponsored Linux distribution...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Michael, bzip2 NOT Pbzip2.
    I think they should stop bitching and just replace the command line tool with lbzip2.
    On the other side Mikolaj Izdebski should reassure us he will start writing an ABI compatible library. "if there will be demand" means NEVER considering there is ALREADY demand.
    ## VGA ##
    AMD: X1950XTX, HD3870, HD5870
    Intel: GMA45, HD3000 (Core i5 2500K)

    Comment


    • #3
      My first thought was: "Finally, some distro that defaults to parallel (de)compression."

      However, while looking in it more closely it seems like a pet project of redhat (lbzip2 creator: Mikolaj Izdebski works for redhat) being pushed in fedora.

      The biggest concern that i have is no API at all. bzip2 has an API and applications (or other libraries) use that. Replacing bzip2 with lbzip2 won't help those apps/libs. They will have to keep linking against bzip2 because there is no replacement there.So the result you're going to get if they (or any distribution) is going to include lbzip2 as "default" is:
      - command line utility uses lbzip2 (you will notice a big speedup)
      - apps/libs keep using bzip2 (you won't notice any changes)
      - results in both lbzip2 and bzip2 being installed

      I think the idea of replacing bzip2 with a parallel friendly version is nice and needed, however, the replacement should be fully compatible with bzip2 (also in API terms) so that bzip2 can slowly move away and apps link against lbzip2. As long as this is not the case, i see no real point from a desktop perspective to replace it. From a command line perspective it's fine, but then it's just a command line tool.

      Comment


      • #4
        We shouldn?t worry about that, because systemd will soon include transparent parallel decompression of all files.

        Comment


        • #5
          Originally posted by darkbasic View Post
          Michael, bzip2 NOT Pbzip2.
          I think they should stop bitching and just replace the command line tool with lbzip2.
          On the other side Mikolaj Izdebski should reassure us he will start writing an ABI compatible library. "if there will be demand" means NEVER considering there is ALREADY demand.
          lol, fedora as biggest proponent of gnome/gtk is concerned about ... stable API/ABI in zip library??? don't know why, but something sure is screwed in this message

          Comment


          • #6
            Originally posted by justmy2cents View Post
            lol, fedora as biggest proponent of gnome/gtk is concerned about ... stable API/ABI in zip library??? don't know why, but something sure is screwed in this message
            GNOME/GTK has a stable API/ABI for the core platform. Any ABI breakages within a stable release series is considered a bug. Perhaps something is screwed up about your understanding. You might be thinking of theme breakages which aren't part of either API or ABI.

            Comment


            • #7
              Originally posted by stqn View Post
              We shouldn?t worry about that, because systemd will soon include transparent parallel decompression of all files.
              LOL Nicely spotted!

              Comment


              • #8
                IMO its time to forget bzip2 at all. If someone needs relatively fast compression, gzip would do (and LZ4 and LZO if gzip is "too slow" for you). If someone needs strong compression, LZMA is their choice. Bzip2 is clearly obsolete these days in terms of speed vs compression ratio. So IMO there is just too much buzz on this outdated algo.

                Comment


                • #9
                  bzip2 still wins on sparse (mostly zero-filled) files, such as usb stick images. Just tested 100mb of zeroes:

                  gzip -9: 100k
                  bzip2 -9: 113 bytes
                  xz -9: 15k

                  Comment


                  • #10
                    Not that 100MB → 15kB is bad. From some tests I did on highly repetitive text files I remember LZMA being a lot better than bzip2 (like 1000 times better, albeit on an unusual example). So yeah, why care about bzip2 any more.

                    Comment

                    Working...
                    X