Announcement

Collapse
No announcement yet.

Rumblings in the Linux Audio World

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

  • #16
    Originally posted by kraftman View Post
    Isn't this meaningless screaming? If they released sources under GPL why someone should care about their terms of use? If he should, maybe that's a reason why Linux devs aren't interested?
    It seems like the lack of interest mostly has to do with technical concerns (e.g. suspend/resume support and the division of labor between user and kernel space) rather than licensing issues. A more cynical observer might suppose the influence of NIH syndrome as well.

    Originally posted by RealNC
    The OSS specification is fully open and you are not required to follow ANY license in order to implement it.
    Thankfully, that much is clear (even to a nitpicker like me )

    Comment


    • #17
      We've been beating around the central issue though. Well, the "issue" as I see it, if there's an issue at all:

      ALSA has the following problems:
      • High latency (only an issue with games and virtual instruments/synthesizers.)
      • Poor OSS support (dmix doesn't work).
      • Sound quality issues (sampling/mixing quality and playback during high system load).
      OSS4 doesn't have the above problems. Furthermore, one can have the best of both worlds by using OSS4 over the ALSA libs:

      Install the "ALSA Plugins" package of your distro ("alsa-plugins" in Gentoo, "libasound2-plugins" in Debian and Ubuntu). Put this in /etc/asound.conf:

      Code:
      pcm.!default {
              type oss
              device /dev/dsp
      }
      
      ctl.!default {
              type oss
              device /dev/mixer
      }
      Nothing more is required. So, a distro that would have this setup, would provide the following features:
      • Full ALSA compatibility.
      • Full OSS API support.
      • The benefits of OSS4 also apply to ALSA applications (sound quality, CPU load).
      That would be (in my very, very humble opinion) the best set-up of any distro. Scratch PulseAudio, don't enable sound in the kernel, take ALSA-libs and run them over OSS4.

      Am I the only one to see a lot of sense in this?

      Comment


      • #18
        Originally posted by RealNC View Post
        That would be (in my very, very humble opinion) the best set-up of any distro. Scratch PulseAudio, don't enable sound in the kernel, take ALSA-libs and run them over OSS4.

        Am I the only one to see a lot of sense in this?
        Any way to get rid of pulseaudio is good Maybe Ubuntu brainstorm will be good place to give OSS4 more attention?

        EDIT:

        I have some problems with this. When I'm listening to the music in Amarok 2 and I open movie in SMPlayer there's no sound.
        Last edited by kraftman; 06-22-2009, 02:54 PM.

        Comment


        • #19
          Originally posted by RealNC View Post
          We've been beating around the central issue though. Well, the "issue" as I see it, if there's an issue at all:

          ALSA has the following problems:
          • High latency (only an issue with games and virtual instruments/synthesizers.)
          • Poor OSS support (dmix doesn't work).
          • Sound quality issues (sampling/mixing quality and playback during high system load).
          OSS4 doesn't have the above problems. Furthermore, one can have the best of both worlds by using OSS4 over the ALSA libs:

          Install the "ALSA Plugins" package of your distro ("alsa-plugins" in Gentoo, "libasound2-plugins" in Debian and Ubuntu). Put this in /etc/asound.conf:

          Code:
          pcm.!default {
                  type oss
                  device /dev/dsp
          }
          
          ctl.!default {
                  type oss
                  device /dev/mixer
          }
          Nothing more is required. So, a distro that would have this setup, would provide the following features:
          • Full ALSA compatibility.
          • Full OSS API support.
          • The benefits of OSS4 also apply to ALSA applications (sound quality, CPU load).
          That would be (in my very, very humble opinion) the best set-up of any distro. Scratch PulseAudio, don't enable sound in the kernel, take ALSA-libs and run them over OSS4.

          Am I the only one to see a lot of sense in this?
          If only it was that simple. OSS 4 doesn't even support midi, it has a much smaller list of supported devices and many of those devices are limited to basic functionality (Forget about SoC setups as well). Latency is not an issue if one configures their system correct (pulseaudio on the otherhand is horrible when it comes to padding on the latency). Dmix I haven't had an issue with for years, (it's setup by default since 1.0.9 and no .asoundrc configuration is really needed anymore.) In fact the 3 issues you list sound more like a kernel config issues then alsa issues. Using a preempt kernel and setting your PCI latency timers as well as getting rid of cycle sucking crap like pulse can address your issues with alsa.

          Comment


          • #20
            This is quite funny to follow. It seems to me like one of the kill Lennart Poettering Threads, his stuff isn't working for me. I quite like his work.
            You can't exactly match a combination of Alsa and Pulse to OSS. Pulse is more like an welcome addon to the whole stack because it offers per stream volume adjustments, network transparency and autoconfiguration, synchronous sound multicasting etc. I'm pretty sure you won't get this results with neither pure Alsa or OSS.
            While Pulse already put me some major headaches it's quite fun once you get it working. Leave alone professional sound production here which always has its own very special needs. I'm talking of the normal Desktop Userbase. Pulse is also Multiplatform and available for BSDs,Windows,...
            One of the Ideas in Pulse is to have an abstraction from any OS-specific api which you're talking about. And while Pulse is getting a lot of attraction on Linux already others are a little slow to follow while Pulse is also still maturing. Let's see what future brings.
            So it's basically boiling down to comparing alsa to oss, especially on Linux. I think the state of alsa is rather well compared to oss here. See the Arguments of my predecessor: More supported soundcards, sometimes also with more functionality, midi support, fair working oss compatibility. I think this is the same thing like MS keeping DOS support. You can't put OSS away, like DOS it has always worked well for many people, but its a little 1982 at least. If you always just keep the things in their current state there is not much room for improvment.
            Another DOS comparison: It was nearly unable to do anything useful, but it took you really fast and reliable to the point where it was failing gracefully
            So if you're still happy with OSS talk to the Developers of all that evil Software instead of moaning here to convince them putting their effort in developing for OSS. See if your oppinion matches the wishes of other people, their userbase. If you're mostly alone you're out of luck here. And i think this is going to be the case.

            Greetz !
            Hibbelharry

            Comment


            • #21
              FWIW, Hannu has disavowed the comments about runtime licensing and said that the "common interpretation" of the GPL will be followed.

              Comment


              • #22
                Judging from the discussion over at Insane Coding and what you guys have been addressing here as well, two things stand out as currently being problematic:

                i) license: is it Free or not? Do different licenses apply to source and binaries? License issues *always* have to be settled before distros can depend on large projects like this.

                ii) sleep/resume: does OSS4 work on a laptop that goes to sleep a few times a day? Has someone around here used it on a laptop in this way?

                Comment


                • #23
                  Originally posted by korpenkraxar View Post
                  i) license: is it Free or not? Do different licenses apply to source and binaries? License issues *always* have to be settled before distros can depend on large projects like this.
                  The binaries have a few drivers in them that have not been open sourced (due to NDAs). However, distros don't need the binaries. Fortunately, the binary-only drivers are for one or two very rare cards.

                  Comment


                  • #24
                    About licensing:

                    What this means is that we can decide to license OSS under some other license if we wish - eg. lets say Google comes out with a Google Public License for their operating system in the future. 4Front can basically say, we license OSS under the Google PL. This has NO bearing on the GPL licensed version but essentially since we are the original developer we reserve the right to relicense any code under another license - so basically GPL isn't the highest license authority of Open Sound.

                    We are no longer in the proprietary business - instead we are in the community business. We want to serve multiple communities that may not fall under the same license - so 4Front license supercedes all other licenses.

                    When you work with the GPL'ed version you don't care about the other licenses. You are given 100% rights under the GPL. There is no exception in GPL'ed version OSS what so ever. You can contribute your changes back under GPL but if you want it to appear under the Open Sound master repository - you would need to sign a OSS contributor agreement under which you would relicense your work so that all other licenses benefit (including GPL).

                    The contributor agreement is basically a mashup of the Fedora CL and the Open Solaris CL.


                    THis is our policy at 4Front and we apologize if there has been any miscommunication.

                    Comment


                    • #25
                      Originally posted by RealNC View Post
                      The binaries have a few drivers in them that have not been open sourced (due to NDAs). However, distros don't need the binaries. Fortunately, the binary-only drivers are for one or two very rare cards.
                      Sounds reasonable indeed.

                      Comment


                      • #26
                        Originally posted by kraftman View Post
                        About licensing:
                        Ok. This also sounds pretty reasonable

                        Comment


                        • #27
                          I don't know if you've noticed, but OSS4 looks like a dying project:

                          There are three main problems with OSS at this moment

                          1. Nobody is paying anything for “free” software. [...]
                          2. The current sound card architectures like USB, FireWire, HDaudio and X-Fi are all bogus. [...]
                          3. Practically all legacy OSS/Free applications are seriously broken. [...]

                          [...]

                          However what I will probably NOT do are:

                          * Fixing the USB audio driver. [...]
                          * Also fixing the HDaudio driver [...]
                          * SB X-Fi [...]
                          * [...] get the legacy OSS applications fixed.
                          The architectures are bogus? Won't fix HDA and X-Fi? This covers about 95% of all new sound chips out there!

                          These are the words of a dying project. OSS support is failing while ALSA is still going strong. The choice is pretty clear at this point.
                          Last edited by BlackStar; 06-23-2009, 09:06 AM.

                          Comment


                          • #28
                            It's not dead yet though. Maybe its user base is keeping it alive, but without more developers, the future doesn't look very bright.

                            Comment


                            • #29
                              Nobody mentioned it yet, but OSS 4.1 supports per-application volume control.

                              Comment


                              • #30
                                The last time I checked, OSS4 had problems with sleep, and one had to run the 'soundoff' script before sleeping and run the 'soundon' script after.

                                It seems to me like one of the kill Lennart Poettering Threads, his stuff isn't working for me.
                                Lennart Poettering is irrelevant to this thread, just like he claims OSS4 irrelevant to Linux.
                                Off-topic: libcanberra is a nice replacement for esd when using the gstreamer backend, so there's no need to kill Lennart.

                                Maybe Ubuntu brainstorm will be good place to give OSS4 more attention?
                                From what I've seen, the OSS4 threads on Ubuntu brainstorm generally get voted down.

                                Comment

                                Working...
                                X