Uwe, I truly appreciate the work that you have done on the sound source API. It's an important change and it does fix some long-standing issues. The choice has never been between merging the code or throwing it away.
Based on the risks and benefits and our experience with long debugging periods after major feature merges, the prudent choice is to keep the current soundsource API for the 1.12 release. Once it's out the door, then we can immediately look at merging the new API. And as far as I'm concerned, as soon as it's stable, we can do another beta right away. I know it's frustrating to have a feature not make it into a release, but it's a self-perpetuating problem. Because releases are rare, we all want everything we do in the next release. Merging everything in slows down the release cycle, and the cycle repeats. If we released more often, it'd be easier to allow a feature to miss a release, and releases would speed up. Therefore, to stop the current cycle of non-releases, we have to agree to hold off on merging this code until after 1.12. I honestly believe that not merging this code now will make it possible to release this code *sooner*! Owen On Fri, 2015-04-17 at 10:55 +0200, Uwe Klotz wrote: > Just some words from my side. I really don't want to give any > recommendations to you, because I'm biased ;) > > The NewSoundSourceAPI branch has gone through many iterations and > Daniel spent a LOT of time reviewing my code very thoroughly. He also > provided the key ideas to fix the MP3 decoding issues! During the last > months the code base has become pretty stable and we have not > discovered any severe defects lately. It is secured by new tests for > many codecs and moreover the critical parts of the code are guarded by > lots of debug assertions that should indicate potential issues or > wrong assumptions early during development in debug builds. > > Of course, I would feel much safer if the new code had been tested by > experienced developers first before rushing it out to the public. On > the other hand, I personally never switched back to master without any > regrets after what I have found there! > > Main features: > * API: Floating-point sample processing > * API: Explicit notion of #channels instead of implicitly assuming > always stereo > * Update, optimization or even re-implementation (MP3, FLAC, > AAC/M4A, ...) of audio decoding > * Many metadata improvements (parsing, character encoding) > > Open issues: > * Windows build with SoundSourceMediaFoundation > * OS X build with SoundSourceCoreAudio > > Uwe > > On 04/17/2015 09:25 AM, Daniel Schürmann wrote: > > > Hi Owen, > > > > We have collected some test results, they are proving that the new > > sound source branch is more reliable then the master solution. > > > > We have no hints for the opposite. > > > > > > So IMHO is worth to consider the 1.12 merge again. > > > > > > > > Yes, master the sound source api has not changed since 1.11, but > > Mixxx 1.12 actually makes a difference since it use a lot of seeks > > when using the new shiny > > > > beat matching algorithms. > > > > They simply fail with effected mp3 tracks. User may blame the > > syncing code in the first place. > > > > > > > > I hope we will get some results from Ryan Kramer before the desired > > beta date. > > > > > > We are still looking for a user that is able to test the Mac > > builds. > > > > It would be nice, if we could provide build server builds for the > > sound source branch. This would allow much more users to run a > > test. > > > > > > Kind regards, > > > > > > Daniel > > > > > > > > > > > > > > > > > > 2015-04-17 1:42 GMT+02:00 Owen Williams <owilli...@mixxx.org>: > > On Thu, 2015-04-16 at 21:40 +0200, Daniel Schürmann wrote: > > > Hi Owen, > > > > > > beta before May 16th should be possible. > > > > > > We have a remaining blocker, the mp3 off beat, click and > > distortion issues. > > > > The current mp3 and soundsource code is the same as in the > > current 1.11 > > release, so by definition it's not really a blocker since we > > already > > have a release with that same code. Yes, it's a bug, but it > > shouldn't > > block the beta or release. We've already discussed this > > issue. > > > > > > > This and the fact that no on is working on a back port is > > a good hint > > > that we should merge > > > Uwes branch before beta. > > > > That no one wants to work on a back port for the mp3 fixes > > is not an > > argument in favor of a premature merge. I am happy to make > > the > > soundsource review a priority after we do the release. > > > > > > > > > > The other issue that does not itches me but might be very > > bad for the > > > Mixxx reputation at all, is > > > https://bugs.launchpad.net/mixxx/+bug/1441046 > > > "Mixxx 1.12: library unusable slow spinning HD Win 8.1 > > notebook" > > > IMHO this does no block the beta, but the release. > > > Does anyone else experience this? > > > > Since we are unable to reproduce this issue, it definitely > > shouldn't > > block the beta. If anything we're more likely to get bug > > reports that > > will help us identify the issue. > > > > > > > > It would be also nice if we can merge the pending fading > > and clicking > > > issues in engine buffer. > > > @Owen will you find time for decide over the pending > > questions? > > > > These are low priority bugs and should not block beta. > > > > The last blocker bugs we fixed were the issue with canceled > > library > > scans breaking the library, and the soundtouch crash. That > > sort of > > data-loss bug and party-crasher are what I mean by blockers. > > > > The only other one I can think of is the issue with win64 > > builds > > crashing on a lot of systems, but we might just have to tell > > people to > > expect that problem and help us get feedback to fix it. > > > > owen > > > > > > > > > > > > ------------------------------------------------------------------------------ > > BPM Camp - Free Virtual Workshop May 6th at 10am PDT/1PM EDT > > Develop your own process in accordance with the BPMN 2 standard > > Learn Process modeling best practices with Bonita BPM through live exercises > > http://www.bonitasoft.com/be-part-of-it/events/bpm-camp-virtual- event?utm_ > > source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF > > > > > > _______________________________________________ > > Get Mixxx, the #1 Free MP3 DJ Mixing software Today > > http://mixxx.org > > > > > > Mixxx-devel mailing list > > Mixxx-devel@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/mixxx-devel > > ------------------------------------------------------------------------------ > BPM Camp - Free Virtual Workshop May 6th at 10am PDT/1PM EDT > Develop your own process in accordance with the BPMN 2 standard > Learn Process modeling best practices with Bonita BPM through live exercises > http://www.bonitasoft.com/be-part-of-it/events/bpm-camp-virtual- event?utm_ > source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF > _______________________________________________ > Get Mixxx, the #1 Free MP3 DJ Mixing software Today > http://mixxx.org > > > Mixxx-devel mailing list > Mixxx-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/mixxx-devel ------------------------------------------------------------------------------ BPM Camp - Free Virtual Workshop May 6th at 10am PDT/1PM EDT Develop your own process in accordance with the BPMN 2 standard Learn Process modeling best practices with Bonita BPM through live exercises http://www.bonitasoft.com/be-part-of-it/events/bpm-camp-virtual- event?utm_ source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF _______________________________________________ Get Mixxx, the #1 Free MP3 DJ Mixing software Today http://mixxx.org Mixxx-devel mailing list Mixxx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mixxx-devel
