teotigraphix / Framework4Bitwig

Bitwig controller API framework extensions.
GNU Lesser General Public License v3.0
58 stars 13 forks source link

Cannot turn on audio engine (Application.activateEngine()) #249

Closed git-moss closed 4 years ago

git-moss commented 4 years ago

After the Audio engine has been turned off (Application.deactivateEngine()), it cannot be turned on again (Application.activateEngine()).

The Bitwig log shows the following Exception:

[2020-01-12 10:14:18 ramona-server error] Connection broken with client: java.io.EOFException at com.bitwig.ramona.serial.bNN.EUM(SourceFile:263) at com.bitwig.ramona.serial.bNN.tYM(SourceFile:413) at bEY.E3n(SourceFile:847) at bEY.EkP(SourceFile:669) at bEY.EkP(SourceFile:650) at com.bitwig.ramona.protocol.ubZ.EkP(SourceFile:88) at bFR.hlF(SourceFile:385) at bFT.run(SourceFile:276)

NickJAllen commented 4 years ago

I can't reproduce this at all.

git-moss commented 4 years ago

Did you try only on Linux?

NickJAllen commented 4 years ago

Not necessary to check windows or mac. The API just calls the same thing as when the user clicks does activate / decactivate via the GUI. So I guess there are some problems depending on the audio backend used and audio settings and these problems would also be there for the user doing it via the GUI. Can you confirm this? If so the bug has nothing to do with the API at all. But would still be a serious bug we need much more info on

git-moss commented 4 years ago

It happens only via the API. 1) Deactivate via API -> Audio engine is off 2) Activate via API -> Crash 3) Activate via API -> Still crashing 4) Activate via UI -> Crash 5) Activate via UI -> Works again

Will test on Linux tonight.

NickJAllen commented 4 years ago

Did you send engine crash reports? I don't see any...

git-moss commented 4 years ago

There was no exception in the audio engine log. Seems to be more like a communication problem.

NickJAllen commented 4 years ago

The exception is the log is completely harmless and expected

git-moss commented 4 years ago

The explanation is that the MIDI processing runs now in the audio thread, which is deactivated after the function call. Therefore, a conceptual problem not a real bug.

git-moss commented 4 years ago

Fixed in Bitwig 3.1.3.