Specifically, I always start with
--speaker=off, and then when I need sound I turn out speaker via context menu. When both server and client run 1.0.7, and I start speaker from context menu, I get this error message:
2017-08-14 14:14:23,462 sound output play: invalid sink: pulsesrc 2017-08-14 14:14:23,499 the wav sound sink has stopped and sound forwarding doesn't work.
When I run 1.0.6 on the client (while the server is still 1.0.7),the error does not appear and the sound forwarding works.
It is natural to speculate that the bug is introduced, on the client side, in 1.0.7
There were only two changes in the sound subsystem for 1.0.7: xpra sound commits
This is now fixed up in r16700 and will be included in 1.0.8. You can apply by hand, or workaround it using:
XPRA_SOUND_SINK=pulsesink xpra ...
this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/1623