This to bring common fomatting to the whole project,
including submissions. To format a file to this style:
astyle --style=linux -t8 <file>
Future patches / pull-requests should be in this format.
Fixes dsp bug in how the master diff was calculated,
which left some room for error and hence some just-not-silence
levels sneak through the fader.
Increase the calculation threshold to be more sensitive, so
there is now 10x less creep. Testing here proved that it never
bleeds audio anymore
[HvH] I modified this commit - a few changes I didn't like, for
example removing the src/config.hxx debug variables - I know they
are commented, but I use the regularly during development to get
extra prints. So this commit removed some of your changes - just
a note that you're not wondering where the changes went :)
Thanks for the PR!
DSP and GUI level defaults for JackSend are now 1, so when
you click the FX button, by default the audio will be routed
to the send, returned, and audible again. This is better UX.
Fixed 2 build warnings to do with string termination
Looks better than before, but still not ideal. Perhaps adding
a global (master) FX send to JACK could fill the space?
Or else re-layout the items (labels under dials?) to fill void.
Although this allows loops to go out of sync, I think its more
"musical" than the re-trigger noises of scrubbing the BPM dial...
In short - in a live performance I do not expect musicians to
change tempo in a song - between songs it possible, and then
triggering a scene will bring everything back into sync.
This needs a better solution, but right now I'm not sure how
to best do that. The other changes in the patches that changed
this behaviour are valuable, so commiting to get closer to a
solution - but I'm not comfortable with the retrigger noises.
This quick-fix ensures that when the tap-average is
really slow, that the TimeManager doesn't try to
set it as a BPM value. When Transport is stopped,
the average FPB could be 0, causing nasty things.
This commit adds an architecture check to CMake,
detecting if we're running on x86_64 or not.
If not, disable the sse instructions. This can
be improved to handle i686 and some ARM optimization.
