The gist of the problem is that I habitually use more than one monophonic sequencer & merge the CV/Gate outputs to drive a polyphonic voice. It seems like if I do that and stop one sequencer, the VCV Merge module that handles pitch CVs is sending the last pitch of channel out another channel.
Has anyone else noticed VCV Merge getting confused about CVs and polyphonic channels? @jeremy is there some variable shared across Gridseq module instances that might be causing this?
@jeremy I don’t think that Gridseq is the problem. I read the entire source code and saw nothing suspicious. What would have really been suspicious is a static class member or static var inside of process, which would have been very bad, unless there was a good reason for multiple module instances to share data.
A good reason would be as a way for a ModuleWidget to share graphic resources created at runtime, for example.
I downloaded the patch from your github issue and I couldn’t reproduce the issue. I was using a poly volt meter (computerscare debug or NYSTHI MultiVoltmetro will do) to look at channel 2 of the V/Oct poly cable and never saw it do what you described. Have you tried using a voltmeter to get a visual confirmation of the problem or are you noticing it based on the sound? My laptop is too weak to run that patch without problems so I was only going by the meters.
It may not be the Merge thing, as much as @almostEric’s Probably Note scrambling up the triggers. I trimmed the patch down to where @computerscare can run it.
If I turn off ‘Run’ on the left-hand gridseq, the trigger input only gets triggers on the 2nd channel of the polyphonic trigger input, but it isn’t clear what’s getting sampled. The pitch just stays the same.