You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The whole business with having to cycle autoassign in NMFMorph is silly and bad, so it's time to fix it. There's actually two aspects to this:
Most urgent: we keep and fix the current behaviour so that autoassign actually does what it says. We can do this, I think, by change tracking the pointers on the input buffers (we don't need to deref them, just compare stored with current). This should tell the client if a buffer has changed in some drastic way, and can be used to set an 'needsAssign' flag to pass through. If the pointer hasn't changed, but the buffer content has, then a toggle would still be needed.
Get that whole business out of the audio thread. That's a much more complex proposition, but really the assignment calculation needs to happen in response to a message and then there needs to be some sort of copy-update to pass the new assignment to the audio thread. That entails a sort of hybrid object we don't yet have a model for.
The text was updated successfully, but these errors were encountered:
The whole business with having to cycle
autoassign
in NMFMorph is silly and bad, so it's time to fix it. There's actually two aspects to this:autoassign
actually does what it says. We can do this, I think, by change tracking the pointers on the input buffers (we don't need to deref them, just compare stored with current). This should tell the client if a buffer has changed in some drastic way, and can be used to set an 'needsAssign' flag to pass through. If the pointer hasn't changed, but the buffer content has, then a toggle would still be needed.The text was updated successfully, but these errors were encountered: