Minor suggestion for scope

ben_wizards

Would it possible to have the #of scope channels detected at compile time rather than specified in the setup function?

It's a really minor quibble, but a few times more than I'd care to admit during late night sessions, I've added some scope channels to look at something, compiled, realized that I've forgotten to change the number of scope channels, gone back, changed it to the WRONG # of channels, compiled...Etc.

LiamDonovan

ben_wizards The only way to do this would be to detect the number of arguments given to scope.log() and I don't know of any way of doing that at compile time. It could be done at runtime, but that would increase CPU usage which is not really worth it.

It would be possible to allow the number of channels to be changed through the UI, but that would only be useful in this particular situation, so I'm not sure that's worth it either... Though if what is suggested here is implemented, that will fix this problem.

LiamDonovan

Actually thinking about this it would be possible for scope.log to take a C++11 parameter list that would allow the number of channels to be detected automatically, but the syntax would have to change to
scope.log({ch1, ch2, ...})
with curly brackets. This could maybe be implemented as an overload.

giuliomoro

I do not see a point for this feature to be added, to be honest, though it is feasible.

If you skip the call to "setup", then the first time you call log() you will have to allocate memory for the Scope object. @LiamDonovan, I believe that C++11 feature you describe would make it do the same, right?
It would require either a mode switch (and consequent audio dropout) upon the first call, or a threading system, so that upon the first call to log a background task is started to allocate the buffers and successive calls to log are inactive until the buffers are ready.

I think that providing flexibility to compensate for programming laziness is going to cause troubles later on and I would not encourage that.

Incidentally, there is an overload for log which takes a float*, that might somehow work around your problem in specific usage patterns: http://docs.bela.io/classScope.html#a1fb0a945352e031679916b512375b7ad

Even more incidentally, I am not currently happy with the fact that log behaves differently in WriteFile and Scope, so maybe we should converge to a coherent behaviour.

LiamDonovan

giuliomoro yeah it is quite an awkward thing to do, which is why I settled on specifying the number of channels in setup, even though I too find that a bit annoying.

I agree about the differences in the behaviour of log, in fact I'm not currently happy with the way all of the non-core classes in Bela (including the scope, OSC, UDP, writeFile etc) are handled in general, especially their APIs. We should definitely review them at some point.

ben_wizards

LiamDonovan , you folks have something that is just a few revisions away from being a middle ground between an audio/CV breadboard and a high end Arduino replacement. Anything that makes this more sound design/EE friendly is going to cause this product to blow up. Ease of use, better documentation, some cool libraries, you've got an Arduino/Processing killing machine.

I've never had any interest in Beagleboard, but the Bela is a hell of a thing.