Given that supporting 32-bit floats is an option, I’d like to test it out a proper environment. Which hosts should I use for this?
I currently have a compatibility hack (wrapper class for Vst::AudioBusBuffers which copies data from 32-bit to 64-bit buffers, and copies back on destructor). It works (tested by temporarily rejecting 64-bits in “canProcessSampleSize()”), but obviously it’s going to slow things down a bit.
Before judging whether I should change it, I should check what the performance impact actually is (measured on a proper system), and also know how likely it is to actually be used.