LFPulse UGen gives asymmetrical waveform when duty (width) is 0.5

[totalgee]

If you set the width (duty cycle) argument to exactly 0.5, you don’t get a symmetric waveform (equal number of ups and downs). To test this, it's most obvious when you set a high frequency (use an even multiple of the sample rate, so you should expect equal number of ups and downs). If you change the width to slightly over 0.5, it works as expected.

Compare:

var freq = s.sampleRate/4; {[LFPulse.ar(freq, 0, 0.5), LFPulse.ar(freq, 0, 0.500001)]}.plot(freq.reciprocal * 5);
// First one is [H, L, L, L, H, L, L, L, ...]
// Second one (correct) is [H, H, L, L, H, H, L, L, ...]

I tested this in v3.6.5 on OS X, but the code (in LFPulse_next_a() doesn't seem to have changed in current). Fixing this might change the sound of existing synths (at high frequencies). On the other hand, it definitely seems wrong; you'd expect a duty of 0.5 to produce equal numbers of highs and lows over time!

[totalgee]

Fixed and merged with PR #1503.