|
I'm having a strange problem with Phasor initializing incorrectly when its 'start' input is a Latch. Illustrating with a couple of units from a more complex SynthDef -- first, this tells how much time elapsed between the synth start time and the time the trace was taken. unit 17 Line in 0 500 500 0 out 0.0522449 0.0522449 sec * 44100 samp/sec * (1/64) control/samp ≈ 36 control blocks Then... unit 19 Phasor in 1 0.015625 128.002 5.17337e+06 128.002 out 5.17325e+06 But, 128.002 + (0.015625 * 36) = 128.5645 -- that's a bit more than rounding error, no? I can reproduce it more simply this way. The original SynthDef pulls the starting buffer position (normalized to 0-1 for GrainBuf) from a control bus and latches it because the control bus value will change continuously. n = 5173374; z = Bus.control(s, 1); z.set(128 / n); { var debugTrig = 1 - Trig1.kr(Impulse.kr(0), 0.0522449), rate = ControlRate.ir / SampleRate.ir, inPos = Latch.kr(In.kr(z, 1), 1) * n; Phasor.kr(1, rate, inPos, n, inPos) .poll(debugTrig); FreeSelf.kr(debugTrig); Silent.ar(1) }.play; UGen(Phasor): 5.17325e+06 Removing Latch: { var debugTrig = 1 - Trig1.kr(Impulse.kr(0), 0.0522449), rate = ControlRate.ir / SampleRate.ir, inPos = In.kr(z, 1) * n; Phasor.kr(1, rate, inPos, n, inPos) .poll(debugTrig); FreeSelf.kr(debugTrig); Silent.ar(1) }.play; UGen(Phasor): 128.562 - that's the right one! I think the issue is that Latch should initialize its output to the input value, instead of as is currently assumed. After changing the last line of Latch_Ctor to this: ZOUT0(0) = ZIN0(0); // 0.f; ... the test case above (with Latch) returns the expected 128.562 value. Commit? Please say yes :-) I've had other issues with Latch where the expected init value was not passed further down the chain. hjh : H. James Harkins .::!:.:.......:.::........:..!.::.::...:..:...:.:.:.:..: "Come said the Muse, Sing me a song no poet has yet chanted, Sing me the universal." -- Whitman |