Bug: *transientslice do not give the same results across OSes

[tremblap]

Please tell us what you were doing! You can include code and files by drag and dropping them into the text area.

I am running a simple transient extraction. In MacOS, Max, Pd and SC give the same values. In Windows, again the same values between CCEs... but not between OSes! In Linux (arm) again another set of values.

What was the expected result?

Since I cannot see any random seed in the extraction code (the randomSampling function seems not in use), and that I get the save values between CCEs within an OS, I was expecting the same values between OSes.

What was the actual result?

different slices points after the first few ones. the first 2 or 3 indices are the same, then they diverge.

What operating system were you using?

Mac

Operating system version

macos 15 Ubuntu 24lts win11

FluCoMa Version

pre-release 1.0.8

[tremblap]

example code in SC for readability:

(
~refBasic = [146, 19189, 34710, 47225, 49466, 58303, 68190, 86944, 105695, 106754, 117448, 122139, 139530, 150492, 152884, 161525, 167575, 186297, 223797, 250361, 256305, 280173, 306507, 310674, 312509, 319116, 327659, 335219, 346779, 364673, 384784, 413937, 429154, 431229, 434504, 439537, 441625, 452032, 453392, 465470, 481516, 494517, 496122, 514126];
~refAdvanced = [667, 19184, 34706, 47229, 58300, 68183, 86942, 106752, 117438, 122135, 150487, 161523, 167572, 179048, 186294, 205050, 220493];

~src = Buffer.read(s,FluidFilesPath("Tremblay-AaS-SynthTwoVoices-M.wav"));
~indices = Buffer(s);
)

FluidBufTransientSlice.processBlocking(s,~src,indices:~indices,action:{~indices.loadToFloatArray(action: {|x|~basic = x})});

FluidBufTransientSlice.processBlocking(s,~src,indices:~indices, numFrames: 220500, order: 200, blockSize: 2048, padSize: 1024, skew: 1, threshFwd: 3, threshBack: 1, windowSize: 15, clumpLength: 30, minSliceLength: 4410, action:{~indices.loadToFloatArray(action: {|x|~advanced = x;\done.postln;})});

~refBasic == ~basic
~refBasic.size == ~basic.size
~refBasic.do{arg v,i; if (v != ~basic[i], {"% (%) ".postf(i, v - ~basic[i])}) };

~refAdvanced == ~advanced
~refAdvanced.size == ~advanced.size
~refAdvanced.do{arg v,i; if (v != ~advanced[i.min(~advanced.size - 1)], {"% (%) ".postf(i, v - ~advanced[i.min(~advanced.size - 1)])}) };

[tremblap]

actually, in Linux I don't even get the same values between Pd and SC...

[tremblap]

@AlexHarker might confirm there is no randomisation used - or that I'm wrong, and that would make this a non-bug...

[tremblap]

"good" news - it is not new. So I will plough on with the release checks.