Hi forum,
I found an issue trying to sample a list and a bpf (both with length=11).
om-sample gives me different list lengths if I use sample-rate=27
Please, see attached picture for a better explanation.
Let me know if someone has the same problem. Thanks in advance!
OM 6.8_r4
MacOSX 10.6.8
Best regards,
RC
Yep, it happens. With 14 samples (samples 4), then with 27 (samples 26)… I can only expect it to sample it wrongly many more times as number of samples goes up.
Anyway, you should consider adding a preventive patch of code. Perhaps with an if testing whether the right amount of samples is being output. You would then add one more sample: perhaps a mean value between any two adjacent values. If you don’t want the if, just sample more than you want (I’d say just one more sample) and then use first-n with the correct number of samples you want.
Hope this helps.
Sampling two times also seems to work.
Here it goes.
…After more inspection, it seems the problem is really with sampling a BPF on its first output (self). If you sample two times by using the self output of OM-SAMPLE, the same error recurs. If you sample the list of Y-values (third output), it works OK.
Hello,
Using the y-list is ok as long as you don’t need to x-positions of the points (or when they are evenly placed like here).
The general case (with BPF) fails in some particular cases due to floating point approximations.
I think the attached fix will solve it in this particular case. It uses double-float precision intead of single float.
(put the lisp file in OM 6.8/patches/)
[Here, we want 27 points between 0 an 1000, the “step” of the result is 1000/26 = 38.46154 (in single float approx.) The 27th point (26x38.46154=1000.00007) is then > 1000 and not included in the result. In double float precision, the step would be 38.46153846153846 and the 27th point is 1000.]
om-sample-double-precsision.lisp (2.24 KB)
Hi Gato, hi Jean,
Thank you very much for your answers and solution, now it works! 
best,
rc