I would like to write a signal into an array (tabwrite~) and I want to know at all times exactly which sample index is being written.

In SC I would do this with:

var phase = Phasor.ar(0, 1, 0, BufFrames.kr(bufnum));
var signal = ... something...;
BufWr.ar(signal, bufnum, phase);

In Pure Data.... [tabwrite~] takes no phase inlet. Nor does it provide a phase outlet.

I could do a sample-by-sample integrator with [rpole~ 1] but then... how to reset it exactly at the end of the array? I could just let it integrate for a few minutes, but without resetting, it will hit the limits of floating point precision after 6 minutes or so. So it needs to be reset.

[phasor~ (1 / array duration in seconds)] --> [*~ array frames] gives fractional sample indices, which of course does not accurately reflect sample-by-sample integration of DC = 1 -- therefore rounding error will eventually kill this approach.

Is there an external to do this?

My intention was to make a demo of recording into a circular buffer and catching event onsets with bonk~, saving the sample indices into a list to use with a buffer player. This presumes accurate knowledge of the buffer position being written to at any moment. That's the rationale for the question.

hjh

@ddw_music This thread...... https://forum.pdpatchrepo.info/topic/10972/distorted-sound-trying-to-simulate-a-bucket-brigade-delay might carry some useful information.
And [BD2.1] is worth poking about in anyway but for windows [ipoke~] is still only 32-bit I think, while for Linux there is a 64bit binary..
[ipoke~] takes an audio rate index which might help?......... ipoke_v.3_test1.zip
There could be an update somewhere.
David.

@whale-av

[ipoke~] takes an audio rate index which might help?

That does look interesting, thanks.

I think this would still need a signal-rate phasor that increments by integers, wouldn't it?

The following does actually work (except, if you try to reset it with the button at the top at the wrong time in the calculation block, the behavior is not consistent -- but it should be OK to loadbang it once and let it keep running). I confirmed in array1's list view that it does cycle through integers 0 to 99 repeatedly (as Phasor.ar(0, 1, 0, 100) would do) -- so a ten-second circular buffer would run 0 to 440999 (with [==~ 440936]).

In theory, if [tabwrite~] and this little patch are both ticking along in samples, they should never go out of sync. But I guess [ipoke~] might be even more certain.

but for windows [ipoke~] is still only 32-bit I think

Couldn't use it in my class, then.

IMO signal-rate writing into an array is a basic feature (kinda like signal-rate comparators are basic) -- I found it rather startling that it isn't available out of the box. What would it take for me to build a 64-bit binary for Windows?

hjh