Hi All,
I have been working on this dynamic looping patch and it all seems to be working well except for one thing. I am dynamically creating abstractions to build infinite layers of looped material, but when each abstraction is created I need to go into that abstraction and disconnect one of the signal connections (any connection) and then reconnect it for the whole patch to work.

Why would this happen?

I have attached the patch with instructions and everything. Any theories/help/insight would be greatly appreciated.

Thanks!

Andrew

http://www.pdpatchrepo.info/hurleur/dynamicmetloop.pd

Sorry, I can't help you at all. I can't get it to work at all. And I get this error:
error: .xbab8e0: no such object

@glitchpop said:

Sorry, I can't help you at all. I can't get it to work at all. And I get this error:
error: .xbab8e0: no such object

I think that's the window name of one of the sub patches being created http://puredata.hurleur.com/sujet-4933-control-pos-other-window wonder why that would be happening?

The reason why it works the first few times is because you accidentally saved some instances of [metloop] inside [dynamicmetloop]. The recording is actually going into those, not the newly created ones.

The problem is that for whatever reason, Pd doesn't update it's dsp graph for the newly dynamically created abstraction. It does, however, cause the previously created abstraction to work. So, the trick to getting it to work right is to also create a dummy abstraction that contains dsp objects after you create an instance of [metloop]. I just added the line "$1 obj 10 30 hilbert~" to then end of your message box, and it works fine now.

Thanks Maelstrom, that did the trick.

So it just needed any dsp abstraction to be created, not necessarily [hilbert~], yes?

Right. I just mentioned [hilbert~] because it's included in Pd, so you could try it right away. But it's probably better to just make your own dummy abstraction with something stupid, like [sig~ 0], inside. [hilbert~], if I recall, has four [biquad~]s in it. If you're not using it, that's a wast of dsp.

bah! i have found a new glitch (thankfully unrelated to the first one, i think).

The full sampled input does not seeming to be written to the array if the sample is more than 10 seconds long. If the input is more than 10 seconds, it will play in full the first repeat (which is played by the delay line and then written to the array) and then only play about 10 seconds and cut off for every subsequent repeat (played from the array via [tabplay~])

also, Maelstrom, dsp objects other than [hilbert~] don't seem to work in initiating the audio, but thats not that big of a deal to me, as long as it works somehow.

I have attached the newest versions of the patch. any thoughts?

http://www.pdpatchrepo.info/hurleur/dynamiclooper.zip

false alarm! it was a rouge [send] that had a slightly different name. thanks for all your help, pd community!

@abbernstein said:

also, Maelstrom, dsp objects other than [hilbert~] don't seem to work in initiating the audio, but thats not that big of a deal to me, as long as it works somehow.

[hilbert~] is not a compiled object. It is an abstraction containing dsp objects. That's what you'll want create.

Got it.

It is now working more or less fine, except that after recording 2 or 3 loops there is major latency to the point of making the patch unusable. Is the patch really just that taxing? Do I just need to get some more ram? I have 1GB on a Macbook running OSX 10.5.8.

newest version here if anyone wants to test it out.

http://www.pdpatchrepo.info/hurleur/dynamiclooper_2.zip