You are right, I tried it without "pd-".
It work's perfectly. Thank you, David!

The funniest part is, PD complains "my-patch. pd: no such object", but the window is activated

Most of the Windows apps are active after I open them, but PD needs to be activated with a mouse for keyboard to become functional. This is very inconvenient if I have automated installation with no monitor. I need Keyboard buttons to control some parameters without monitor, at least turn a PD patch on/off with keyboard only.

Is there any solution, workaround or . . . ?

Thanks everyone!
At first glance [bag] seems promising, but I didn't find it useful, probably of "do nothing" when no values are stored.
Even if poly works I found Arrays more simple. Yes, Max or Sum of array does the same job - when "zero value" is detected it bangs. That's it!
Off_Bang.pd

I want to stop "note off" messages if at least one of the MIDI keys was not released. Or to put in other words I need to activate "note off" messages only if the key released was the only pressed. If there are at least two keys pressed, no "note off" messages should be sent. I think this is doable, because "note off" goes after "note on". So maybe I should count "note on" messages or. . .?

Interesting. .. it is not possible to connect the audio out to control in, but it is possible to SEND audio via throw~ to toggle inlet, and that gives me "no method for float" error.

I tried to send line~ from throw~ to catch~ but it gives me "catch~: no method for 'float'" is it normal behavior or a bug? And if normal, what I'm doing wrong? Line~ has audio connections so it is logical to think that signal can be sent via trew~/catch~ method?

@whale-av

1. readsf~ works fine with long files, no complaints, but yes, I sometimes want to scrub those long files, so. . . . this why I'm looking for arrays
2. I can't get rid of that nasty "... couldn't create" for iemlib, that is installed in standard path

@solipp

3. Pd version 0.49.0
4. In the documentation you referred: "it (soundfiler?) still reads the samples only to throw them away". I think this is why ~12 min array is only half filled, i.e ~16 million samples form ~32 million are visible.

As I understand soundfiler is pretty limited, so it would be interesting to load long audio into multiple arrays tabread~ can play afterwards. Is it doable? Maybe soundfiler can read part of audio file with start/end positions?

I have 100+ audio files associated with "open" message to be played by "readsf~". Filenames are formal so I can easily manage large amount of files by predefined names without any change in PD. So far so good, but sometimes I need to know exactly the content of the audio file I'm going to play. Interesting, is it possible to extract some meta data (or anything like that?) that would function like filenames?

P.S. I have tried to use colors and groups, but it is too vague.