@jancsika: I'm no expert of RT audio but, first, as the number generation is (here) the result of a manually generated event, I cannot see why a few clock cycles would matter timing-wise ; second, the chance these solutions never return exist but are really low, eg: with random 5, not getting a new number out of 4 possibilities in 10 tries is already a 1ppm event! Pd is for me about livecoding, so I prefer to go with easy to code blocks and manually edit something in case of need. That's why I was perplexed by the complexity of the original solution.

@gsagostinho: Ok. Thanks! So I'll have to ask Johannes Kreidler what was intended in the solution he provided.

PS: New version of randiff.pd with a slight change to centralize initialization.

I'm new to Pd and going through the official tutorial. And I'm am overcome by the complexity of the proposed solution to the first exercise, "2.2.1.2.7 Random without repetitions".

More precisely, I'd like to know why such complexity? Why would my attached patch cause problems?

Thanks in advance for your insights