Maybe you can use one [netreceive] to serve available ports and others in [clone].

• The instance first ask the fixed port for a dynamic port number, then close the connection (and don't take care anymore of its output so if the disconnection is delayed and the server still push something, it's discarded)
• The instance open then another connection with the given dynamic port.

At server side the [netreceive]s get their ports numbers from the $1 which contain the clone instance number. You can start the port minimal number where you want with the -s argument of [clone].
The callback is sent back to the main patch throw the clone instance outlet.

Don't you simply look for abstraction instead of sub-patch?

Based on your previous patch:

It maybe can work in message space if we start with a [bang~] to trigger the computation until it fill a 64b array and then play.

EDIT : Yes it does

Without [expr] it should like this

You can even use a single [tabread4~] to play the array twice throw the multichannel support.
Patch from the [tabread4~] help with a small modification:

Another possible algorithm:

1. Fill a list with the samples, randomize it
2. Drop the samples one by one, use it and fill another list
3. When the first list is empty, switch with the second, randomize it and goto 2.

If you pick a random sample of the second list before the last of the first is consumed then you prevent repetition.

Hi,
You don't need the parent name if you use [iemguts/sendcanvas]. For an example see
https://forum.pdpatchrepo.info/topic/15118/tooltip-display-a-message-about-an-abstraction-in-its-parent

The gui ping when you change the array values by mouse:

FYI it opens with PlugData.

@willblackhurst you generate the combinations but the main subject is the permutations I think.

EDIT: my two cents

perm4.pd

FYI I've played with dynamic display here https://forum.pdpatchrepo.info/topic/15118/tooltip-display-a-message-about-an-abstraction-in-its-parent/1

The Euclidean definition of the functions div and mod https://dl.acm.org/doi/10.1145/128861.128862

it is unfortunate that the definition of these functions appears to be
handled rather casually in the computer science literature and in the design
of programming languages, as one might infer from various poor “defini-
tional engineering” decisions down to wrong or erroneous definitions, as in
the ISO Standard for Pascal [11, 13], Algol 68 [201], and some other languages.

In this paper we clarify the differences between the various definitions, in
particular those based on division by truncation (T-definition) and on division
by flooring (F-definition) as defined by Knuth [141]. We also propose still
another definition, which we call Euclidean because it is based on Euclid’s
theorem (E-definition).

April 1992…

1s. between input bangs

1s. between output bangs

@Moothart @oid's patch with a different arrangement. Patch style is very personal, there is no accounting for taste.

@freq63 said:

What I find confusing is the fact that all 4 values are printed without issue separately before unpack.

separately is the word. [unpack] get the values one (two) after each other so only its (two) first inlet(s) is (are) triggered.

The messages come one by one (or two by two) so the unpack never get a 4 items list.

You can accumulate the messages to 4 this way:

This is your triggerized patch:

You can see the right branch is triggered before the top-most [t f f] feeds the [pack] so the numbers are all generated by the loop before they was printed.
EDIT: you stack computation branch and they are released deepest to shallowest so numbers are reverted.

1. Do not split the control flow without [trigger]
2. Why don't you use [until] ?

I think it's easier to dynamically upgrade the variable name with the [list prepend] right inlet.

[netsend] help for the [send( message says "same as list" so you can also do

[hslider]
|
[list prepend z]
|
[netsend]