I got a few ideas from the code to which you linked. Thanks for that. Here's a modified MooselessRequireHookTest.pm:

package MooselessRequireHookTest;

use 5.032;
use warnings;

sub new {
    my ($class) = @_;

    return bless {} => $class;
}

sub dynamic_require {
    my ($self, $ns_extension) = @_;

    state $hook_source = {};
    state $inc_updated = 0;
    my $class = join '::', __PACKAGE__, $ns_extension;
    my $filename_key = $class =~ s{::}{/}gr . '.pm';

    unless (exists $hook_source->{$filename_key}) {
        my $source = <<~EOF;
            package $class;
            use parent 'MooselessRequireHookTest';
            1;
            EOF
        $hook_source->{$filename_key} = \$source;
    }

    state sub inc_hook {
        my ($coderef, $filename) = @_;
        warn "\$coderef[$coderef] \$filename[$filename]\n";

        return $hook_source->{$filename};
    };

    unless ($inc_updated) {
        push @INC, \&inc_hook;
        $inc_updated = 1;
    }

    eval "require $class;";

    return;
}

1;
[download]

I added this to the end of sscce_mooseless_require_hook_test.t:

warn '-' x 10, ' @INC ', '-' x 10, "\n";
warn "$_\n" for @INC;
warn '-' x 26, "\n";
[download]

Here's a new, sample prove run:

$ prove -v sscce_mooseless_require_hook_test.t
sscce_mooseless_require_hook_test.t ..
1..8
ok 1 - Test MooselessRequireHookTest::->new()
ok 2 - 'Test MooselessRequireHookTest::->new() ISA' isa 'MooselessRequ
+ireHookTest'
ok 3 - Test MooselessRequireHookTest::Test1->new()
ok 4 - 'Test MooselessRequireHookTest::Test1->new() ISA' isa 'Mooseles
+sRequireHookTest::Test1'
ok 5 - Test MooselessRequireHookTest::Test2->new()
ok 6 - 'Test MooselessRequireHookTest::Test2->new() ISA' isa 'Mooseles
+sRequireHookTest::Test2'
ok 7 - Test MooselessRequireHookTest::A::B::C->new()
ok 8 - 'Test MooselessRequireHookTest::A::B::C->new() ISA' isa 'Moosel
+essRequireHookTest::A::B::C'
$coderef[CODE(0x600867f00)] $filename[MooselessRequireHookTest/Test1.p
+m]
$coderef[CODE(0x600867f00)] $filename[MooselessRequireHookTest/Test2.p
+m]
$coderef[CODE(0x600867f00)] $filename[MooselessRequireHookTest/A/B/C.p
+m]
---------- @INC ----------
/home/ken/tmp/pm_sscce_require_hook/lib
/home/ken/perl5/perlbrew/perls/perl-5.32.0/lib/site_perl/5.32.0/cygwin
+-thread-multi
/home/ken/perl5/perlbrew/perls/perl-5.32.0/lib/site_perl/5.32.0
/home/ken/perl5/perlbrew/perls/perl-5.32.0/lib/5.32.0/cygwin-thread-mu
+lti
/home/ken/perl5/perlbrew/perls/perl-5.32.0/lib/5.32.0
CODE(0x600867f00)
--------------------------
ok
All tests successful.
Files=1, Tests=8,  0 wallclock secs ( 0.02 usr  0.01 sys +  0.08 cusr 
+ 0.05 csys =  0.15 CPU)
Result: PASS
[download]

So, the CODEREF is only run once for each invocation and @INC only has one hook. Compare that output with what I previously showed in "Re^6: require() @INC hooks problem [non-Moose]".

I put the changes (not the warn statements) into the original module: all (98) tests passed without any problems.

With regard to your suggestion of "Just eval the code and update %INC", I'm not following you in terms of how that might be implemented. Could you provide a code example?

I notice you're referencing the OP code. Please see the sub-thread starting at "Re^4: require() @INC hooks problem [non-Moose]". Moose was never an issue here but, given your request, and stated aversion to Moose, that provides a non-Moose version of the SSCCE code.

I was hoping for an answer to "Why is it calling the subroutine with all the previous CODEREF values then the newly created CODEREF?" (see "Re^2: require() @INC hooks problem").

> "Why is it calling the subroutine with all the previous CODEREF values then the newly created CODEREF?"

I don't know what that means.

> With regard to your suggestion of "Just eval the code and update %INC", I'm not following you in terms of how that might be implemented. Could you provide a code example?

something like (totally untested)

sub create_class {
    my ($class) = @_;
    my $source = <<~EOF;
            package $class;
            use parent 'MooselessRequireHookTest';
            1;
            EOF
    
    eval $source;
    
    my $name = $class =~ s(::)(\/)gr;
     
    $INC{"$name.pm"} = 'imported via eval';
}
[download]

Cheers Rolf
_{(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery}

here tested code

use strict;
use warnings;
use Data::Dump qw/pp dd/;


sub create_class {
    my ($class) = @_;

    my $source = <<"__CODE__";

    package $class;

    sub import {
        warn "$class imported";
    }

__CODE__


    eval $source;

    my $name = $class =~ s(::)(\/)gr;

    $INC{"$name.pm"} = 'imported via eval';
}


BEGIN {
    create_class('Jabba::Dabba');
}

pp \%INC;

use Jabba::Dabba;
[download]

<Reveal this spoiler or all in this thread>

Cheers Rolf
_{(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery}

I wrote my last reply at 3.13am (my time). I did see your response about 10 minutes later but my eyes were getting pretty blurry by then: I decided some sleep was in order before looking into that further. :-)

Twelve hours later, and much refreshed, I see you've posted a lot more code. Thanks for that — I'll look into it and see how that fits with my existing code.

— Ken