Perl keeps the information where (file, line and package) a function was declared before exporting.

There multiple ways to introspect° it, I chose Devel::Peek out of laziness.

Introducing the line 123 in the code just before the redefinition happens will show the culprit.

  DB<121> use Cwd

  DB<122> use Devel::Peek

  DB<123> Dump \&getcwd if defined &getcwd
SV = IV(0x3458550) at 0x3458560
  REFCNT = 1
  FLAGS = (TEMP,ROK)
  RV = 0xdc79e0
  SV = PVCV(0xe02188) at 0xdc79e0
    REFCNT = 11
    FLAGS = (DYNFILE,HASEVAL)
    COMP_STASH = 0xd54290       "Cwd"                  # PACKAGE
    START = 0x2985788 ===> 1
    ROOT = 0x29ade90
    GVGV::GV = 0xdc7b48 "Cwd" :: "_win32_cwd"          # NAME
    FILE = "c:/Perl_524/lib/Cwd.pm"                    # FILE
    DEPTH = 0
    FLAGS = 0x5000
    OUTSIDE_SEQ = 633
    PADLIST = 0x29d5aa8
    PADNAME = 0x29bb378(0x29ca0a8) PAD = 0xdc79f8(0x29c8968)
       1. 0xdc7a28<1> (634,635) "$pwd"
    OUTSIDE = 0xd540e0 (UNIQUE)

  DB<124>
[download]

°) the B backend has methods for it, some modules offer wrapper for it:

• Sub::Identify
• Sub::Util

I simply thought that this problem might be a good way for me to get my feet wet with the debugger. I thought it might bring me close to perl's machine model.

Yes indeed. I needed to change that line with caller:

my $rvars = \%vars;
## this function is to be debugged using the perl debugger
my $return2 = make_initial_captions($rvars);
say "return2 is $return2";
[download]

Let's consider this. I know that we're encouraged to come up with an SSCCE, but this is not, nor can it easily be made so. It's an html templating system that has evolved according to my needs. I've tried to make a distribution of it, but I couldn't manage how to deal with filesystems on github. What would you use instead of Net::SFTP::Foreign?

This was my effort to be concise. I don't see how the use statements in subroutines affects other modules:

Update: put module source between readmore tags

Question: what use statements can be removed from everything I've posted?

This routine has a breakpoint now:

sub make_initial_captions {
  use 5.016;
  use warnings;
  use POSIX;
  use Path::Tiny;
  use Encode;
  use open OUT => ':encoding(UTF-8)', ':std';
  use Data::Dumper;

  my $rvars = shift;
  my %vars  = %$rvars;

  #print Dumper $rvars;
  
  my $image_path   = $vars{"to_images"};
  my $caption_path = $vars{"eng_captions"};
  # put a break point here
  $DB::single = 1;
  return "nothing yet";
}
[download]

, and, I got to see some values at this mark:

$ perl -d 1.debug.11.pl

Loading DB routines from perl5db.pl version 1.55
Editor support available.

Enter h or 'h h' for help, or 'man perldebug' for more help.

Subroutine debug1::getcwd redefined at /usr/share/perl/5.30/Exporter.p
+m line 66.
 at template_stuff/debug1.pm line 400.
    debug1::BEGIN() called at template_stuff/debug1.pm line 400
    eval {...} called at template_stuff/debug1.pm line 400
    require debug1.pm called at 1.debug.11.pl line 4
    main::BEGIN() called at template_stuff/debug1.pm line 400
    eval {...} called at template_stuff/debug1.pm line 400
main::(1.debug.11.pl:12):    my $ts          = "template_stuff";
  DB<1> n                                                             
+          
main::(1.debug.11.pl:13):    my $images      = "aimages";
  DB<1> c                                                             
+          
title is 1.debug.1
path1 is /home/hogan/6.scripts/1.debug.1
abs is /home/hogan/6.scripts/1.debug.1/1.debug.11.pl
debug1::make_initial_captions(template_stuff/debug1.pm:43):
43:      return "nothing yet";
                                                          
...
  DB<3> p $vars{to_images}                                            
+          
/home/hogan/6.scripts/1.debug.1/template_stuff/aimages
  DB<4> p $vars{eng_captions}                                         
+          
/home/hogan/6.scripts/1.debug.1/template_stuff/captions
  DB<5> c                                                             
+          
return2 is nothing yet
ini path is /home/hogan/Documents/html_template_data/6.values.ini
...[normal execution]
return is 1.debug.13.html
http://www.merrillpjensen.com/perlmonks/1.debug.13.html
Debugged program terminated.  Use q to quit or R to restart,
use o inhibit_exit to avoid stopping after program termination,
h q, h R or h o to get additional info.
  DB<5> q                                                             
+          
$
[download]

It then executed to the end and behaved.

Let's consider this. I know that we're encouraged to come up with an SSCCE, but this is not, nor can it easily be made so. It's an html templating system that has evolved according to my needs.

Your questions were not about HTML templating systems, they were about manipulating lists of filenames. It would have been easy enough for you to construct an SSCCE which demonstrated a single, isolated problem. If your SSCCE did not exhibit the same problem as your larger code then the problem is somewhere else in your larger code and you need to determine that first.

Question: what use statements can be removed from everything I've posted?

Let's just look at this one package. You have use 5.010; and  use 5.011; in package utils1; - the former is quite obviously superfluous and should be removed (it does nothing and if read in isolation is potentially misleading).

In some of the subroutines you use strict, in others warnings and in still others both (or neither). This inconsistency is perplexing to say the least. Far better to use both throughout: either within your package or in the scope enclosing it.

In subroutine invert_aoa you use 5.010; but there is apparently nothing in this sub which requires that version. What is the statement doing here? Remove this.

In subroutine highest_number you use both File::Basename and Cwd but then do absolutely nothing with them. Remove these.

In subroutines print_aoa and print_hash you use 5.011; but there is apparently nothing in these subs which requires that version. What is the statement doing here? Remove this.

Other things which aren't helping you: inconsistent indenting, use of the special variables $a and $b as lexicals, using a package name with all lowercase when it isn't a pragma and exporting everything.

If you had created an SSCCE then it is probable that some or perhaps even all of these issues would not appear in it. That would make the work of everyone reading the SSCCE much easier and therefore make them (ie. me) much more likely to offer help. There are therefore 2 benefits. Firstly the construction of the SSCCE helps you to hone down the problem better and in doing so may even solve it for you. Secondly, it allows others to see the actual problem you are having without wading through hundreds of lines of unnecessary code.

---

🦛

I simply thought that this problem might be a good way for me to get my feet wet with the debugger.

Fair enough. But then - I couldn't find a question related to how to use the debugger in your post.

I thought it might bring me close to perl's machine model.

Oh, I don't think the debugger will help here. The Perl debugger may help to understand Perl code you're currently maintaining, or maybe understand the interaction with a CPAN module in cases where its documentation isn't clear enough - but it doesn't go into Perl's guts.

I know that we're encouraged to come up with an SSCCE, but this is not, nor can it easily be made so.

Then, why show the code? What do you expect us to do with it?

Question: what use statements can be removed from everything I've posted?

That question has already been answered by other monks, so let me add just a general remark: It is of little use to have use statements within a subroutine. Typically you see them near the top of the file so that it is easy to spot the dependencies of your code.

This routine has a breakpoint now:

Good - that's one of the methods to get a breakpoint into the source. Since you've already read the docs about the Perl debugger, you should be aware that you could also have done it like that:

$ perl -d 1.debug.11.pl
DB<1>f debug1.pm
DB<2>b 42
[download]

You could, of course, also b debug1::make_initial_captions and then step to line 42.