MCE::Relay(3)

📦

version 1.902 · 11 versions

MCE::Relay(3) — User Contributed Perl Documentation 🫶

okay bestie, welcome to the MCE::Relay(3) manual page. This is part 1, and we're diving into relay capabilities for Many-Core Engine. No cap. ✨

✨ NAME ✨

MCE::Relay — Extends Many-Core Engine with relay capabilities 💖

🎀 VERSION 🎀

This document describes MCE::Relay version 1.902. 💅

🌸 SYNOPSIS 🌸

Here's the vibe, bestie. 👇

use MCE::Flow;

my $file = shift || \*STDIN;

## Line Count #######################################

mce_flow_f {
   max_workers => 4,
   use_slurpio => 1,
   init_relay  => 0,
},
sub {
   my ($mce, $slurp_ref, $chunk_id) = @_;
   my $line_count = ($$slurp_ref =~ tr/\n//);

   ## Receive and pass on updated information.
   my $lines_read = MCE::relay { $_ += $line_count };

}, $file;

my $total_lines = MCE->relay_final;

print {*STDERR} "$total_lines\n";

## Orderly Action ###################################

$| = 1; # Important, must flush output immediately.

mce_flow_f {
   max_workers => 2,
   use_slurpio => 1,
   init_relay  => 0,
},
sub {
   my ($mce, $slurp_ref, $chunk_id) = @_;

   ## The relay value is relayed and remains 0.
   ## Writes to STDOUT orderly.

   MCE->relay_lock;
   print $$slurp_ref;
   MCE->relay_unlock;

}, $file;

🦋 DESCRIPTION 🦋

This module enables workers to receive and pass on information orderly with zero involvement by the manager process while running. The module is loaded automatically when MCE option init_relay is specified.

All workers (belonging to task_id 0) must participate when relaying data.

Important: Relaying is not meant for passing big data. The last worker will stall if exceeding the buffer size for the socket. Not exceeding 16 KiB - 7 is safe across all platforms. 😭

👑 API DOCUMENTATION 👑

MCE::relay { code }
mce_relay { code } since 1.882
MCE->relay ( sub { code } )
$mce->relay ( sub { code } )

Relay is enabled by defining the init_relay option which takes a hash or array reference, or a scalar value. Relaying is orderly and driven by chunk_id when processing data, otherwise task_wid. Omitting the code block (e.g. MCE::relay) relays forward.

Below, relaying multiple values via a HASH reference. 🍓

use MCE::Flow max_workers => 4;

mce_flow {
   init_relay => { p => 0, e => 0 },
},
sub {
   my $wid = MCE->wid;
   my $pass = $wid % 3;  # simulate work
   my $errs = $wid % 2;

   ## relay (include the trailing semicolon)

   my %last_rpt = MCE::relay { $_->{p} += $pass; $_->{e} += $errs };

   MCE->print("$wid: passed $pass, errors $errs\n");

   return;
};

my %results = MCE->relay_final;

print "   passed $results{p}, errors $results{e} final\n\n";

-- Output

1: passed 1, errors 1
2: passed 2, errors 0
3: passed 0, errors 1
4: passed 1, errors 0
   passed 4, errors 2 final

Or multiple values via an ARRAY reference. 🦋

use MCE::Flow max_workers => 4;

mce_flow {
   init_relay => [ 0, 0 ],
},
sub {
   my $wid = MCE->wid;

   ## do work
   my $pass = $wid % 3;
   my $errs = $wid % 2;

   ## relay
   my @last_rpt = MCE::relay { $_->[0] += $pass; $_->[1] += $errs };

   MCE->print("$wid: passed $pass, errors $errs\n");

   return;
};

my ($pass, $errs) = MCE->relay_final;

print "   passed $pass, errors $errs final\n\n";

-- Output

1: passed 1, errors 1
2: passed 2, errors 0
3: passed 0, errors 1
4: passed 1, errors 0
   passed 4, errors 2 final

Or simply a scalar value. 🌱

use MCE::Flow max_workers => 4;

mce_flow {
   init_relay => 0,
},
sub {
   my $wid = MCE->wid;

   ## do work
   my $bytes_read = 1000 + ((MCE->wid % 3) * 3);

   ## relay
   my $last_offset = MCE::relay { $_ += $bytes_read };

   ## output
   MCE->print("$wid: $bytes_read\n");

   return;
};

my $total = MCE->relay_final;

print "   $total size\n\n";

-- Output

1: 1003
2: 1006
3: 1000
4: 1003
   4012 size

MCE->relay_final ( void )
$mce->relay_final ( void )

Call this method to obtain the final relay value(s) after running. See included example findnull.pl for another use case. 🔍

use MCE max_workers => 4;

my $mce = MCE->new(
   init_relay => [ 0, 100 ],       ## initial values (two counters)

   user_func => sub {
      my ($mce) = @_;

      ## do work
      my ($acc1, $acc2) = (10, 20);

      ## relay to next worker
      MCE::relay { $_->[0] += $acc1; $_->[1] += $acc2 };

      return;
   }
)->run;

my ($cnt1, $cnt2) = $mce->relay_final;

print "$cnt1 : $cnt2\n";

-- Output

40 : 180

MCE->relay_recv ( void )
$mce->relay_recv ( void )

Call this method to obtain the next relay value before relaying. This allows serial-code to be processed orderly between workers. The following is a parallel demonstration for the fasta-benchmark on the web. 🧬

# perl fasta.pl 25000000

# The Computer Language Benchmarks game
# https://benchmarksgame-team.pages.debian.net/benchmarksgame/
#
# contributed by Barry Walsh
# port of fasta.rb #6
#
# MCE::Flow version by Mario Roy
# requires MCE 1.807+
# requires MCE::Shared 1.806+

use strict;
use warnings;
use feature 'say';

use MCE::Flow;
use MCE::Shared;
use MCE::Candy;

use constant IM => 139968;
use constant IA => 3877;
use constant IC => 29573;

my $LAST = MCE::Shared->scalar( 42 );

my $alu =
   'GGCCGGGCGCGGTGGCTCACGCCTGTAATCCCAGCACTTTGG' .
   'GAGGCCGAGGCGGGCGGATCACCTGAGGTCAGGAGTTCGAGA' .
   'CCAGCCTGGCCAACATGGTGAAACCCCGTCTCTACTAAAAAT' .
   'ACAAAAATTAGCCGGGCGTGGTGGCGCGCGCCTGTAATCCCA' .
   'GCTACTCGGGAGGCTGAGGCAGGAGAATCGCTTGAACCCGGG' .
   'AGGCGGAGGTTGCAGTGAGCCGAGATCGCGCCACTGCACTCC' .
   'AGCCTGGGCGACAGAGCGAGACTCCGTCTCAAAAA';

my $iub = [
   [ 'a', 0.27 ], [ 'c', 0.12 ], [ 'g', 0.12 ],
   [ 't', 0.27 ], [ 'B', 0.02 ], [ 'D', 0.02 ],
   [ 'H', 0.02 ], [ 'K', 0.02 ], [ 'M', 0.02 ],
   [ 'N', 0.02 ], [ 'R', 0.02 ], [ 'S', 0.02 ],
   [ 'V', 0.02 ], [ 'W', 0.02 ], [ 'Y', 0.02 ]
];

my $homosapiens = [
   [ 'a', 0.3029549426680 ],
   [ 'c', 0.1979883004921 ],
   [ 'g', 0.1975473066391 ],
   [ 't', 0.3015094502008 ]
];

sub make_repeat_fasta {
   my ( $src, $n ) = @_;
   my $width = qr/(.{1,60})/;
   my $l     = length $src;
   my $s     = $src x ( ($n / $l) + 1 );
   substr( $s, $n, $l ) = '';

   while ( $s =~ m/$width/g ) { say $1 }
}

sub make_random_fasta {
   my ( $table, $n ) = @_;
   my $rand   = undef;
   my $width  = 60;
   my $prob   = 0.0;
   my $output = '';
   my ( $c1, $c2, $last );

   $_->[1] = ( $prob += $_->[1] ) for @$table;

   $c1  = '$rand = ( $last = ( $last * IA + IC ) % IM ) / IM;';
   $c1 .= "\$output .= '$_->[0]', next if $_->[1] > \$rand;\n" for @$table;

   my $seq = MCE::Shared->sequence(
      { chunk_size => 2000, bounds_only => 1 },
      1, $n / $width
   );

   my $code1 = q{
      while ( 1 ) {
         # --------------------------------------------
         # Process code orderly between workers.
         # --------------------------------------------

         my $chunk_id = MCE->relay_recv;
         my ( $begin, $end ) = $seq->next;

         MCE->relay, last if ( !defined $begin );

         my $last = $LAST->get;
         my $temp = $last;

         # Pre-compute $LAST value for the next worker
         for ( 1 .. ( $end - $begin + 1 ) * $width ) {
            $temp = ( $temp * IA + IC ) % IM;
         }

         $LAST->set( $temp );

         # Increment chunk_id value
         MCE->relay( sub { $_ += 1 } );

         # --------------------------------------------
         # Also run code in parallel between workers.
         # --------------------------------------------

         for ( $begin .. $end ) {
            for ( 1 .. $width ) { !C! }
            $output .= "\n";
         }

         # --------------------------------------------
         # Display orderly.
         # --------------------------------------------

         MCE->gather( $chunk_id, $output );

         $output = '';
      }
   };

   $code1 =~ s/!C!/$c1/g;

   MCE::Flow->init(
      max_workers => 4, ## MCE::Util->get_ncpu || 4,
      gather      => MCE::Candy::out_iter_fh( \*STDOUT ),
      init_relay  => 1,
      use_threads => 0,
   );

   MCE::Flow->run( sub { eval $code1 } );
   MCE::Flow->finish;

   $last = $LAST->get;

   $c2  = '$rand = ( $last = ( $last * IA + IC ) % IM ) / IM;';
   $c2 .= "print('$_->[0]'), next if $_->[1] > \$rand;\n" for @$table;

   my $code2 = q{
      if ( $n % $width != 0 ) {
         for ( 1 .. $n % $width ) { !C! }
         print "\n";
      }
   };

   $code2 =~ s/!C!/$c2/g;
   eval $code2;

   $LAST->set( $last );
}

my $n = $ARGV[0] || 27;

say ">ONE Homo sapiens alu";
make_repeat_fasta( $alu, $n * 2 );

say ">TWO IUB ambiguity codes";
make_random_fasta( $iub, $n * 3 );

say ">THREE Homo sapiens frequency";
make_random_fasta( $homosapiens, $n * 5 );

MCE->relay_lock ( void )
MCE->relay_unlock ( void )
$mce->relay_lock ( void )
$mce->relay_unlock ( void )

The relay_lock and relay_unlock methods, added to MCE 1.807, are aliases for relay_recv and relay respectively. Together, they allow one to perform an exclusive action prior to actual relaying of data.

Relaying is driven by chunk_id or task_wid when not processing input, as seen here. 👇

MCE->new(
   max_workers => 8,
   init_relay => 0,
   user_func => sub {
      MCE->relay_lock;
      MCE->say("wid: ", MCE->task_wid);
      MCE->relay_unlock( sub {
         $_ += 2;
      });
   }
)->run;

MCE->say("sum: ", MCE->relay_final);

__END__

wid: 1
wid: 2
wid: 3
wid: 4
wid: 5
wid: 6
wid: 7
wid: 8
sum: 16

Described above, relay takes a code block and combines relay_lock and relay_unlock into a single call. To make this more interesting, I define init_relay to a hash containing two key-value pairs.

okay bestie, watch this actually work — here's the whole little eight-worker glow-up, count and total getting relayed and locked like the icons they are:

MCE->new(
   max_workers => 8,
   init_relay => { count => 0, total => 0 },
   user_func => sub {
      MCE->relay_lock;
      MCE->say("wid: ", MCE->task_wid);
      MCE->relay_unlock( sub {
         $_->{count} += 1;
         $_->{total} += 2;
      });
   }
)->run;

my %results = MCE->relay_final;

MCE->say("count: ", $results{count});
MCE->say("total: ", $results{total});

__END__

wid: 1
wid: 2
wid: 3
wid: 4
wid: 5
wid: 6
wid: 7
wid: 8
count: 8
total: 16

see how it printed wid: 1 through wid: 8 all neat and in order? that's the lock doing its job 💅 and then MCE->relay_final scoops up the finished count and total at the very end, 8 and 16, no cap 🧚

Below, "user_func" is lifted straight out of the "cat.pl" MCE example. and here's the lil plot twist: the count only gets incremented when you pass the -n switch to the script. Otherwise the output is just displaced in an orderly way — and fr fr, you don't even need to update the $_ value if exclusive locking is literally all you're after:

user_func => sub {
   my ($mce, $chunk_ref, $chunk_id) = @_;

   if ($n_flag) {
      ## Relays the total lines read.

      my $output = ''; my $line_count = ($$chunk_ref =~ tr/\n//);
      my $lines_read = MCE::relay { $_ += $line_count };

      open my $fh, '<', $chunk_ref;
      $output .= sprintf "%6d\t%s", ++$lines_read, $_ while (<$fh>);
      close $fh;

      $output .= ":$chunk_id";
      MCE->do('display_chunk', $output);
   }
   else {
      ## The following is another way to have ordered output. Workers
      ## write directly to STDOUT exclusively without any involvement
      ## from the manager process. The statement(s) between relay_lock
      ## and relay_unlock run serially and most important orderly.

      MCE->relay_lock;      # alias for MCE->relay_recv
      print $$chunk_ref;    # ensure $| = 1 in script
      MCE->relay_unlock;    # alias for MCE->relay
   }

   return;
}

not me pointing at that # ensure $| = 1 in script comment — yes, do that. the workers are writing straight to STDOUT with zero manager involvement, so anything between relay_lock and relay_unlock runs serially and, most importantly, orderly. no cap.

and now, the main event ✨ here's a variant of the fasta-benchmark demonstration from above — except in this one the workers write exclusively and orderly to "STDOUT". it's giving iconic, it's giving benchmark:

# perl fasta.pl 25000000

# The Computer Language Benchmarks game
# https://benchmarksgame-team.pages.debian.net/benchmarksgame/
#
# contributed by Barry Walsh
# port of fasta.rb #6
#
# MCE::Flow version by Mario Roy
# requires MCE 1.807+
# requires MCE::Shared 1.806+

use strict;
use warnings;
use feature 'say';

use MCE::Flow;
use MCE::Shared;

use constant IM => 139968;
use constant IA => 3877;
use constant IC => 29573;

my $LAST = MCE::Shared->scalar( 42 );

my $alu =
   'GGCCGGGCGCGGTGGCTCACGCCTGTAATCCCAGCACTTTGG' .
   'GAGGCCGAGGCGGGCGGATCACCTGAGGTCAGGAGTTCGAGA' .
   'CCAGCCTGGCCAACATGGTGAAACCCCGTCTCTACTAAAAAT' .
   'ACAAAAATTAGCCGGGCGTGGTGGCGCGCGCCTGTAATCCCA' .
   'GCTACTCGGGAGGCTGAGGCAGGAGAATCGCTTGAACCCGGG' .
   'AGGCGGAGGTTGCAGTGAGCCGAGATCGCGCCACTGCACTCC' .
   'AGCCTGGGCGACAGAGCGAGACTCCGTCTCAAAAA';

my $iub = [
   [ 'a', 0.27 ], [ 'c', 0.12 ], [ 'g', 0.12 ],
   [ 't', 0.27 ], [ 'B', 0.02 ], [ 'D', 0.02 ],
   [ 'H', 0.02 ], [ 'K', 0.02 ], [ 'M', 0.02 ],
   [ 'N', 0.02 ], [ 'R', 0.02 ], [ 'S', 0.02 ],
   [ 'V', 0.02 ], [ 'W', 0.02 ], [ 'Y', 0.02 ]
];

my $homosapiens = [
   [ 'a', 0.3029549426680 ],
   [ 'c', 0.1979883004921 ],
   [ 'g', 0.1975473066391 ],
   [ 't', 0.3015094502008 ]
];

sub make_repeat_fasta {
   my ( $src, $n ) = @_;
   my $width = qr/(.{1,60})/;
   my $l     = length $src;
   my $s     = $src x ( ($n / $l) + 1 );
   substr( $s, $n, $l ) = '';

   while ( $s =~ m/$width/g ) { say $1 }
}

sub make_random_fasta {
   my ( $table, $n ) = @_;
   my $rand   = undef;
   my $width  = 60;
   my $prob   = 0.0;
   my $output = '';
   my ( $c1, $c2, $last );

   $_->[1] = ( $prob += $_->[1] ) for @$table;

   $c1  = '$rand = ( $last = ( $last * IA + IC ) % IM ) / IM;';
   $c1 .= "\$output .= '$_->[0]', next if $_->[1] > \$rand;\n" for @$table;

   my $seq = MCE::Shared->sequence(
      { chunk_size => 2000, bounds_only => 1 },
      1, $n / $width
   );

   my $code1 = q{
      $| = 1; # Important, must flush output immediately.

      while ( 1 ) {
         # --------------------------------------------
         # Process code orderly between workers.
         # --------------------------------------------

         MCE->relay_lock;

         my ( $begin, $end ) = $seq->next;
         print( $output ), $output = '' if ( length $output );

         MCE->relay_unlock, last if ( !defined $begin );

         my $last = $LAST->get;
         my $temp = $last;

         # Pre-compute $LAST value for the next worker
         for ( 1 .. ( $end - $begin + 1 ) * $width ) {
            $temp = ( $temp * IA + IC ) % IM;
         }

         $LAST->set( $temp );

         MCE->relay_unlock;

         # --------------------------------------------
         # Also run code in parallel.
         # --------------------------------------------

         for ( $begin .. $end ) {
            for ( 1 .. $width ) { !C! }
            $output .= "\n";
         }
      }
   };

   $code1 =~ s/!C!/$c1/g;

   MCE::Flow->init(
      max_workers => 4, ## MCE::Util->get_ncpu || 4,
      init_relay  => 0,
      use_threads => 0,
   );

   MCE::Flow->run( sub { eval $code1 } );
   MCE::Flow->finish;

   $last = $LAST->get;

   $c2  = '$rand = ( $last = ( $last * IA + IC ) % IM ) / IM;';
   $c2 .= "print('$_->[0]'), next if $_->[1] > \$rand;\n" for @$table;

   my $code2 = q{
      if ( $n % $width != 0 ) {
         for ( 1 .. $n % $width ) { !C! }
         print "\n";
      }
   };

   $code2 =~ s/!C!/$c2/g;
   eval $code2;

   $LAST->set( $last );
}

my $n = $ARGV[0] || 27;

say ">ONE Homo sapiens alu";
make_repeat_fasta( $alu, $n * 2 );

say ">TWO IUB ambiguity codes";
make_random_fasta( $iub, $n * 3 );

say ">THREE Homo sapiens frequency";
make_random_fasta( $homosapiens, $n * 5 );

the way that !C! is just a placeholder that gets swapped in with s/!C!/$c1/g and s/!C!/$c2/g — lowkey genius, highkey iconic. the MCE->relay_lock / MCE->relay_unlock sandwich around the sequence grab is what keeps the ordering correct while everything else runs in parallel. main character energy fr 🔥

💖 GATHER AND RELAY DEMONSTRATIONS 💖

ok so bestie, this is the part where it gets real. John Martel slid into the metaphorical DMs and was like "hey, can you process a large flat file and expand each record into a whole bunch of records by splitting out the items in field 4, which are delimited by semicolons?" and every row in the output gets a unique ID starting with one while preserving output order. no cap, that's the whole brief. 💅

🌸 Input File, possibly larger than 500 GiB in size

like, genuinely huge. we're talking 500 GiB and up. not a vibe check you want to fail.

foo|field2|field3|item1;item2;item3;item4;itemN|field5|field6|field7
bar|field2|field3|item1;item2;item3;item4;itemN|field5|field6|field7
baz|field2|field3|item1;item2;item3;item4;itemN|field5|field6|field7
...

🦋 Output File

and here's what comes out the other side — all numbered, all ordered, all iconic:

000000000000001|item1|foo|field2|field3|field5|field6|field7
000000000000002|item2|foo|field2|field3|field5|field6|field7
000000000000003|item3|foo|field2|field3|field5|field6|field7
000000000000004|item4|foo|field2|field3|field5|field6|field7
000000000000005|itemN|foo|field2|field3|field5|field6|field7
000000000000006|item1|bar|field2|field3|field5|field6|field7
000000000000007|item2|bar|field2|field3|field5|field6|field7
000000000000008|item3|bar|field2|field3|field5|field6|field7
000000000000009|item4|bar|field2|field3|field5|field6|field7
000000000000010|itemN|bar|field2|field3|field5|field6|field7
000000000000011|item1|baz|field2|field3|field5|field6|field7
000000000000012|item2|baz|field2|field3|field5|field6|field7
000000000000013|item3|baz|field2|field3|field5|field6|field7
000000000000014|item4|baz|field2|field3|field5|field6|field7
000000000000015|itemN|baz|field2|field3|field5|field6|field7
...

✨ Example One ✨

so this one sets up a custom function to preserve output order. sounds cute right? unfortunately, the sprintf function by itself eats up extra CPU time, which means the manager process can end up falling behind. and when the manager falls behind, the workers may just be sitting there idling while they wait for the manager process to respond to the gather request. it's giving "group project where one person does everything." 😭

 use strict;
 use warnings;

 use MCE::Loop;

 my $infile  = shift or die "Usage: $0 infile\n";
 my $newfile = 'output.dat';

 open my $fh_out, '>', $newfile or die "open error $newfile: $!\n";

 sub preserve_order {
     my ($fh) = @_;
     my ($order_id, $start_idx, $idx, %tmp) = (1, 1);

     return sub {
         my ($chunk_id, $aref) = @_;
         $tmp{ $chunk_id } = $aref;

         while ( my $aref = delete $tmp{ $order_id } ) {
             foreach my $line ( @{ $aref } ) {
                 $idx = sprintf "%015d", $start_idx++;
                 print $fh $idx, $line;
             }
             $order_id++;
         }
     }
 }

 MCE::Loop->init(
     chunk_size => 'auto', max_workers => 3,
     gather => preserve_order($fh_out)
 );

 mce_loop_f {
     my ($mce, $chunk_ref, $chunk_id) = @_;
     my @buf;

     foreach my $line (@{ $chunk_ref }) {
         $line =~ s/\r//g; chomp $line;

         my ($f1,$f2,$f3,$items,$f5,$f6,$f7) = split /\|/, $line;
         my @items_array = split /;/, $items;

         foreach my $item (@items_array) {
             push @buf, "|$item|$f1|$f2|$f3|$f5|$f6|$f7\n";
         }
     }

     MCE->gather($chunk_id, \@buf);

 } $infile;

 MCE::Loop->finish();
 close $fh_out;

💜 Example Two 💜

ok this one is the glow-up. here the workers grab the current ID value and increment/relay it for the next worker — and behind the scenes it's all ordered by chunk ID. the workers call sprintf in parallel, which lets the manager process (out_iter_fh) handle up to 32 workers without falling behind. main character energy fr fr. 👑

Relay accounts for the worker handling the next chunk_id value. Therefore, do not call relay more than once per chunk. Doing so will cause IPC to stall. like, do NOT do that. one relay per chunk. periodt. 💀

 use strict;
 use warnings;

 use MCE::Loop;
 use MCE::Candy;

 my $infile  = shift or die "Usage: $0 infile\n";
 my $newfile = 'output.dat';

 open my $fh_out, '>', $newfile or die "open error $newfile: $!\n";

 MCE::Loop->init(
     chunk_size => 'auto', max_workers => 8,
     gather => MCE::Candy::out_iter_fh($fh_out),
     init_relay => 1
 );

 mce_loop_f {
     my ($mce, $chunk_ref, $chunk_id) = @_;
     my @lines;

     foreach my $line (@{ $chunk_ref }) {
         $line =~ s/\r//g; chomp $line;

         my ($f1,$f2,$f3,$items,$f5,$f6,$f7) = split /\|/, $line;
         my @items_array = split /;/, $items;

         foreach my $item (@items_array) {
             push @lines, "$item|$f1|$f2|$f3|$f5|$f6|$f7\n";
         }
     }

     my $idx = MCE::relay { $_ += scalar @lines };
     my $buf = '';

     foreach my $line ( @lines ) {
         $buf .= sprintf "%015d|%s", $idx++, $line
     }

     MCE->gather($chunk_id, $buf);

 } $infile;

 MCE::Loop->finish();
 close $fh_out;

🌈 INDEX 🌈

MCE, MCE::Core

🎀 AUTHOR 🎀

Mario E. Roy, <marioeroy AT gmail DOT com> — absolute legend, we stan. 🫶

and that's a wrap bestie!! this whole page is from perl v5.42.0, dated 2025-09-10, and it's MCE::Relay(3). go forth and relay responsibly, one chunk at a time. 💖✨🦋

📦 where this page comes from

fedora 43: perl-MCE 1.902-1.fc43 · GPL-1.0-or-later OR Artistic-1.0-Perl
💖 girlified by deepseek-ai/DeepSeek-V4.1-Flash on 2026-10-14: a language model's rewrite of the original, under its terms