$ echo "one
> two
> three" > foo
$ cat foo
one
two
three
$ tac foo
three
two
one
[download]

If you don't have tac on your system (which some of our AIX boxes don't), you can also use this short script,

#!/bin/sh

file=$1;
i=1
max=`wc -l $file | cut -d' ' -f1,1 `

while [ $i -le $max ];
do
        tail -$i $file | head -1
        i=$((i+1))
done
[download]

Call it much like tac,

$ ./reverse your.file > reverse.file
[download]

It's not as fast as tac obviously, but if you don't have 'tac' and want to reverse a file it does work.

---
s;;:<).>|\;\;_>?\\^0<|=!]=,|{\$/.'>|<?.|/"&?=#!>%\$|#/\$%{};;y;,'} -/:-@[-`{-};,'}`-{/" -;;s;;$_;see;
Warning: Any code posted by tuxz0r is untested, unless otherwise stated, and is used at your own risk.

my @input = <STDIN>;
my $i = $#input;
while ($i >= 0) {
  print $input[$i--];
}
[download]

As it's pure perl, I installed File::ReadBackwards locally and it's definitely the way to go - very impressive - only 90seconds to reverse the 180MB file and uses less than 5MB of memory to do it.

I diff'd the original .txt file with the .rev.rev file and no differences, so it seems to work ok!

problem solved - once again - many many thanks!

Probably it is not the fastest way, but Tie::File should work

use Tie::File;

my $file = shift @ARGV;

tie my @data, 'Tie::File', $file or die $!;
tie my @reversed, 'Tie::File', $file."rev" or die $!;

@reversed = reverse(@data);
untie @data;
untie @reversed;
[download]

Tie::File doesn't load the file into memory, it indexes it, so this should work even for very big files

Your reverse puts all 1 million lines in memory.

That's on top of the 2 million indexes you place in memory when only 1 million are needed.

Fix:

use Tie::File;

tie(my @data, 'Tie::File', $fname_in)
   or die("Unable to open \"$fname_in\": $!\n");

open(my $fh_out, '>', $fname_out)
   or die("Unable to create \"$fname_out\": $!\n");

for (my $i=@data; $i--; ) {
   print $fh_out $data[$i];
}

untie @data;
[download]

Although I recommend File::ReadBackwards.

use File::ReadBackwards qw( );

my $fh_in = File::ReadBackwards->new($fname_in)
   or die("Unable to open \"$fname_in\": $!\n");

open(my $fh_out, '>', $fname_out)
   or die("Unable to create \"$fname_out\": $!\n");

while (defined(my $line = $fh_in->readline())) {
   print $fh_out $line;
}
[download]

Unfortunately, this firm doesn't have File::ReadBackwards and it's a real pain getting approval for non-core modules!

I tried the Tie::File idea (thanks for that), but killed the job after 10mins of max'd out cpu (it was about half way through) - the file is only 180MB

There must be a faster way??