use RTF::Writer;
  my $rtf = RTF::Writer->new_to_file("greetings.rtf");
  my $foo = \$rtf;
  ${$foo}->prolog();
  ${$foo}->paragraph(
    "Hi there!"
  );
  ${$foo}->close;
[download]

See Why it's stupid to use a variable as a variable name for why you are opening yourself up for pain. The above also passes strict.

In the end, however, these types of games generally raises the specter of an XY Problem. Why do you need a reference to the object? Is this a caching thing, which might be addressed with proper scoping?

I agree with kennethk, if you want to use symbolic references, there's a high probability that you are doing it wrong, or do not have a clear idea of what you are trying to do. But, if you know that you are asking for trouble, here is something that might help you :

our $value = "package var";
my $value = "lexical var";
my $ref = "value";
$, = ", ";
print ${$ref}, $value, eval "\$$ref";
[download]

package var, lexical var, lexical var

You should see there that you can only access dynamix variables with symbolic references, not lexicals.

Thanks for your reply! I've added a description of what I'm actually trying to accomplish below. I'm also going to spend more time reading some of the links you've suggested.

$rtf isn't a string. It's a RTF::Writer object created by the call to RTF::Writer->new_to_file(). Your problem isn't with references, so your search is bound to end in frustration. Perhaps you could explain what you're really trying to accomplish? Is it that you don't want to write output to a file? If so you can look at the RTF::Writer documentation which provides a way for you to send the output to a string instead with RTF::Writer->new_to_string(). Afterward, you can do whatever you like with the string.

Thanks for the replies! Since people have asked what I'm trying to do, I've included a copy of the actual code below.

I have another program that regex-searches lengthy documents and tries to sort the paragraphs into categories based on keywords. The other program colours the text of the various paragraphs based on what category it thinks it fits into, and then human assistants go over the coloured document to check that the categorization is correct. If it isn't, the human assistants change the text colour to the appropriate value manually. The files are in RTF format.

The purpose of the second program is to take the correctly-coloured document and split in into six new files (one for each category). The code I have written effectively (although inelegantly) identifies the text colour of each paragraph. I then want to write the text of the paragraph to the appropriate file, based on the text colour of the paragraph.

use diagnostics;
use File::Slurp;
use RTF::Writer;

my @destinations = ('IAs','Defs','Term','Payoffs','RWs','Boilerplate2'
+);
my $targetfolder = "contract/new2";

opendir(DIR, "$targetfolder");
my @files = readdir(DIR);
closedir(DIR);

my $count = 0;

foreach $file (@files) {
         if ($file =~ /rtf/i) {
              my $IAs = RTF::Writer->new_to_file("contract/IAs/$file")
+;
              my $Defs = RTF::Writer->new_to_file("contract/Defs/$file
+");
              my $Term = RTF::Writer->new_to_file("contract/Term/$file
+");
              my $Payoffs = RTF::Writer->new_to_file("contract/Payoffs
+/$file");
              my $RWs = RTF::Writer->new_to_file("contract/RWs/$file")
+;
              my $Boilerplate2 = RTF::Writer->new_to_file("contract/Bo
+ilerplate2/$file");
              my $contract = read_file("$targetfolder/$file");
              my @paragraphs = split('cf', $contract);
              foreach $clause (@paragraphs) {
                  my @lines = split("\n", "$clause");
                  if ($lines[0] =~ /^\d$/){
                      my $colour = $lines[0];
                      my $goal = \${$destinations["$colour"]};      #P
+ROBLEM CODE
                      my $destinationfolder = ("contract/$goal");
                      foreach $line (@lines) {
                               if ($line =~ /^\s*\w/ && $line !~ /^\d$
+/){
                                           ${$goal}->paragraph (
                                                    "$line",
                                           );
                               }
                      }
                  }

              }
         }
$count++;
}
[download]

Essentially, I want to fix the line marked "#PROBLEM CODE", above. I'm going to go over the replies and links above to see if I can figure out a solution. Thanks again!

You are quite close to a clean (IMHO) solution here. Rather than using symbolic references, I would just use a hash to store the writer references:

use diagnostics;
use File::Slurp;
use RTF::Writer;

my @destinations = ('IAs','Defs','Term','Payoffs','RWs','Boilerplate2'
+);
my $targetfolder = "contract/new2";

opendir(DIR, "$targetfolder");
my @files = readdir(DIR);
closedir(DIR);

my $count = 0;

foreach $file (@files) {
         if ($file =~ /rtf/i) {
              my %writer;
              for my $dest (@destinations) {
                   $writer{$dest} = RTF::Writer->new_to_file("contract
+/$dest/$file");
              }
              my $contract = read_file("$targetfolder/$file");
              my @paragraphs = split('cf', $contract);
              foreach $clause (@paragraphs) {
                  my @lines = split("\n", "$clause");
                  if ($lines[0] =~ /^\d$/){
                      my $colour = $lines[0];
                      my $goal = $writer{$destinations[$colour]};     
+ #NO-LONGER-PROBLEMATIC CODE
                      my $destinationfolder = ("contract/$destinations
+[$colour]"); # Also changed this one
                      foreach $line (@lines) {
                               if ($line =~ /^\s*\w/ && $line !~ /^\d$
+/){
                                           $goal->paragraph (
                                                    "$line",
                                           );
                               }
                      }
                  }

              }
         }
$count++;
}
[download]

Untested, though it passes strict. I think there are some simplifications you could make in the logic in your @lines loop as well.

Remember that objects are just a subclass of scalars; and consider perusing perlreftut and perllol.

---

#11929 First ask yourself `How would I do this without a computer?' Then have the computer do it the same way.

I had some trouble for a while with a bug relating to my first program, but I can now confirm that the above code works perfectly. Thanks very much!