That was surprising. It appears that Text::Balanced alters some of the magic innards of the variable. I changed your code a bit:

sub eteb {  
  my $data = shift;
  my $orig = $data;
  my @array = extract_multiple(
    $data,  
    [ sub{extract_tagged($_[0], 'a', 'b', undef,)}, ],
    undef,  
    1       
  );        
  print "data='$data'\n";
  display('eteb1', @array);
  @array = extract_multiple(
    $orig,                                                            
+                  
    [ sub{extract_bracketed($_[0], '()')}, ],
    undef,  
    1       
  );        
  display('eteb2', @array);
}
[download]

And get the desired results. The funny thing is, I was expecting that $data would be empty after the call or something, but was surprised to see that the value looked unchanged. I then changed the second extract_multiple to:

  @array = extract_multiple(
    $data."",                                                         
+                     
    [ sub{extract_bracketed($_[0], '()')}, ],
    undef,  
    1       
  );
[download]

and it worked as you expect. I haven't read the Text::Balanced docs to see if it's expected behaviour or not. But if it isn't, you may want to file a bug report on it.

Update: I remember a module (Devel::Peek) that lets you look at the magic goo inside of variables, so I changed your program to look at the $data variable before and after the call:

sub eteb {   
  my $data = shift;
  my $orig = $data;
  Dump($data);
  my @array = extract_multiple(
    $data,   
    [ sub{extract_tagged($_[0], 'a', 'b', undef,)}, ],
    undef,
    1
  );
  Dump($data);
  $data = $data."";
  Dump($data);
  print "data='$data'\n";                                             
+                                  
  display('eteb1', @array);
  @array = extract_multiple(
    $data,
    [ sub{extract_bracketed($_[0], '()')}, ],
    undef,
    1
  );
  display('eteb2', @array);
} 
<c>
<p>And sure enough, some stuff inside changed:</p>
<c>
$ perl 1005372.pl
et:a)(b
eb:(a)(b)
SV = PV(0x8458478) at 0x84fe160
  REFCNT = 1
  FLAGS = (PADMY,POK,pPOK)
  PV = 0x83c47b8 "(a)(b)"\0
  CUR = 6
  LEN = 8
SV = PVMG(0x83ef3a8) at 0x84fe160
  REFCNT = 7
  FLAGS = (PADMY,SMG,POK,pPOK)
  IV = 0
  NV = 0
  PV = 0x83c47b8 "(a)(b)"\0
  CUR = 6
  LEN = 8
  MAGIC = 0x83c2ec0
    MG_VIRTUAL = &PL_vtbl_mglob
    MG_TYPE = PERL_MAGIC_regex_global(g)
    MG_LEN = 5
SV = PVMG(0x83ef3a8) at 0x84fe160
  REFCNT = 7
  FLAGS = (PADMY,SMG,POK,pPOK)
  IV = 0
  NV = 0
  PV = 0x83c47b8 "(a)(b)"\0
  CUR = 6
  LEN = 8
  MAGIC = 0x83c2ec0
    MG_VIRTUAL = &PL_vtbl_mglob
    MG_TYPE = PERL_MAGIC_regex_global(g)
    MG_LEN = -1
data='(a)(b)'
eteb1:a)(b
eteb2:(a)(b)
[download]

After seeing this, I reviewed the docs for Text::Balanced, and noticed this:

Note that in a list context, the contents of the original input text (the first argument) are not modified in any way. However, if the input text was passed in a variable, that variable's pos value is updated to point at the first character after the extracted text. That means that in a list context the various subroutines can be used much like regular expressions. For example:

In short, it's supposed to do that. That way it's ready to pull out the *next* bits of balanced text for you. Appending a null to the end of the string simply resets the string.

Sigh! Had I read the docs before playing with the code, I'd've saved myself a little time. Ah, well...

...roboticus

When your only tool is a hammer, all problems look like your thumb.

Thanks for the explanation.