So dom style http://xmltwig.org

#!/usr/bin/perl --
use strict; 
use warnings; 
use XML::Twig;
my $t = XML::Twig->new( pretty_print => 'indented' );
$t->xparse(q{<roooooo>
  <citreference id="mycite">
    <linktext>
      <keyword conref="ds"/> support for IPv6 addresses</linktext>
  </citreference>
  <moops></moops>
</roooooo>});
$t->print;
my( @links ) = $t->findnodes( '//linktext' );
my( $moops ) = $t->findnodes( '//moops' );
for my $link ( @links ){
    print $link->path, "\n";
    $link->cut->paste( $moops );
}
$t->print;
__END__
<roooooo>
  <citreference id="mycite">
    <linktext><keyword conref="ds"/> support for IPv6 addresses</linkt
+ext>
  </citreference>
  <moops></moops>
</roooooo>
/roooooo/citreference/linktext
<roooooo>
  <citreference id="mycite"></citreference>
  <moops>
    <linktext><keyword conref="ds"/> support for IPv6 addresses</linkt
+ext>
  </moops>
</roooooo>
[download]

Thank you for the suggestion. My solution was this:

$topicref->set_content(@$h);

Sorry i wasn't clear about that. I'd like everything inside of linktext to go inside of xref.

<xref><keyword conref="ds"/> support for IPv6 addresses</xref>
[download]

I also didn't mention there might be other elements midway.

<linktext><keyword conref="ds"/> support for <b>IPv6</b> addresses</li
+nktext>
[download]

So I need to iterate through the @children array and handle each child, which is either text or an element.

thanks, will try these other suggestions.

Sorry i wasn't clear about that.

You're still not very clear

Anyway, see Re^2: XML::Twig - setting text and elements ( reparent move cut detach ), instead of moops, paste to something else

Or if you want to rename the tag just https://metacpan.org/pod/XML::Twig#set_name-name