note BrowserUk <blockquote><i></i></blockquote> <p>Reading between the lines of your last couple of OPs, you seem to be parsing multiple XML files with XML::Simple; and doing so in threads in the hope of preventing leaks? <p>From a couple of quick experiments, it seems that whichever underlying parser -- XML::SAX, XML::SAX::PurePerl, XML::Parser -- you use, they all leak substantial amounts of memory. XML::SAX more so that XML::Parser, but still in the range of 10 to 30MB per iteration for a 12MB xml file. <p>My suggestion to cure that -- essentially what [Zentara] suggested earlier, but with a twist -- is to run the parser in a separate process. I used this to parse the file and then export the structure back to the parent:<code> #! perl -slw use strict; use Storable qw[ freeze ]; use XML::Simple; $XML::Simple::PREFERRED_PARSER = 'XML::Parser'; binmode STDOUT; print freeze XMLin( $ARGV[ 0 ] ); </code> <p>And then in the main script just a simple backticks command and [thaw]:<code> for( 1 .. 1000 ) { my $xml = thaw `xmlSto.pl junk.xml`; print mem $_; <STDIN>; } </code> <p>It took about 1 extra second or so for the parent process to get access to the data structure, but it cures the leak completely. <div class="pmsig"><div class="pmsig-171588"> <hr /> <font size=1 > <div>With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'</div> <div>Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.</div> <div>"Science is about questioning the status quo. Questioning authority". </div> <div>In the absence of evidence, opinion is indistinguishable from prejudice. <p align=right> [http://thebottomline.cpaaustralia.com.au/|RIP Neil Armstrong]</p></div> </font> </div></div> 1001200 1001200