Beyond Firebug to NYTProf

Clearly, the problem is in the core application, not the CSS and JS surrounding it. And Firebug only covers the outside of the application.

So I got NYTProf going. A little search gave me the knowledge that calling a page at http://example.com/myprog.cgi?foo=bar is the same as calling it with perl myprog.cgi 'foo=bar', which is so good to know, especially since the addition of the NYTProf step is perl -d:NYTProf myprog.cgi 'foo=bar' .

So, I was able to shave off a second by caching. I could have HOP'd it and just used Memoize, but I like having all the details of a program visible so I don't get bit by something I can't see.

    # all lines of code with  %url_cache are new
    # URLs have been changed to protect the innocent.
    my %url_cache ;
    sub get_service_page {
        my ( $pi ) = @_ ;
        if ( $url_cache{ $pi } ) {
            return $url_cache{ $pi } ;
        my $readfile = pi_Readfile() ;
        my $url = 'http://www.example.edu/~user/projects/XXXXX/' ;
        my $alt =
        my $attr   = 'SGNAME_PUTATIVE' ;
        my $sgname = $readfile->{ $pi }->{ $attr } ;

        if ( ! defined $sgname || '' eq $sgname ) {
            $url_cache{ $pi } = $alt ;
            return $alt ;
        $url =~ s/XXXXX/$sgname/ ;
        $url_cache{ $pi } = $url ;
        return $url ;

So, simply by holding onto that little piece of information instead of checking against the same pi_Readfile() each time, I was able to go from 2-3 seconds to 1.2 seconds. And, now that I'm seeing it, I could hold onto the data structure I get from  pi_Readfile() the same way I hold onto the cache, and could probably tighten up even more.

I don't know why it didn't occur to me to do that in the first place....
Post a Comment