I'm looking at converting a large CGI installation over to mod_perl. The code is clean, so moving it to Apache::Registry isn't causing any problems, but I'd like to get the most bang out of the conversion that I can.
Currently the code is a set of scripts that all call modules that are subclasses of a local subclass of CGI::Application (I realize the subclassing is hitting our efficency, but it's deemed worthwhile for the advantages it brings). These modules have run-modes as per standard CGI::App, and return to CGI::App a scalar ref of HTML, which CGI::App outputs with the appropriate headers via CGI.pm. The scalar ref our modules return are content files processed through a Template Toolkit template. I'm seeing the following potential areas of speed improvements: - CGI::App could be altered to use Apache::Request. I haven't looked into the code yet, but I recall from previous scans that it isn't overly complex (the beauty is in the simplicity, after all), so if the efficiency is worth it, and if CGI::App can't be easily coaxed into it, I should be able to fork my own to do Apache::Request. Of course, I'm doubtless not the first to consider this, so has someone already invented this wheel? - My Template Toolkit object could be reused, along the lines of Apache::DBI. (Since that object isn't changed in any session, just processes some data and returns a result). Sadly, I'm not clear on how to do this. Apache::Template is a handler in it's own right, so it won't do what I seek. Should I jump into Apache::DBI to see how it does it, or is there some documentation on how to create persistant objects that survive between connections? [Actually, I want more than a persistant TT2 object, since I have a wrapper module that has some helpful interface routines, so I'd want that to be persistant as well.] I've looked at perl.apache.org and skimmed the eagle book without finding this, but it is a bit specific, so I could have missed it. Thanks in advance for any help.
