On Sunday, June 16, 2002, at 07:02 AM, Eric Frazier wrote:
> The big points I want to achieve right now, is to make everything I > write > OOP, separate HTML from code as much as possible, and to not make it > impossible to deal with for the people I work with who don't know as > much > perl as I do. One thing that has proven immensely beneficial on the project that I'm working on now, which is very similar to yours in that it's a complete redo intended to be 'right', is this: Write each page's functionality in a package that is 100% removed, interface-wise, from HTML, then 'glue' that module to HTML (via HTML::Template for output and CGI.pm for input) in a module name identical to, but preceded by Apache:: So, we have a page module like: Functionality in CompanyName::page1, and glue code in Companyname::Apache::page1 Modules NOT under the CompanyName:: namespace cannot be dependent IN ANY WAY on mod_perl, but must be mod_perl clean, of course. This makes the module functionality very easy to test WITHOUT firing up Apache. We have a complete white box test suite that tests at the Perl level, and complete black box test suite that tests at the HTTP level. This proved a good strategy when we decided to implement a SOAP interface to some of our functionality. We made some new glue code in CompanyName::SOAP::page1 that was based on SOAP::Lite and it was just as easy as it should be, with 100% code reuse. -- -- Tom Mornini -- InfoMania Printing and Prepress -- -- ICQ: 113526784, AOL: tmornini, Yahoo: tmornini, MSN: tmornini
