On 12/14/05, Michael Graham <[EMAIL PROTECTED]> wrote: > > On Wed, 14 Dec 2005 15:18:42 -0500 > Cees Hek <[EMAIL PROTECTED]> wrote: > > [...excellent explanation snipped...] > > > So I don't think Rob is talking about changing the whole philosophy of > > CGI::Application, he is just talking about refactoring some bloated > > methods into smaller pieces (which is generally considered a good > > thing). > > Cees, thank you for this most excellent post - it makes everything > *much* clearer!
And thank you, Cees, for putting what I was trying to say into the words needed to communicate to the audience. I was clearly not doing so and I apologize to the list for that. > I think I misinterpreted this "Everything is a Plugin" idea to be > something much more abstract and general purpose, e.g. mucking about > with the dispatcher or the execution pipeline in highly flexible ways. > But that's obviously not the intent here. > > Splitting out 'run' or other monolithic parts of C::A into self-contained, > single-purpose bits is clearly a good thing. Longer term, I can also > imagine how making some of these bits pluggable would also be useful. To me, those are one and the same thing. By making each subroutine responsible for one and only one thing, we make the pieces of the whole easier to work with. We can also provide a callback hook for each of these steps. For example, add a callback hook for "translate_request_into_runmode". I can see benefit in multiple translators, each able to translate one set of request types. If it cannot handle it, it can punt, allowing the next guy in turn to try his hand. The default would be there. But, getting each bit into its own function is definitely the first step. Rob --------------------------------------------------------------------- Web Archive: http://www.mail-archive.com/[email protected]/ http://marc.theaimsgroup.com/?l=cgiapp&r=1&w=2 To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
