Application design - Opinions?

[Vigilante]

This isn't really a php specific thought however what I have in mind is something every php coder should, at some point, put a little thought into.

That is, what kind of code design to use for a given project. By code design I mean OOP vs procedural method and the system used to create additional pages on a site, as in having each php file include a config which includes a global script and a function library or having an index php file call different modules depending on it's arguments.

Personally, I like OOP using the index-as-a-gateway method as it allows infinite extension with modules and offers relatively low redundancy. I was wondering if there would be any unforeseen issues (I haven't encountered any) with making users always load the same php file that includes others as needed on large applications.

This brings me (in a non-linear logical sort of way) to the specifics of OOP design. Do you gain any possible advantage from using oop for it's own sake?

For example.. (code not meant to be great, just illustrative)

class page
{
  function page()
  {
    GLOBAL $_GET;
    $action = $_GET['action'];  

    switch ($action) {
      case 'index':
        $this->index();
        break;
      case 'form':
        $this->form();
        break;
    }
  }
  function index()
  {
    // do index stuff
  }
  function form()
  {
    // do form stuff
  }
}

If that is the same as...

$action = $_GET['action'];  

switch ($action) {
  case 'index':
    // do index stuff
    break;
  case 'form':
    // do form stuff
    break;
}

Would I be better off not using a class in this situation? Maybe there is some more efficient way of writing the class... I haven't been writing oop so I'm probably missing obvious ways to make better classes. I was just wondering whether anyone thinking about this had anything to say on their favorite way to plan projects.

[AstroTeg]

I used to do the switch() approach as well until I figured out I could do this instead:

$function = '';
if(isset($_GET['Action']))
	$function = 'Do_' . $_GET['Action'];

if(function_exists($function))
	$function($_POST, $_GET);
else
	Do_Default($_POST, $_GET);

Basically, "Action" as a parameter on the URL will define which function gets executed. I can go into more details, but I post this as an example to my discussion below...

In the above code, its pretty easy to add additional functionality. Its procedural and nothing too exciting is going on here. Its possible it could be wrapped up in a class, but I haven't seen the value in it.

I could see value in setting the get and post variables as a member of the class (as well as maybe server, cookie, and file variables). But I find when I go from working on one site to the next, they have different requirements and I end up modifying the code to fit the needs.

I have heard some arguments which say OOP for web development is overkill (the amount of complexity versus the advantages in performance, readability, and maintenance). I would say if I was with a group of web developers all coding for 1 large site, then OOP would definitely be a great way to go. OOP forces structure and to some degree documentation.

But then I believe in using the right tool for the right job. If you're comfy with OOP, go for it. If not, procedural work will great too...