Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:4570
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Date: Wed, 24 Sep 2003 16:20:51 -0400 (EDT)
To: internals@lists.php.net
Message-ID: <Pine.LNX.4.21.0309241602260.18150-100000@miranda.org>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Subject: pres2 hacking and refactoring
From: adam@trachtenberg.com (Adam Maccabee Trachtenberg)

I posted this to the pres2 list, but either nobody read it, or
nobody's interested. So, I'll try here.

I've been futzing around with pres2 and I'd like to propose a change
in how the objects are organized. I think this will make it easier for
people to work on designing new formats and also reduce the total
number of LOC.

Right now, we do something like this:

class _example {

   function _example() { /* construct */ }
   function display() { $this->$mode; /* mode is html, plainhtml, 
flash, pdf */ }

   function html() { /* html display */ }
   function plainhtml() { /* plain display */ }
   function flash() { /* flash display */ }
   function pdf() { /* pdf display */ }
}

This makes it really hard to work through all the tag classes because 
there's tons on formatting code embedded in each object. Also, many 
display methods for one mode are implemented by calling the same method 
for another mode. (i.e.: function flash() { $this->html(); })

I propose we rip all these individual display functions out of the tag 
objects and put them in a separate object for each display format, like 
so:

class html {
   function _example(&$example) {
     /* html() function previously in the _example class */
   }
}

And make the display() function be:

function display($mode) {
             $class = get_class($this);
             $mode->$class($this);
}

Where $mode is one of html, plainhtml, etc.

This, IMHO, has a few advantages:

1) By extending from one class, we can easily override just the methods 
we want to modify to create new display formats. (For example, the 
flash class only needs to implement one method since everything else is 
just calling the html() method.)

2) From my perspective, people generally hack on the output for one 
format at a time instead of one tag at a time. This consolidates all 
the code they're working on into one common place instead of spreading 
it across a whole file.

3) When new tags are added, we can just add a basic implementation in a 
parent display class and nothing will break.

Thoughts? Comments?

-adam

PS: In case you were wondering, I've actually already done all this and 
it works "on my machine." But I've only really tested the html output; 
I decided to wait on flash and pdf until I got some feedback on way or 
another.

-- 
adam trachtenberg
adam@trachtenberg.com