Re: PHP 4.0 Bug #5152: Object passed in a session generateserrors when member functions are called
| From: | Stanislav Malyshev | Date: | Thu, 22 Jun 2000 21:36:57 +0000 |
| Subject: | Re: PHP 4.0 Bug #5152: Object passed in a session generateserrors when member functions are called | ||
| References: | 1 | Groups: | php.dev |
| Request: | Send a blank email to php-dev+get-21989@lists.php.net to get a copy of this message | ||
RL>> But since the class data is really not much use without the class
RL>> definition, I don't see the point in separating them. And it wouldn't be
Is there a point in defining classes at all?e
RL>> that much additional data. It's not lie this stuff is getting passes
Yes it will be. Try to take some live class library (PHPLIB is a good
example). Now try to derive two levels of classes from them (that's a real
life example too). Not try to write down what you have to serialize to get
full class definition for this object.
But this is not the point. The point is that carrying class definitions
along with objects defies all concept of classes. You could just as well
carry PHP code setting variables and eval it on session load. That looks
like a dirty hack and quacks like a dirty hack.
RL>> back and forth across the wire. It's all local, so I don't see the size
RL>> as being any sort of issue. Although for the shared memory backend I
Size is not an issue. Wrong concept is. Serialization never carried along
class definitions. I know no serialization structure that requires to
carry data sematics along with the data. Why should we do it?
--
Stanislav Malyshev stas@zend.com
+972-3-6139665