Question

Currently, I build two classes for each database table. For instance, if I have the table person, I will have the classes Person_List and Person.

Design-wise, is it better

1. for Person_List to output an array of Person; or
2. for it to output an array containing arrays of rows in the table.

Performance-wise, which is better?

Answer 1

I believe that design-wise, and taking performance into account, would be to (if you insist on Person_List class to represent table and Person to represent single record):

• use Iterator interface for Person_List class, so you can iterate through the table without the need to pull all the records at once (it should be significant performance gain in some cases),
• additionally use Countable interface for Person_List class, so you are able to count all the results if necessary by getting count directly from database,

This should give you flexibility and allow you to use Person_List class objects similarly as arrays.

If you still have problems employing these two interfaces, here is some explanation:

• every time you do foreach ($table as $record) (where $table is an instance of Person_List), the current() method of Person_List class will be invoked (because it is a part of Iterator interface – see docs here), which should return an object of Person class; this should happen using eg. mysql_fetch_object();
• when you call count($table) (where $table is an instance of Person_List), the count() method of Person_List class will be invoked, which in turn can use eg. mysql_num_rows() function to return all the results instead of pulling them from database and then counting (this will be again significant performance gain),