Re: Re: Pagination Optimization

Thanks for your help, in fact helped a lot... Now my function only performs 2 queries, here it is: ======================================= function fetch_paginated($query='',$page=1,$itens=20) { $this->query($query); $total_rows = $this->num_rows(); if($total_rows > $itens) { $this->total_pages = ceil($total_rows/$itens); $this->pages_before = ceil($page - 1); $this->pages_after = ceil($this->total_pages - $page); $this->query($query.' LIMIT '.(($page*$itens)-$itens).','.$itens); while($this->fetch_array()) { $results[] = $this->row; } } elseif($total_rows > 0) { while($this->fetch_array()) { $results[] = $this->row; } $this->total_pages = '1'; $this->pages_before = '0'; $this->pages_after = '0'; } else { return null; } return $results; } ======================================= Regards, Bruno B B Magalhães On Jan 7, 2005, at 10:09 PM, M. Sokolewicz wrote:

first of all, you're running 4 queries here. 4 queries is a lot! Especially when you don't need more than 2 ;) the problem here is that your queries are pretty "unknown" to this function. Although it does a nice result for that unknowing, there's a few minor things that make it faster. First of all, would be using less queries. What I usually do is issue a query like this: "SELECT count(some_unique_col) WHERE (that_where_clause_youre_using_in_the_select_query)" then, we do some math. $pages_before = $page-1; $rows_before = $pages_before*$itens; $rows_after = $total_number_of_rows-($page*$itens); $pages_after = ceil($rows_after/20); Then do the actual selecting of the rows using the limit. The thing that makes it slow in your example is the fact that 4 times you're selecting ALL data from the relevant rows, and buffer it. You buffer it, but don't use any of it, except for the number of rows. Mysql does a far quicker job at this than PHP would, so use mysql. :) Then, you're using 3 queries to determine the rows around the page; even though, with a bit of simple math, you can calculate it. And trust me on this, simple math is faster ;) anyway, hope that helped.