What is more object-oriented?

PHP_Novice2

My scripts call functions from classes in a library.php file. I have several functions that query the database, loop through the results, and print an HTML table. To ensure that a blank table isn't displayed, these functions include logic that counts the number of rows, and displays an error message when zero are returned.

From an OOP standpoint, is it better to check for a null result within the function, or in the script that calls this function? Also, which is considered better practice from a more general programming perspective?

rowanparker

I'd do it in the script that calls the function.
Because, returning zero rows may be useful (e.g. page not found, user doesn't exist, whatever) and therefore you can check for that without throwing errors.

PHP_Novice2

rowanparker wrote:
I'd do it in the script that calls the function.
Because, returning zero rows may be useful (e.g. page not found, user doesn't exist, whatever) and therefore you can check for that without throwing errors.

Perhaps I should explain my functions in greater detail. In pseudocode:

function displayRows()
{

$sql = "SELECT * FROM mytable";
$query = mysql_query($sql);
$numRows = mysql_num_rows($query);

if ($numRows == 0)
{
echo "No rows found";
return;
}

// Loop through result, print table

}

Isn't this doing the same thing?

laserlight

A function should do one thing and do it well.

The function in question prints a table by querying a database. As such, it should handle the case where the database table has no rows to retrieve. Now, it might call other functions that handle some of its implementation details, one of which could be a function that counts the number of rows in the result set. However, this function already exists in the MySQL API, i.e., mysql_num_rows(). As such, you do not need to write another function for that task. If you do, it would have to query the database, resulting in an unnecessary query.

rowanparker

Instead of echoing something, it could also return false.

laserlight

Instead of echoing something, it could also return false.

To do that, the check must be done within the function. If an empty result set is really considered an error as opposed to merely a special case, then this approach may be good in that it decouples the table printing from the error handling.

Shrike

Following on from what Laserlight is saying, from an OO perspective a class (and so an object) should have a discrete responsibility. This generally promotes reuse of code and reduces duplication.

For example, you have a class which both queries a database and generates HTML. What if you want to display an HTML table from another data source? You'd end up either inheriting the base class and overriding certain methods, or duplicating code. Neither of which are particularly appealing.

A quick example:

class DbQuery
{
  function fetchRow()
  {}
}
class SoapQuery
{
  function fetchRow()
  {}
}
class HtmlTable
{
  function draw( $datasource )
  {
    while( $row = $datasource->fetchRow() )
    {
      // draw HTML
    }
  }
}

PHP_Novice2

laserlight wrote:
However, this function already exists in the MySQL API, i.e., mysql_num_rows(). As such, you do not need to write another function for that task. If you do, it would have to query the database, resulting in an unnecessary query.

In terms of performance overhead, does mysql_num_rows() count as a query?

laserlight

In terms of performance overhead, does mysql_num_rows() count as a query?

It should not result in another query as it works on the current result set retrieved.