If the connection is successful, a MySQL link identifier will be returned. Notice that I'm
testing for failure and performing the connection on one line. Link identifiers are always
greater than zero, and zero is returned when the connection cannot be made. So, testing
for a FALSE return value allows us to detect a failed connection. If that happens, we just
abort the entire script.
The function used to connect to the database is mysql_pconnect. If you've flipped
through the descriptions of the MySQL functions in Chapter 13, "Database
Functions," you might remember another function called mysql_connect. These
two functions operate identically inside a script, but mysql_pconnect returns
persistent connections.
Most of the database functions that PHP offers incorporate the idea of a persistent
connection—a connection that does not close when your script ends. If the same Web
process runs another script later that connects to the same database server, the connection
will be reused. This has the potential to save overhead. In practice, the savings are not
dramatic, owing to the way Apache 1.3.x and earlier use child processes instead of
threads. These processes serve a number of requests and then are replaced by new
processes. When the process ends, it takes its persistent connection with it, of course.
Only under high loads will your script benefit from persistent connections, but that's
exactly the time when it needs to benefit from them. Using mysql_pconnect costs
nothing, so I use it by default. At the time of this writing, Apache 2.0 is nearing release. It
promises a multithreaded approach that will certainly take full advantage of persistent
connections.
The next step is to select a database. Here I've selected the database named store.
Once we tell PHP which database to use, we get all rows from the catalog table. This
is done with the mysql_query function. It executes a query on the given link and
returns a result identifier. We will use this result identifier to fetch the results of the
query.
Before we begin pulling data from the results, we must begin building an HTML table.
This is done, as you might expect, by using an opening table tag. I've created a header
row with a gray background and left the rest of the table behavior as default.
Now that the header row is printed, we can fetch each row from the result set. The fastest
way to do this, executionwise, is to use mysql_fetch_object. This expresses each
column in the result as the property of an object. The names of the columns are used for
the names of the properties. You could also use mysql_fetch_row or
mysql_fetch_array, which are equally efficient. Most of the time using an object