In terms of a real-world example of what goes in these 4 different layers, assume we
are writing a database-backed application that needs to fetch a user record containing
various fields. This is a very common thing to do in a web app. Our template layer
might look something like this:
auto_prepend_file = "./helpers.inc"
include_path = "/usr/local/lib/php"
Heading 1
Page content
Heading 2
Yada Yada
Heading 3
Page content
]]>
';
?>
\n";
if($age=birthday($user)) {
echo "Congratulations, today is your ";
echo "$age birthday!
\n";
}
}
function title($title) {
echo "$title\n";
echo "\n";
}
function footer() {
echo "\n";
}
?>]]>
10 && ($age%100)<20) $ap='th';
else switch($age%10) {
case 1: $ap = 'st'; break;
case 2: $ap = 'nd'; break;
case 3: $ap = 'rd'; break;
default: $ap = 'th'; break;
}
return $age.$ap;
else
return false;
}
?>]]>
In this case the final layer written in C contains the ~mysql_*~ functions, and the ~date()~
function. These happen to be standard PHP functions. If ~birthday()~ is called many times on
every single request and since how you figure out if it is someone's birthday is unlikely
to change very often, this may be a good candidate to translate into C. Although, in this
example, the birthday function is probably too simple to see much of a performance improvement.
On the other hand, other than a little bit of added parameter parsing, if you compare the C
version of ~birthday()~ to the PHP version, it isn't that much harder to write it in C.
10 && (age%100)<19) sprintf(ret_age,"%dth",age);
else switch(age % 10) {
case 1: sprintf(ret_age,"%dst",age); break;
case 2: sprintf(ret_age,"%dnd",age); break;
case 3: sprintf(ret_age,"%drd",age); break;
default:sprintf(ret_age,"%dth",age); break;
}
} else {
RETURN_FALSE;
}
RETURN_STRING(ret_age,1);
}
]]>