Understanding PHP Type Coercion

Question

I saw this small piece of code that is evading my understanding:

Answer 1

In the official documentation, the test of equality betwen 2 vairiables is made as follow:

$a == $b # Equal TRUE if $a is equal to $b after type juggling.

Example :

$a = 13;   # integer type
$b = "13"; # string type
var_dump($a == $b); # will say TRUE, because juggling was made
var_dump($a === $b); # will say FALSE, because PHP will also evaluate the type of variables :)

Answer 2

I think this article explains it pretty well:

Type-coercing comparison operators will convert numeric strings to numbers

Just to quote the main issue here :

According to php language.operators.comparison, the type-coercing comparison operators will coerce both operands to floats if they both look like numbers, even if they are both already strings:

where both strings are using exponential notation, hence are treated as numeric strings, making loose comparison (==), coerce these strings to floats before actually "loosely" comparing them.

As a best practice and to prevent unexpected behaviour, always try to use identity equality (===), especially when dealing with strings

Answer 3

PHP attempts to convert to type float because the string begins with a 0. It stops after 0 because the next character is not a number. The same thing happens when you use type coercion to convert scientific notation to integer:

$x = (float)"12E-1x";  // $x == 1.2
$x = (int)"12E-1x";  // $x == 12 (stops at E because it's not an integer)

Answer 4

It is not really an answer, but if you try:

$a = '0e4620974319065090195629887368549';
$b = '0e8304004519934940580242199033918';
echo floatval($a) . '<br>' . floatval($b);var_dump($a == $b);

You get:

0

0

bool(true)

Now, if you try:

$a = '0e4620974319065090195629887368549';
$b = '1e8304004519934940580242199033918';
echo floatval($a) . '<br>' . floatval($b);var_dump($a == $b);

You get:

0

INF

bool(false)

My guess is that PHP converts the strings to floats and gives comparison result using the floats obtained, which are not correct anyway, but that is another story.