Quotation Marks in PHP

dr0n3

Items within single quotation marks are treated literally. Items within double quotation marks are extrapolated (that is, a variable’s name is replaced with its value).

From my point of view, if I were to comply with this rule, I would end up with messy code.

In the example below, there is a mixture of single and double quotation marks. In my opinion this looks tremendously ugly.

<?php

$name='John';

echo '<p>Hello '.$name.', how are you?</p>'."\n"; // Hello John, how are you?

?>

Is it perfectly acceptable to re-write the code as follows?

<?php

$name="John";

echo "<p>Hello ".$name.", how are you?</p>\n"; // Hello John, how are you?

?>

Now you can see that I have used double quotation marks for anything that requires quotation marks. I think this looks much neater. I am guessing that it takes slightly longer to process the script this way because PHP has to figure out whether to extrapolate everything between double quotation marks. Is this really a problem? Please post your thoughts and opinions.

laserlight

The difference is negligible.

Actual extrapolation would take slightly longer:

echo "<p>Hello $name, how are you?</p>\n";

But otherwise there isnt really a difference.

Also, using braces to delimit the variable names when extrapolating should reduce the time taken, possibly back to when concatenation was used instead of extrapolation:

echo "<p>Hello {$name}, how are you?</p>\n";

Of course, in this trivial example it doesnt matter either way.

Mordecai

But, since echo takes multiple parameters, splitting the variables and parameters up would be fastest:

echo '<p>Hello ', $name, " how are you?</p>\n";

😃

yelvington

Actual results, 100,000 iterations on a P2-400:

Test run single quotes in 0.862855911255 seconds

Test run double quotes in 0.747938156128 seconds

Test run simple string no braces in 2.09341287613 seconds

Test run simple string with braces in 2.08907198906 seconds

Test run multiple args in 0.601377010345 seconds

(Output was buffered and flushed to remove i/o as a factor. )

As you can see, Mordedai wins. But all of this is dangerously close to debating the number of angels that can dance on the head of a pin. It is likely that the time spent thinking about this issue will not be recovered by the "speed increase" during the lifecycle of the program. Your best practice would be to write clean, readable code first, and not worry about optimization until you understand where the program is spending meaningful amounts of time.

laserlight

Provide the code used when benchmarking, yelvington.

There has already been auch a benchmarking done around here a few years back, and I believe that it did show using multiple arguments when echo to be relatively faster than the rest, but the results for using braces when interpolating differed greatly from your own.

yelvington

Here you go. I just took the examples from the postings, added timers and buffering, and printed the output.

You might want to try multiple runs to smooth over any contention for CPU with other processes. (I ran this on a web and mail server that's handling the usual constant bombardment of spam and IIS virus attacks.)

<pre>
<?
// getmicrotime() from the manual page
// for microtime()
function getmicrotime()
{
   list($usec, $sec) = explode(" ", microtime());
   return ((float)$usec + (float)$sec);
}

$loopsize = 100000;
$name='John';

ob_start();
$time_start = getmicrotime();
for($i=0; $i < $loopsize; $i++)
echo '<p>Hello '.$name.', how are you?</p>'."\n";
$time_end = getmicrotime();
$time = $time_end - $time_start;
ob_end_clean();
echo "Test run single quotes in $time seconds\n";

ob_start();
$time_start = getmicrotime();
for($i=0; $i < $loopsize; $i++)
echo "<p>Hello ".$name.", how are you?</p>\n";
$time_end = getmicrotime();
$time = $time_end - $time_start;
ob_end_clean();
echo "Test run double quotes in $time seconds\n";

ob_start();
$time_start = getmicrotime();
for($i=0; $i < $loopsize; $i++)
echo "<p>Hello $name, how are you?</p>\n";
$time_end = getmicrotime();
$time = $time_end - $time_start;
ob_end_clean();
echo "Test run simple string no braces in $time seconds\n";

$time_start = getmicrotime();
for($i=0; $i < $loopsize; $i++)
echo "<p>Hello {$name}, how are you?</p>\n";
$time_end = getmicrotime();
$time = $time_end - $time_start;
ob_end_clean();
echo "Test run simple string with braces in $time seconds\n";

ob_start();
$time_start = getmicrotime();
for($i=0; $i < $loopsize; $i++)
echo '<p>Hello ', $name, " how are you?</p>\n";
$time_end = getmicrotime();
$time = $time_end - $time_start;
ob_end_clean();
echo "Test run multiple args in $time seconds\n";

?>
</pre>

Drakla

Ooops - you missed out the start ob in one of those lines.

I just shut down everything I could and just left Apache 2 and with the the cli version of php running and got these results.

Test run single quotes in 2.31234908104 seconds
Test run double quotes in 2.33806204796 seconds
Test run simple string no braces in 2.84585499763 seconds
Test run simple string with braces in 2.79856300354 seconds
Test run multiple args in 2.16451311111 seconds

I ran it loads of times, and there was hardly any devition.

Using the Apache module got this:

Test run single quotes in 2.31322193146 seconds
Test run double quotes in 2.29662084579 seconds
Test run simple string no braces in 2.76868391037 seconds
Test run simple string with braces in 2.7918510437 seconds
Test run multiple args in 2.09991002083 seconds

Hardly any difference.

Warning - this is Apache running on XP and a 1.8Ghz Celeron, anyone got a different OS to test this with?

Weedpacket

Nope, those results look much like all the others I've seen. The multiple args method has an additional advantage that won't be seen in these tests: if one of the args takes a bit of time to evaluate, the args before it will be output before the evaluation is carried out, instead of being held on to for concatenation or interpolation. Overall the speed is the same as concatenation (and faster than interpolation, which isn't as versatile), but output is produced sooner.