Another exploding head

[goobernutz]

What is going on here???? 😕

i'm trying to write a function that will return the integer value + html entity for 1/4 1/2 and 3/4 when given a float value.

i wrote this simple test that loops from 0 to 2 in .05 increments, outputting everything in an ugly little table. BUT!! for some reason .5 always comes out as 1/4??? what did i miss?

<table border="2">
<tr><td>i</td><td>left</td><td>right</td><td>fraction</td></tr>
<?php
for($i=0; $i<2; $i=$i+(.05)){
	$left = intval($i);
	$right = $i - $left;
	$frac = "&amp;nbsp;";
	if($right < 0.25){
		$frac = "&amp;nbsp;";
	} elseif ($right < .5){
		$frac = "&amp;frac14;";
	} elseif ($right < 0.75){
		$frac = "&amp;frac12;";
	} else {
		$frac = "&amp;frac34;";
	}

echo "<tr><td>$i</td><td>$left</td><td>$right</td><td>$frac</td></tr>";
}
?>
</table>

oh yea, and 1 minus 1 comes out as 2.22044604925E-16 !!! is this a problem with the intval function or more likely my inelegant coding?

[Weedpacket]

Probably the latter.....🙂

Floating point binary arithmetic can't represent numbers like five hundredths exactly, for exactly the same reason that floating point decimal can't represent numbers like one seventh exactly. One seventh in decimal is 0.14285714285714285714285714285714...... and five hundredths in binary is 0.000011001100110011...

So the processor has to round the number off. You're getting roundoff errors.

Multiply everything through by 100 and work with integers instead, and only convert to floating point when you need to.

[The_Chancer]

That is very true of Weed, a simple workaround would be something like:

<table border="2">
<tr><td>i</td><td>left</td><td>right</td><td>fraction</td></tr>
<?php
for($i=0; $i<2; $i=$i+(0.05)){
	$left = intval($i);
	$right = $i - $left;
	switch (round($right,2)){
		case 0.25:
		$frac="&frac14;";
		break;
		case 0.5;
		$frac="&frac12;";
		break;
		case 0.75;
		$frac="&frac34;";
		break;
		default:
		$frac="&nbsp;";
		break;
	}
	echo "<tr><td>$i</td><td>$left</td><td>$right</td><td>$frac</td></tr>";
}
?>
</table>

And just round effectively to two decimal places, before trying to work it out.

[goobernutz]

cheers for that guys! what a quick response.

i figured there wouldn't be any problems when dealing with a built-in function and a float with only 1 digit.

the website i work for is expanding to the states and the yanks wouldn't be too happy to see measurements like 34.21 inches! now i can go to the pub 😃

BTW - Weedpacket, i work in Mayfair. I may have passed by your "office". Or chucked some letters in.