In my desire to find the “right” UUID function to use in my application, I grew to like the MySQL UUID() results produced because they are partially time based. But I didn’t like that I have to make a database call every time I want a new UUID.  I didn’t find any similar functionality in PHP or from user based PHP functions, so here’s my efforts.

It clocks in at 100,000 results in about 1.36 seconds on my 2.8Ghz Intel iMac.

function uuid()
{
	// Time based PHP Unique ID
	$uid = uniqid(NULL, TRUE);
	// Random SHA1 hash
	$rawid = strtoupper(sha1(uniqid(rand(), true)));
	// Produce the results
	$result = substr($uid, 6, 8);
	$result .= substr($uid, 0, 4);
	$result .= substr(sha1(substr($uid, 3, 3)), 0, 4);
	$result .= substr(sha1(substr(time(), 3, 4)), 0, 4);
	$result .= strtolower(substr($rawid, 10, 12));
	// Return the result
	return $result;
}

We produce results looking like this:

1f8fec61-4a6f-9bd5-7f8e-e12325c464c3
1f8fed49-4a6f-9bd5-7f8e-d665dedc0297
1f8fee17-4a6f-9bd5-7f8e-29d1ce3e9e30
1f8feef2-4a6f-9bd5-7f8e-8ce03bebdbdf
1f8fefc7-4a6f-9bd5-7f8e-c22fbedb5390

I recently had to deal with a SQLite database that stored the timestamps as ticks.  Having to first research what a tick is, I created a simple function that will convert a tick value to a timestamp or MySQL datetime value since I couldn’t find anyone else who had done such a thing yet.

You can download it here.

The only tricky part (aside from figuring out what a tick is) was calculating the number of ticks between 0001-01-01 to 1970-01-01. Also of note: a tick is apparently a popular way to store date/time values for Microsoft.

Ran into a little problem hitting the memory limit for PHP when serving 100MB+ files through a script like:

$file = @fopen($filename,"r");

if ($file)
{
while(!feof($file))
{
    print(fread($file, 1024*4));
    flush();
    ob_flush();
}
@fclose($file);
}

Found that you need to call ob_flush() as well as flush() since flush() has no effect on the buffering scheme of your web server.

Just thought I’d throw that out there for anyone else.