Here's how to do server-push using ajax long-polling. The browser makes an ajax request which initiates server-side self-polling. The ajax request remains open, waiting for a response until the file changes, and as soon as it gets a response, it makes a new long-polling request.
Here's what it looks like with jQuery and php, implementing the example of live-updating a div in the html showing the number of clients currently connected:
index.html:
<html>
<head>
<title>Comet Test</title>
<script type="text/javascript" src="jquery.js"></script>
<script type="text/javascript" src="longpolling.js"></script>
</head>
<body>
Number of connected users: <div id="total">0</div>
</body>
</html>
longpolling.js:
$(document).ready(function() { connectToServer(1); });
function connectToServer( incp ) {
$.get("LongPolling.php",
{ inc: incp },
function(resp) {
$('#total').html(resp);
connectToServer(0);
}
);
}
LongPolling.php:
<?php
# (over)write file with contents, locking the file while doing so.
# just barf and die if there's an error.
function update($file, $contents)
{
$f = fopen($file, 'w');
if(!$f) { echo "ERROR1"; exit; } # couldn't open file for writing.
if(!flock($f, LOCK_EX)) { echo "ERROR2"; exit; } # couldn't get lock.
fwrite($f, $contents);
fclose($f); # this also releases the lock.
return $contents;
}
# fetch the contents of the given file.
# if the file doesn't exist, create it with contents "0"
function fetch($file)
{
if(file_exists($file)) {
if(!is_readable($file)) { echo "ERROR3"; exit; }
$x = file_get_contents($file);
} else {
$x = 0;
update($file, $x);
}
return $x;
}
$fc = 'connx.txt'; # file that stores the number of connections.
if ( $_REQUEST['inc'] == 1 ) { # someone just connected.
echo update($fc, fetch($fc)+1);
} else { # someone is reconnecting (also happens immediately after connect).
$last = filemtime($fc);
do { # wait until some other instance causes $fc to change...
sleep(1);
clearstatcache(); # otherwise filemtime() results are cached!
} while(filemtime($fc) == $last);
echo fetch($fc);
}
?>
NOTE: This does not track disconnects, so it's more like live-tracking the total number of pageviews.
See http://stackoverflow.com/questions/531001/running-server-side-function-as-browser-closes/531007#531007 for info on keeping track of browser disconnects, ie, server-side action on client disconnect.