I'm trying to reliably identify when a browser window/tab is activated and deactivated. Normally, window
's focus
and blur
events would do, but the document contains several iframes.
When an iframe is focused, the main window gets unfocused and vice versa, so we have the following possibilities of focus events [(none)
means the window/tab is deactivated]:
current focus new focus events
----------------------------------------------------------------------
window (none) window:blur
window iframe window:blur + iframe:focus
iframe (none) iframe:blur
iframe window iframe:blur + window:focus
iframe another iframe iframe:blur + iframe:focus
(none) window window:focus
(none) iframe iframe:focus
It is no problem to register all of these events, as shown by this fiddle. But whenever we switch from the main window to an iframe or vice versa, or between two iframes, the respective blur
and focus
events both fire; and they fire with a small delay at that.
I am worried about the concurrency here, since the blur
handler could go and start doing stuff, but it should have never started because the user actually just switched focus somewhere in between the frames.
Example: A page should do some AJAX requests periodically whenever it is currently not active. That is, it should start requesting whenever the user deactivates the tab and stop requesting as soon as it's activated again. So we bind a function to the blur
event that initiates the requests. If the user just clicks on another iframe, blur
, and shortly after that, focus
is triggered. But the blur
handler already fires away, making at least one request before it can be stopped again.
And that's my problem: How can I reliably detect when a user actually (de-)activates a browser window containing iframes, without risking to get a false alarm caused by two immediate blur
and focus
events?
I wrote a half-baked solution that uses a timeout after a blur
event in order to determine if there was an immediate focus
event after it (fiddle):
var active = false,
timeout = 50, // ms
lastBlur = 0,
lastFocus = 0;
function handleBlur() {
if (lastBlur - lastFocus > timeout) {
active = false;
}
}
function handleFocus() {
if (lastFocus - lastBlur > timeout) {
active = true;
}
}
$(window).on('focus', function () {
lastFocus = Date.now();
handleFocus();
}).on('blur', function () {
lastBlur = Date.now();
window.setTimeout(handleBlur, timeout);
});
$('iframe').each(function () {
$(this.contentWindow).on('focus', function () {
lastFocus = Date.now();
handleFocus();
}).on('blur', function () {
lastBlur = Date.now();
window.setTimeout(handleBlur, timeout);
});
});
But I believe this could be very problematic, especially on slower machines. Increasing the timeout is also not acceptable to me, 50 ms is really my pain threshold.
Is there a way that doesn't depend on the client to be fast enough?