W3C home > Mailing lists > Public > public-web-perf@w3.org > May 2012

RE: [ResourceTiming] Clarifying a few things...

From: Jatinder Mann <jmann@microsoft.com>
Date: Fri, 4 May 2012 18:41:12 +0000
To: Andy Davies <dajdavies@gmail.com>, James Simonsen <simonjam@chromium.org>
CC: "public-web-perf@w3.org" <public-web-perf@w3.org>
Message-ID: <AE5FFD9402CD4F4785E812F2C9929D6504EEB938@SN2PRD0310MB383.namprd03.prod.outlook.com>
> Hmm, that works on the premise that the event handler can be added before the event fires 
> which is problematic for bookmarklets and may be for beacons too...
> If the event handler isn't in place in time, there's no way of telling if data has been lost is there?

We recommend setting the onresourcetimingbufferfull callback prior to the loading of the resources. Unless you increase the buffer size by using the setResourceTimingBufferSize method, if you do not have the callback handler set, resource data greater than the buffer size will be lost.

> Based on this data I suspect the 150 mark is probably OK, though some articles I've read suggest
> the number of resources per page is on the up (particularly once third party resources get involved).

We found that a buffer size of 150 resources is sufficient for most web pages. However, if you know the page you are going to measure has a large number of resources, you can always increase the size of the buffer by using the setResourceTimingBufferSize method.

Received on Friday, 4 May 2012 19:02:33 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:04:32 UTC