WebPagetest 2.8


From SVN (requires a client that can follow externals references for files): svn co https://webpagetest.googlecode.com/svn/tags/webpagetest/2.8/

Changes from WebPagetest 2.7.2:

Web Server

  • Added an installation check script that checks for common install issues
  • Added data comparison graphs to the filmstrip view (comparing times, bytes and requests across the tests being compared)
  • Added a simple method for issuing SPOF (Single Point Of Failure) tests
  • Added a new metric "Visually Complete" that is the first point in time when the video frames are 100% complete (also an option for ending the filmstrips and videos at this point)
  • Changed the displayed location to not lookup CDN requests and just list which CDN served the response (the Maxmind data isn't very good for CDN edge locations)
  • Added support for checking for gzip of icon files
  • Added support for properly handling portrait-orientation for thumbnails in the filmstrip (mobile)
  • Added logic to smooth the bandwidth chart to the configured traffic shaping bandwidth to counter the effect of spiky measurements.
  • Added a 3G mobile profile for latency simulation based on data collected by the Google mobile team in Seattle
  • Added a "Native" profile that disables traffic shaping
  • Fixed an issue where URLs were not being escaped when added to the image map for the waterfalls. Usually not a problem but came across a few URL's with quotes and HTML in them.

Test Agent

  • Improved Chrome/Firefox agent:
    • Fixes to support Chrome 24+
    • Fixes to support Chrome Canary 23+
    • Fixed support for multiple content blocking strings
  • Improved IE Agent:
    • Added support for custom viewport sizes
    • Fixed clearing of cached 301 redirects on Windows 8 and Server 2008
    • Fixed clearing of the OS certificate caches (OCSP and CRL)
    • Added support for disabling IE compatibility view on a per-test basis
  • Improvements common to both agents:
    • Added support for disabling javascript (not just blocking but putting the browsers in noscript mode)
    • Added a metric that keeps track of the number of IP addresses returned for each host name
    • Added a metric for the estimated RTT to the server for each request (and at a page-level to the base page).
    • Added reporting of the CDN used by the base page and changed the CDN detection logic to look up the CDN name for all of the resources, not just static (only static resources will still contribute to the score).
    • Automatically set the desktop resolution as high as possible up to 1920x1200 to allow for larger viewports on headless test machines