search

joelvh / webpagetest

Automatically exported from code.google.com/p/webpagetest
0 stars 0 forks source link

Chrome early flush does not appear to work #76

Closed GoogleCodeExporter closed 8 years ago

GoogleCodeExporter commented 8 years ago 
Something is going on with the testing infrastructure that is causing Chrome to 
wait until after the base HTML completely downloads before it starts parsing 
and executing resources.

Need to investigate if this is something we are causing or an issue with Chrome 
17.

Original issue reported on code.google.com by PatMee...@gmail.com on 16 Mar 2012 at 2:08

GoogleCodeExporter commented 8 years ago 
Turns out it IS actually working correctly but Chrome on the test machines is 
taking significantly longer than IE to request subsequent resources (so it can 
happen after the base page completes).  Looks like IE starts the next request 
in ~100ms and Chrome takes 250ms: 
http://www.webpagetest.org/forums/showthread.php?tid=10756&pid=16746#pid16746

This is very similar to the problem where a new renderer was being spawned but 
Chrome appears to be honoring the --process-per-tab parameter and not starting 
a new renderer so there is some other bottleneck going on.

Original comment by PatMee...@gmail.com on 19 Mar 2012 at 3:24

GoogleCodeExporter commented 8 years ago 
Changed from using about:blank to serving a blank page from the local web 
server which was enough to trigger Chrome to do the heavy lifting at browser 
launch instead of navigation (my guess is browser cache initialization since 
it's a clean profile).

Original comment by PatMee...@gmail.com on 20 Mar 2012 at 6:48