Bug 100991 - Refactor ReplayPerfTest and PageLoadingPerfTest
Summary: Refactor ReplayPerfTest and PageLoadingPerfTest
Status: NEW
Alias: None
Product: WebKit
Classification: Unclassified
Component: Tools / Tests (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Zoltan Horvath
URL:
Keywords:
Depends on: 111027
Blocks: 77037
  Show dependency tree / graph
 
Reported: 2012-11-01 14:49 PDT by Zoltan Horvath
Modified: 2013-02-27 18:24 PST (History)
1 user (show)

See Also:

Attachments
Work in progress (7.00 KB, patch)
2012-11-07 01:09 PST, Ryosuke Niwa 		no flags Details | Formatted Diff | Diff
Demonstrates interesting cache-dependent behavior (26.14 KB, text/html)
2012-11-07 01:12 PST, Ryosuke Niwa 		no flags Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Zoltan Horvath 2012-11-01 14:49:17 PDT 
Since we removed the PageLoadingPerfTest conception we don't need that class, but ReplayPerfTest uses it, so it's time to do some cleanup.
Comment 1 Ryosuke Niwa 2012-11-01 15:16:24 PDT 
I think we can change ReplayPerfTest to use the same iframe approach.
Comment 2 Zoltan Horvath 2012-11-01 16:05:22 PDT 
(In reply to comment #1)
> I think we can change ReplayPerfTest to use the same iframe approach.

Good point and it makes sense, since it would be really interesting to measure the memory usage after a page loads. I'll do it with this in mind.
Comment 3 Ryosuke Niwa 2012-11-07 01:08:41 PST 
I think we can make ReplayPerfTest inherit from PerfTest and use the same iframe approach.
Comment 4 Ryosuke Niwa 2012-11-07 01:09:52 PST 
Created attachment 172732 [details]
Work in progress

Here's a snippet of code that turns replay tests into an iframe-based tests.
Comment 5 Ryosuke Niwa 2012-11-07 01:12:38 PST 
Created attachment 172733 [details]
Demonstrates interesting cache-dependent behavior

For some reason, immediately changing iframe's src makes subsequent page loads really fast. Delaying it with 50ms or so will make the results more uniform across iterations. I'm not sure whether this behavior is caused by WebCore or web-page-replay.