Bug 270791 - [perfs] Regression on TypeHierarchyPerfTest#testOpenObjectHierarchy() test
Summary: [perfs] Regression on TypeHierarchyPerfTest#testOpenObjectHierarchy() test
Status: VERIFIED FIXED
Alias: None
Product: JDT
Classification: Eclipse Project
Component: UI (show other bugs)
Version: 3.5   Edit
Hardware: PC Windows XP
: P3 normal (vote)
Target Milestone: 3.5 M7   Edit
Assignee: Markus Keller CLA
QA Contact:
URL:
Whiteboard:
Keywords: performance, test
Depends on:
Blocks: 270824
  Show dependency tree
 
Reported: 2009-04-01 10:14 EDT by Frederic Fusier CLA
Modified: 2009-04-15 09:57 EDT (History)
1 user (show)

See Also:

Attachments
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 Frederic Fusier CLA 2009-04-01 10:14:44 EDT 
Verifying results for build I20090331-0901, it appears that there's a 
regression on TypeHierarchyPerfTest#testOpenObjectHierarchy() test.

Looking at history, it seems (at least for 2 of test boxes) that something happened for buildd I20090120-0920...
Comment 1 Markus Keller CLA 2009-04-03 15:49:49 EDT 
The strange thing is that the step in the charts happened on
I20090120-0920 for SLED 10 Sun 1.5.0_10 and Win XP Sun 1.5.0_10
, but already on
I20081202-1812 for RHEL 5.0 Sun 6.0_04

I couldn't reproduce any relevant slowdown between 3.4 and I20090401-1325.
Dani, can you see something?

In HEAD, I made the fingerprint testOpenObjectHierarchy2() more robust (repeat 10 times) and changed the cold tests to use assertPerformanceInAbsoluteBand(..)
Comment 2 Dani Megert CLA 2009-04-06 09:28:55 EDT 
Looked at this and could not reproduce it either. However, looking closer at what happens and at the changes I found out that in 3.4.x the Welcome page obscures all other views even if a view gets explicitly shown. In 3.5 the Welcome page minimizes and the views are visible. Painting of the Type Hierarchy view in 3.4 is therefore affected depending on the graphic driver which explains why it's faster in 3.4 and why we don't see it on our machines.

We now close all views before running the test and will backport this to perf_34x. The problem will persist until the baseline is rebuilt.
Comment 3 Markus Keller CLA 2009-04-06 10:49:12 EDT 
Fixed in TypeHierarchyPerfTest and OrderedTestSuite and backported to perf_34x:
- close perspective and open Java perspective before cold test
- make sure test order is fixed
- measure testOpenObjectHierarchy2() multiple times

(In reply to comment #2)
Thanks for the investigations, Dani.
Comment 4 Frederic Fusier CLA 2009-04-15 09:57:13 EDT 
Verified using I20090414-0800 performance results.