Bug 190840 - SourceMapper#computeAllRootPaths(IType) factorisation suggestion
Summary: SourceMapper#computeAllRootPaths(IType) factorisation suggestion
Status: VERIFIED FIXED
Alias: None
Product: JDT
Classification: Eclipse Project
Component: Core (show other bugs)
Version: 3.3   Edit
Hardware: PC Windows XP
: P3 normal (vote)
Target Milestone: 3.5 M1   Edit
Assignee: Jerome Lanneluc CLA
QA Contact:
URL:
Whiteboard:
Keywords: performance
Depends on:
Blocks:
 
Reported: 2007-06-04 12:26 EDT by Eric Jodet CLA
Modified: 2008-09-16 10:07 EDT (History)
1 user (show)

See Also:

Attachments
Proposed fix and performance test (10.58 KB, patch)
2008-06-10 11:06 EDT, Jerome Lanneluc CLA 		no flags Details | Diff
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 Eric Jodet CLA 2007-06-04 12:26:29 EDT 
Version: 3.3.0
Build id: I20070601-1539

see bug 190094 for the context.

Test scenario:
- have a type that inherits a class located in a big JAR
- open this type in the Java editor, press Crtl+O twice to show inherited members

As there is no source attached to the referenced JAR file, the SourceMapper#computeAllRootPaths(IType) method could be optimized, as we scan
the same zip file twice:
once to search for class files,
once to search for source files.

If no source attached to JAR file, then we'll be scanning the same JAR twice.
During the first scan, we could trigger a flag to remember this JAR has no source in it.
Comment 1 Eric Jodet CLA 2007-06-04 12:35:46 EDT 
(In reply to comment #0)
see also bug 190842
Comment 2 Jerome Lanneluc CLA 2007-06-20 10:21:56 EDT 
Performance work
Comment 3 Jerome Lanneluc CLA 2008-05-07 07:53:54 EDT 
Deferring post 3.4
Comment 4 Jerome Lanneluc CLA 2008-06-10 11:06:13 EDT 
Created attachment 104326 [details]
Proposed fix and performance test

To verify, one can run the performance test locally with and without the fix in SourceMapper
Comment 5 Jerome Lanneluc CLA 2008-06-25 09:22:16 EDT 
Fix and performance test released for 3.5M1.

Still need to release performance test to 3.5 performance baseline.
Comment 6 Jerome Lanneluc CLA 2008-06-27 06:51:21 EDT 
Performance test released in perf_34x branch.
Comment 7 Olivier Thomann CLA 2008-08-08 10:57:58 EDT 
Will be verified once the performance test results are back.
Comment 8 Frederic Fusier CLA 2008-09-16 10:07:14 EDT 
Verified for 3.5M2 using local JDT/Core performance tests.

Here's the results noticed on our test box for the added performance test testGetSourceBigJarNoAttachment():

Version		Value
----------------------
M1		22.62%
v_908		23.07%
v_909		23.79%
v_910		23.86%
v_911		20.97%
v_912		19.16%