Bug 21230

Summary: Rebuilding project fails with ContextStackOverflow (CompilationResult.quickPrioritize)
Product: [Eclipse Project] JDT Reporter: Tom Krauss <tok>
Component: CoreAssignee: Kent Johnson <kent_johnson>
Status: VERIFIED FIXED QA Contact:
Severity: normal    
Priority: P3 CC: alicec
Version: 2.0   
Target Milestone: 2.0.1   
Hardware: PC   
OS: Windows 2000   
Whiteboard:
Attachments:
Description Flags
.log containing stacktrace of reported error none

Description Tom Krauss CLA 2002-07-04 07:52:58 EDT 
I have several projects in my workspace, which I can't rebuild any more
(Message: "rebuild failed - null")
The problem started to show up, after I switched to JDK 1.4 compiler compliance
(although I am not sure, whether that action was actually causing the problem).
Switching back to JDK 1.3 compiler compliance did not solve the problem. In the
.log file of Eclipse, one can find a ContextStackoverflow caused by
CompilationResult.quickPrioritize. I've attached the .log file containing the
stack trace.
Comment 1 Tom Krauss CLA 2002-07-04 07:54:56 EDT 
Created attachment 1634 [details]
.log containing stacktrace of reported error
Comment 2 Kent Johnson CLA 2002-07-04 12:46:48 EDT 
Tom: Which build/drop are you using?

Does changing the number of problems per compilation unit change the behaviour? 
See Preferences->Java->Compiler->Errors and Warnings...
Comment 3 Philipe Mulet CLA 2002-07-04 17:47:37 EDT 
If quicksort is looping, then we might have the same bug in another quick* 
method. Need to investigate.
Comment 4 Kent Johnson CLA 2002-07-05 11:19:09 EDT 
From Tom:

I am using the final 2.0 build (200206271827).

Yes! Setting the Compiler Preferences to not report any deprecations (and there 
are in fact many in one class I am compiling) did fix the problem.
Comment 5 Kent Johnson CLA 2002-07-11 13:23:50 EDT 
The quick sort algorithm was replaced with another (from the search indexing 
class Util) which is log 2 based... the previous one was order N when all the 
items were equal.

It would divide 100 elements into 1-1 and 2-100, 2-2 and 3-100... instead of 1-
49 and 51-100, etc.
Comment 6 Jerome Lanneluc CLA 2002-08-20 07:53:41 EDT 
*** Bug 22392 has been marked as a duplicate of this bug. ***
Comment 7 David Audel CLA 2002-08-21 04:24:39 EDT 
Verified.
Comment 8 David Audel CLA 2002-09-20 09:21:05 EDT 
Verified in 2.1 M1