Summary: | DBCS3.5: Classpath resolution fails to honor the 'Class-Path' header of JAR manifest file in DBCS | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Product: | [Eclipse Project] JDT | Reporter: | Masaihko Maedera <maedera> | ||||||||
Component: | Core | Assignee: | Olivier Thomann <Olivier_Thomann> | ||||||||
Status: | CLOSED FIXED | QA Contact: | |||||||||
Severity: | normal | ||||||||||
Priority: | P3 | CC: | camle, david_audel, kitlo, Olivier_Thomann, pwebster | ||||||||
Version: | 3.5 | ||||||||||
Target Milestone: | 3.5 M7 | ||||||||||
Hardware: | PC | ||||||||||
OS: | Windows XP | ||||||||||
Whiteboard: | |||||||||||
Attachments: |
|
Description
Masaihko Maedera
2009-04-16 04:07:01 EDT
The Manifest file needs to be read using UTF-8 encoding. This is the default encoding on most Linux distribution and this would explain why it worked on Linux. I am investigating. Yes, this is it. Manifest are read using the platform default encoding. Working on it. Created attachment 132278 [details]
Proposed fix
The output is step11/12 depends what is the encoding used to display the file contents. The patch is fixing the retrieval of chained jars in MANIFEST entries. Created attachment 132292 [details]
Proposed fix + regression test
Changed the previous patch to make the UTF-8 encoding usage inside the analysis code.
Adapted all existing tests and added regression test:
org.eclipse.jdt.core.tests.compiler.regression.ManifestAnalyzerTest#testWithOneJarUsingUTF8Name
Released for 3.5M7. Verified for 3.5M7 using I20090428-0100 I verified this bug had been fixed on eclipse-SDK-3.5M7-win32.zip (I20090430-2300). Thank you for your fix. I will close this bug report. |