You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@harmony.apache.org by "Jim Yu (JIRA)" <ji...@apache.org> on 2009/02/26 09:33:01 UTC
[jira] Created: (HARMONY-6101) [classlib][luni] Extraneous path
separator in *.boot.class.path on IBM VME 5
[classlib][luni] Extraneous path separator in *.boot.class.path on IBM VME 5
----------------------------------------------------------------------------
Key: HARMONY-6101
URL: https://issues.apache.org/jira/browse/HARMONY-6101
Project: Harmony
Issue Type: Bug
Components: Classlib
Reporter: Jim Yu
By using IBM VME 5, while running SystemTest.test_getProperty_bootClassPath test case on
both Linux and Windows, there is a failure:
Boot class path contains extra patch separator.
In the boot class path, I can see an extra ':' on Linux after:
deploy/jdk/jre/bin/default/security-kernel.jar
Since similar issue has been resolved in HARMONY-3298 and the test can pass on IBM VME 4.
I suppose this may be caused by the VME 5. After my further investigation, I found out that the root
cause of this issue is a behavior change by VME 5. For VME 4, there is no CP separator at the end
of the boot strap class path set by VM. So it is OK for Harmony to extend the boot class path with
values retrieved from bootclasspath.properties begins with a CP separator . But for VME 5, root class
path set by VM will add a CP separator at the end, so an repeated CP separator is generated. Though
it was caused by the change of VME, I find out a tricky method to make Harmony work well on both
VME 4 and VME 5. I've tested my patch. It works fine on VME4, VME5 and DRLVM.
Can someone take a look at this? Thanks.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HARMONY-6101) [classlib][luni] Extraneous path
separator in *.boot.class.path on IBM VME 5
Posted by "Jim Yu (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HARMONY-6101?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12676979#action_12676979 ]
Jim Yu commented on HARMONY-6101:
---------------------------------
Hi Pavel,
Sorry for the misprint, I've updated my patch. Since this work around only adds a check, it will not affect the usage of current VM.
I think it can be filed at any appropriate time.
For your concerns:
If there are double separators contained in the boot class path, provided by a VM, it is definitely a VM defect. Currently,
the double separators were not introduced by the VME 5 alone. It resulted from the VM added one separator at the end
of the root class path, provided by itself, and classlib added one simultaneously at the start of the boot class path retrieved
from properties. So I think we just need to check whether there is a separator at the end of the root class path, provided by VM.
> [classlib][luni] Extraneous path separator in *.boot.class.path on IBM VME 5
> ----------------------------------------------------------------------------
>
> Key: HARMONY-6101
> URL: https://issues.apache.org/jira/browse/HARMONY-6101
> Project: Harmony
> Issue Type: Bug
> Components: Classlib
> Reporter: Jim Yu
> Attachments: HARMONY-6101_V2.diff
>
>
> By using IBM VME 5, while running SystemTest.test_getProperty_bootClassPath test case on
> both Linux and Windows, there is a failure:
> Boot class path contains extra patch separator.
> In the boot class path, I can see an extra ':' on Linux after:
> deploy/jdk/jre/bin/default/security-kernel.jar
> Since similar issue has been resolved in HARMONY-3298 and the test can pass on IBM VME 4.
> I suppose this may be caused by the VME 5. After my further investigation, I found out that the root
> cause of this issue is a behavior change by VME 5. For VME 4, there is no CP separator at the end
> of the boot strap class path set by VM. So it is OK for Harmony to extend the boot class path with
> values retrieved from bootclasspath.properties begins with a CP separator . But for VME 5, root class
> path set by VM will add a CP separator at the end, so an repeated CP separator is generated. Though
> it was caused by the change of VME, I find out a tricky method to make Harmony work well on both
> VME 4 and VME 5. I've tested my patch. It works fine on VME4, VME5 and DRLVM.
> Can someone take a look at this? Thanks.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HARMONY-6101) [classlib][luni] Extraneous path
separator in *.boot.class.path on IBM VME 5
Posted by "Jim Yu (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HARMONY-6101?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jim Yu updated HARMONY-6101:
----------------------------
Attachment: HARMONY-6101_V2.diff
> [classlib][luni] Extraneous path separator in *.boot.class.path on IBM VME 5
> ----------------------------------------------------------------------------
>
> Key: HARMONY-6101
> URL: https://issues.apache.org/jira/browse/HARMONY-6101
> Project: Harmony
> Issue Type: Bug
> Components: Classlib
> Reporter: Jim Yu
> Attachments: HARMONY-6101_V2.diff
>
>
> By using IBM VME 5, while running SystemTest.test_getProperty_bootClassPath test case on
> both Linux and Windows, there is a failure:
> Boot class path contains extra patch separator.
> In the boot class path, I can see an extra ':' on Linux after:
> deploy/jdk/jre/bin/default/security-kernel.jar
> Since similar issue has been resolved in HARMONY-3298 and the test can pass on IBM VME 4.
> I suppose this may be caused by the VME 5. After my further investigation, I found out that the root
> cause of this issue is a behavior change by VME 5. For VME 4, there is no CP separator at the end
> of the boot strap class path set by VM. So it is OK for Harmony to extend the boot class path with
> values retrieved from bootclasspath.properties begins with a CP separator . But for VME 5, root class
> path set by VM will add a CP separator at the end, so an repeated CP separator is generated. Though
> it was caused by the change of VME, I find out a tricky method to make Harmony work well on both
> VME 4 and VME 5. I've tested my patch. It works fine on VME4, VME5 and DRLVM.
> Can someone take a look at this? Thanks.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HARMONY-6101) [classlib][luni] Extraneous path
separator in *.boot.class.path on IBM VME 5
Posted by "Jim Yu (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HARMONY-6101?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jim Yu updated HARMONY-6101:
----------------------------
Attachment: HARMONY-6101.diff
> [classlib][luni] Extraneous path separator in *.boot.class.path on IBM VME 5
> ----------------------------------------------------------------------------
>
> Key: HARMONY-6101
> URL: https://issues.apache.org/jira/browse/HARMONY-6101
> Project: Harmony
> Issue Type: Bug
> Components: Classlib
> Reporter: Jim Yu
> Attachments: HARMONY-6101.diff
>
>
> By using IBM VME 5, while running SystemTest.test_getProperty_bootClassPath test case on
> both Linux and Windows, there is a failure:
> Boot class path contains extra patch separator.
> In the boot class path, I can see an extra ':' on Linux after:
> deploy/jdk/jre/bin/default/security-kernel.jar
> Since similar issue has been resolved in HARMONY-3298 and the test can pass on IBM VME 4.
> I suppose this may be caused by the VME 5. After my further investigation, I found out that the root
> cause of this issue is a behavior change by VME 5. For VME 4, there is no CP separator at the end
> of the boot strap class path set by VM. So it is OK for Harmony to extend the boot class path with
> values retrieved from bootclasspath.properties begins with a CP separator . But for VME 5, root class
> path set by VM will add a CP separator at the end, so an repeated CP separator is generated. Though
> it was caused by the change of VME, I find out a tricky method to make Harmony work well on both
> VME 4 and VME 5. I've tested my patch. It works fine on VME4, VME5 and DRLVM.
> Can someone take a look at this? Thanks.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HARMONY-6101) [classlib][luni] Extraneous path
separator in *.boot.class.path on IBM VME 5
Posted by "Jim Yu (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HARMONY-6101?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jim Yu updated HARMONY-6101:
----------------------------
Attachment: (was: HARMONY-6101.diff)
> [classlib][luni] Extraneous path separator in *.boot.class.path on IBM VME 5
> ----------------------------------------------------------------------------
>
> Key: HARMONY-6101
> URL: https://issues.apache.org/jira/browse/HARMONY-6101
> Project: Harmony
> Issue Type: Bug
> Components: Classlib
> Reporter: Jim Yu
>
> By using IBM VME 5, while running SystemTest.test_getProperty_bootClassPath test case on
> both Linux and Windows, there is a failure:
> Boot class path contains extra patch separator.
> In the boot class path, I can see an extra ':' on Linux after:
> deploy/jdk/jre/bin/default/security-kernel.jar
> Since similar issue has been resolved in HARMONY-3298 and the test can pass on IBM VME 4.
> I suppose this may be caused by the VME 5. After my further investigation, I found out that the root
> cause of this issue is a behavior change by VME 5. For VME 4, there is no CP separator at the end
> of the boot strap class path set by VM. So it is OK for Harmony to extend the boot class path with
> values retrieved from bootclasspath.properties begins with a CP separator . But for VME 5, root class
> path set by VM will add a CP separator at the end, so an repeated CP separator is generated. Though
> it was caused by the change of VME, I find out a tricky method to make Harmony work well on both
> VME 4 and VME 5. I've tested my patch. It works fine on VME4, VME5 and DRLVM.
> Can someone take a look at this? Thanks.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HARMONY-6101) [classlib][luni] Extraneous path
separator in *.boot.class.path on IBM VME 5
Posted by "Pavel Pervov (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HARMONY-6101?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12676933#action_12676933 ]
Pavel Pervov commented on HARMONY-6101:
---------------------------------------
I have one major concern - shouldn't we file this against VME5 and wait for the next snapshot?
And several concerns about the patch:
1) Probably, misprint: endsWithCPSeprator -> endsWithCPSeparator;
2) If we add workaround into class library, then we have to check all boot class path, provided by a VM, so it would not contain double separators inside it - not only at the end.
> [classlib][luni] Extraneous path separator in *.boot.class.path on IBM VME 5
> ----------------------------------------------------------------------------
>
> Key: HARMONY-6101
> URL: https://issues.apache.org/jira/browse/HARMONY-6101
> Project: Harmony
> Issue Type: Bug
> Components: Classlib
> Reporter: Jim Yu
> Attachments: HARMONY-6101.diff
>
>
> By using IBM VME 5, while running SystemTest.test_getProperty_bootClassPath test case on
> both Linux and Windows, there is a failure:
> Boot class path contains extra patch separator.
> In the boot class path, I can see an extra ':' on Linux after:
> deploy/jdk/jre/bin/default/security-kernel.jar
> Since similar issue has been resolved in HARMONY-3298 and the test can pass on IBM VME 4.
> I suppose this may be caused by the VME 5. After my further investigation, I found out that the root
> cause of this issue is a behavior change by VME 5. For VME 4, there is no CP separator at the end
> of the boot strap class path set by VM. So it is OK for Harmony to extend the boot class path with
> values retrieved from bootclasspath.properties begins with a CP separator . But for VME 5, root class
> path set by VM will add a CP separator at the end, so an repeated CP separator is generated. Though
> it was caused by the change of VME, I find out a tricky method to make Harmony work well on both
> VME 4 and VME 5. I've tested my patch. It works fine on VME4, VME5 and DRLVM.
> Can someone take a look at this? Thanks.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.