Build: NetBeans IDE Dev (Build 080214)
VM: Java HotSpot(TM) Client VM, 1.6.0_10-ea-b07
OS: SunOS, 5.11, x86
try to open help using F1 or using menu Help / Contents
Created attachment 56664 [details]
JavaHelp issue. I filed issue https://javahelp.dev.java.net/issues/show_bug.cgi?id=22 against JavaHelp. Closing here as
we cannot fix it. Hiding exception is probably not good solution as Help window creation fails so pressing F1 or
selecting Help -> Contents (or any other Help menu item) will fail in sich case.
NetBeans IDE 6.1 Beta (Build 200803031027)
This issue has already 50 duplicates
This issue has already 100 duplicates
It seems this is related to a MissingResourceException, somewhere inside NetBeans
Please see http://article.gmane.org/gmane.comp.java.netbeans.modules.openide.devel/36588
I'm reopening it, requesting for an evaluation.
We have xml logs from exception reporter but I do not see MissingResourceException there. So your link showing MRE seems
just coincidence. It looks like some race condition in JavaHelp to me. If you have any sample module or any way how to
reproduce this and show that original cause is MRE please send it to me. Without that I am not able to do anything.
Product Version: NetBeans IDE Dev (Build 200812170242)
Java: 1.6.0_06; Java HotSpot(TM) Client VM 10.0-b22
System: Windows XP version 5.1 running on x86; MS932; ja_JP (nb)
I can't reproduce the issue, probably it depends on jdk version.
I was able to use F1, Help->Help Contents, Help button from dialogs.
Help window was opened properly.
So far this issue looks dependent on JH #22. Will see what's possible to do for 7.0.
*** Issue 155325 has been marked as a duplicate of this issue. ***
Can't reproduce. Need additional data: ideally the copy of NB & user dir, as minimum all jar's that contains *.hs files
and full description of build (date, origin, type, etc.).
The most probable reason of this bug are broken modules obtained from update center. Currently the Jindent module
provoke this error.
Yes, looks like in some cases this issue is caused by plugins from UC.
Can we workaround, so a plug-in problem would not be an IDE problem ? Or improve diagnostics at least.
Technically - yes. But is it correct from methodic point of view? In case we have incorrectly built module I think it
would be better to elimanate a problem with module rather than hide exception.
Diagnostic message had recently been improved to show help set caused the problem.
It's not about hiding the exception but about showing Help.
Two statements are here:
- Keep IDE in a good shape even if some plugin has a problem with HelpSet
- Provide diagnostic message about a problem in certain plugin.
Good to see that 2nd is done. Is 1st technically doable w/o JH modifications ?
I don't exprect problems with changing exception to message and isolating bad helpset. Involved code is already
subclassed. Just not sure that such failure-resistance is really nedeed.
This exception was reported more than 260 times since 6.0 so failure resistance is highly appreciated.
Moving JH issues to Victor.
definitely not INCOMPLETE ...
owner, if you improved message it should be enough for now (I would say) ... so close as FIXED ... we'll wait for new
exception and report all those excs to particular modules.
The latest build when this exception is reproduced is 081110.
This issue claims that the IAE has occurred on the Solaris platform.
The attached stack trace shows that the JavaHelp implementation can't load SearchEngine via the reflection mechanism in
private SearchEngine makeEngine(NavigatorView view) of the javax.help.search.MergingSearchEngine class
Please note, in our case there are guaranties that non-null argument "view" is specified when the method is invoked.
All points mentioned above give me an idea that a cause of this issue is the same as for the Issue #159586 , but it has
AFAIK the original problem was that in very different cases the very same exception was thrown. Now we should be able to
deal with each on case by case basis.
Applied upstream as rev 76.
3 new duplicates in dev builds
*** Bug 187702 has been marked as a duplicate of this bug. ***
Created attachment 104522 [details]
Created attachment 104781 [details]
Created attachment 104857 [details]
hit the help button on the plugins window.
The current selection was 'Java Web and EE' in the Installed tab
Created attachment 105093 [details]
Created attachment 105595 [details]
Just a few reports recently, decreasing priority. Not a P2.
Created attachment 108152 [details]
I cannot avoid this exception. Quick search is unusable.
Created attachment 108189 [details]
- unzip IDE from http://bertram.netbeans.org/hudson/job/jet-main/3979/artifact/nbbuild/NetBeans-dev-jet-main-3979-on-20110508-full.zip
- disable ergonomics cluster in etc/netbeans.clusters
- run IDE
- call "Help|Help Contents" from main menu and exception is thrown
Working fine for me in a fresh dev build from core-main (cluster.config=full, ergonomics cluster deleted) on Ubuntu / JDK 6u24.
It fails for me just with builds from bertram.netbeans.org. Lastly I tried jet-main-3998-on-20110512. I don't know what is different on these builds.
Created attachment 108244 [details]
Correct jar from daily build.
The case I am able to reproduce is caused by corrupted/missing content in org/netbeans/modules/apisupport/project/docs/JavaHelpSearch in org-netbeans-modules-apisupport-kit.jar. I see another difference in manifest.mf but I don't know if this is the reason.
- bertram manifest
Ant-Version: Apache Ant 1.7.1
Created-By: 11.0-b12 (Sun Microsystems Inc.)
- daily manifest
Ant-Version: Apache Ant 1.8.0
Created-By: 1.6.0_18-b07 (Sun Microsystems Inc.)
Created attachment 108245 [details]
Incorrect jar from bertram.
> jbecicka: I cannot avoid this exception. Quick search is unusable.
> jskrivanek: Working fine for me in a fresh dev build. The case I am able to reproduce is caused by corrupted/missing content in org/netbeans/modules/apisupport/project/docs/JavaHelpSearch in org-netbeans-modules-apisupport-kit.jar.
+ only three reports from users in the exception report list in the last three months
(In reply to comment #41)
> The case I am able to reproduce is caused by corrupted/missing content in
> org/netbeans/modules/apisupport/project/docs/JavaHelpSearch in
Indeed, the JAR is corrupt. Also broken in apisupport.kit/build/classes/, so the problem is in indexing, not creating the JAR.
> I see another difference in
> manifest.mf but I don't know if this is the reason.
> - bertram manifest
> Ant-Version: Apache Ant 1.7.1
> Created-By: 11.0-b12 (Sun Microsystems Inc.)
> - daily manifest
> Ant-Version: Apache Ant 1.8.0
> Created-By: 1.6.0_18-b07 (Sun Microsystems Inc.)
Well I tried building this module locally using Ant 1.7.1 and it looked OK to me. Same when building using JDK 6u10.
Also looks broken on web-main-test-dist, built on bertram master; and main-silver-test-dist, built on beefy. Looks OK on NB-Core-Build, on deadlock master, and nbms-and-javadoc, on a slave.
Maybe a Solaris-specific problem? But then I would expect every production build to be broken.
Console log shows tons of exceptions such as:
Note: Indexing 11 HTML files in .../maven/src/org/netbeans/modules/maven/docs into org/netbeans/modules/maven/docs/JavaHelpSearch/
java.lang.Exception: ConfigFile and/or IndexBuilder not set
Treating this as a P2 - build products from bertram are broken.
(In reply to comment #44)
> at com.sun.java.help.search.DefaultIndexerKit.storeToken(DefaultIndexerKit.java:145)
This line number does not correspond to JH 2.0_05. It seems that JDKs on bertram have some obsolete version of JH in their classpath?
On http://bertram.netbeans.org/hudson/script entering
println(new ProcessBuilder(['sh', '-c', '/space/soft/jdk1.6.0_10/bin/jrunscript -e "println(new com.sun.java.help.search.Indexer().getClass().protectionDomain.codeSource.location)"']).start().text)
points to the culprit:
This seems to be 2.0_03, despite the file name.
I am still not sure how this gets into the default Java classpath, but it must be removed.
I downloaded that copy of jhall.jar and tried
ANT_OPTS=-Xbootclasspath/a:/tmp/jhall.jar ant -f apisupport.kit/build.xml clean netbeans
which I can confirm loaded Indexer.class from the mentioned JAR, but the resulting JavaHelpSearch dir looked fine. So I am not sure how to reproduce this problem locally.
I have understood from comments above that this is not the production bits issue. If so, I would say this is not P2.
I have no access to bertram machine itself. Will try to find out somebody who can update JH there.
(In reply to comment #48)
> this is not the production bits issue.
> If so, I would say this is not P2.
I am not sure. It is a serious error thrown when the IDE is built in certain environments, with no warning or explanation.
(As mentioned in comment #47, I do not know how to reproduce on my own computer. If I did, I would just make the build fail in this circumstance.)
>I would say this is not P2.
This is P2 according bug priority guidelines:
* Affects another developer's progress
Affects my progress. I'm trying to fix issues in quick search and I'm always getting this exception.
* The issue has more then 20 exception reports or more then 10 duplicates + votes
This issue has 292 reports.
It doesn't affect end-users of IDE thus doesn't sense to fix it in NB701.
Created attachment 109663 [details]
Created attachment 111303 [details]
Ant build error - stacktrace
I have just seen this error when building a module on Mac:
Note: Indexing 21 HTML files in /data/work/src/netbeans-cm/db/src/org/netbeans/modules/db/docs into org/netbeans/modules/db/docs/JavaHelpSearch/
java.io.FileNotFoundException: /var/folders/ h/ hnakHZqE10Cpgs4i0FeHk TI/-Tmp-/jhindexer-out241111078710371151/TMAP (No such file or directory)
at java.io.RandomAccessFile.open(Native Method)
(In reply to comment #38)
> To reproduce:
jhavlin says this now works (with NetBeans-dev-jet-main-4920-on-20111026-full.zip).
(In reply to comment #44)
> Also broken in apisupport.kit/build/classes/
No longer broken in the Hudson workspace that I can see.
> java.lang.Exception: ConfigFile and/or IndexBuilder not set
> at com.sun.java.help.search.DefaultIndexerKit.storeToken
BTW I made this a fatal error as of SVN revision 85.
(In reply to comment #46)
Seems to be gone; perhaps the OS was upgraded on these machines? As discussed in comment #47, not clear if that was really the culprit anyway.
(In reply to comment #53)
> I have just seen this error when building a module on Mac:
This was an unrelated bug #201194.
No known remaining way to reproduce; closing.
*** Bug 217555 has been marked as a duplicate of this bug. ***