This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
Since a few days Netbeans reports lots of InvalidObjectExceptions that occur for no real cause. Netbeans reports them with the litte red sign in the lower right corner. The logs are appended to this bug report. I am not sure, wether this has something to do with it, but on class of my application (with lots of different classes) now has gets no further help from netbeans. Code completion doesn't work (it only shows the text "Please wait...") and no errors are marked in the code editor. However there is no problem in compiling and running the application. And, only one class has this problems.
Created attachment 33598 [details] messages.log file created by netbeans
Created attachment 33599 [details] another messages.log from netbeans
There are also some IllegalStateExceptions.
I'm consistently getting InvalidObjectException under just about the same circumstances as hupfdule. It happened for the first time while I was editing a file, and NetBeans was simultaneously trying to do its code-checking. It flagged an incomplete section of code as bad, and after I'd added enough to get rid of syntax errors, it remained flagged as an error *much* longer than usual, i.e. the compile took much longer than expected. But it did eventually complete, and the error mark was removed. After I added a few more lines (completed statements with correct syntax), this time, the code-check took even longer. While the code-check was still not completed, I got the InvalidObjectException. I dismissed the error dialog. At this point, the "save file" option was grayed out in the right-mouse menu on the file's editor tab, though the file had unsaved edits. Eventually, the check succeeded, and I was allowed to save the file. I exited and restarted NetBeans. After the restart, when I selected the file (via its editor tab) I had been editing, its Navigator window now said "No view available". Other files that I had not edited were still ok -- when I selected them, their Navigator info was as usual. I commented out the code I'd added (all statements involved in both events above). That set off another code check. This one never completed, but I haven't gotten an InvalidObjectException (yet). I was going to use cvs to revert the file, and see if that made the problem go away, but...the cvs menu item in the right-mouse-on-file-tab menu was gone! It's still there for other, unedited, files. I've been using this project successfully for months, and successfully edited this same file last week. One difference from other reports of InvalidObjectException: I have code completion turned off. I'll attach my messages.log. Then I'm going to go look for damage to project files or cvs files in the project directory.
Try deleting directory with mdr cache. It's located in ${your_userdir}/var/cache/mdrstorage
> Try deleting directory with mdr cache That actually solved the problem (at least for me). Well, I assume in some days/ weeks/month the problem will reoccur, but at least I know a workaround now. Was this already a known issue and is fixed in release 5.5?
Yes, it is known problem. IOE are thrown in different cases. Not sure if your exact case is fixed, in 5.5 you should get it less often, but perhaps you will met it again. That is the bad news. The better one is that in release 6.0 it will be definitely fixed.
There were no apparent problems with project or cvs files. The .java file that was having problems, however, had some odd cvs-related damage: It contained what looked like two nested sets of merge conflict marks: <<<<<<< RemoteGuidance.java <<<<<<< RemoteGuidance.java * @version $Id: RemoteGuidance.java,v 1.78 2006/09/23 07:29:32 ptressel Exp $ ======= * @version $Id: RemoteGuidance.java,v 1.78 2006/09/23 07:29:32 ptressel Exp $ >>>>>>> 1.73.2.11 ======= * @version $Id: RemoteGuidance.java,v 1.78 2006/09/23 07:29:32 ptressel Exp $ >>>>>>> 1.75.2.14 However, I'd not done any merge recently. Does NetBeans do an update as part of its cvs "show changes" check? I was assuming so, but also assuming NetBeans would have the merge result go to some other file name, or else would clean up any change to the real file resulting from the update. This bit of damage happened to be inside a comment, so should not have affected the compile...and the first few compiles after the problem started did complete eventually. In the past, I've run into compilers (for other languages) that were not ignoring everything inside comments, e.g. they'd complain about unmatched angle brackets or quotes... Could the java compiler have acquired such a problem with the introduction of <stuff> as a meaningful construct? Anyhow, I edited out the junk above and restarted NetBeans, and the editor is now behaving normally.
Javacore module was replaced by Retouche infrastructure. This bug is not valid in trunk any more.
Reorganization of java component