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.
Summary: | Code Completion not correctly working, even after Update the Code Completion Database Manager | ||
---|---|---|---|
Product: | editor | Reporter: | vbrabant <vbrabant> |
Component: | Completion & Templates | Assignee: | Martin Roskanin <mroskanin> |
Status: | RESOLVED FIXED | ||
Severity: | blocker | CC: | dkonecny |
Priority: | P3 | ||
Version: | 4.x | ||
Hardware: | PC | ||
OS: | Windows 95/98 | ||
Issue Type: | DEFECT | Exception Reporter: |
Description
vbrabant
2004-05-20 16:09:17 UTC
I have reproduced something similar yet. I create a class One and the class Two. In class One I wrote "new Two().|" => no CC result I have to modified a class Two, save it and return back to One. After that CC after "new Two().|" worked. The same if I add some members to class Two. I.e: add a public void foo() method to Two class, save. Return back to One, invoke "new Two().|", the method foo is not present. I have to modify a Two class (i.e by adding a new method foo2() ) and save the class. After that I see after "new Two().|" the method foo, but foo2 is missing... It looks like some synchro problem. This may be also connected with issue #43520 I have tested release36, it works correctly. We should fix it to promo-d Please note that it was not working for a line, but it was working for another line OF THE SAME FILE, FOR THE SAME METHOD. And that is really strange Re. Martin: I know about this problem and I believed that the same problem was in NB36. But I might be wrong. The problem is that when the file is saved the event comes to editor but at that time the DataObject of the file was not yet reparsed and its SourceCookie is out-of-date. But editor updates PDB with this old source hierarchy. That's why second save solves this. re. Vincent: yes, this is very strange. Wasn't the source modified (accidentaly, by pressing a space or any other key?) during replacing the cursor during testing completion on different places? Unfortunately I cannot reproduce it. So, this will be probably difficult to find out where the problem could be. re. David: I cannot reproduce the problem that I described in my previous comment in NetBeans 3.6. There could be some race condition problem or some other listener misbehaviour, because except the problem mentioned in issue #43520 there is similar problem with deleted classes. If you delete class in explorer, it is still available in completion => it wasn't deleted from CC DB. I believe, these problems should be fixed by meta_stavbicka merge (I cannot reproduce the problem there). That's why I will not fix these problems in maintrunk for now. This should now be resolved as there is no more the ParserDBs updating going on so the problem described by David should no longer happen. Please reopen if the problem would reappear. |