Tor Norbye <tor@netbeans.org> [Tue, 16 Sep 2008 11:39:17 -0700] rev 16762
147342: NullPointerException at org.netbeans.napi.gsfret.source.ClassIndex.getQueries
ffjre@netbeans.org [Tue, 16 Sep 2008 17:40:11 +0000] rev 16761
Automated merge
Tor Norbye <tor@netbeans.org> [Tue, 16 Sep 2008 10:04:30 -0700] rev 16760
146886: [cc] code completion is closed after typing "-"
Tor Norbye <tor@netbeans.org> [Tue, 16 Sep 2008 07:00:59 -0700] rev 16759
133384: NullPointerException at org.netbeans.modules.gsfret.editor.hyperlink.GoToSupport$1.run
Jesse Glick <jglick@netbeans.org> [Tue, 16 Sep 2008 17:17:31 +0200] rev 16758
merged from main
Tor Norbye <tor@netbeans.org> [Mon, 15 Sep 2008 22:57:27 -0700] rev 16757
Go To Declaration unit tests
Tor Norbye <tor@netbeans.org> [Mon, 15 Sep 2008 19:33:30 -0700] rev 16756
Add getPreferences to the HintsManager for use by user-configurable hints
Vadiraj Deshpande <vadirajvd@netbeans.org> [Tue, 16 Sep 2008 04:04:21 +0530] rev 16755
Automated merge with http://hg.netbeans.org/main/
Tor Norbye <tor@netbeans.org> [Mon, 15 Sep 2008 14:27:07 -0700] rev 16754
Fix a bug where the hint to rename a local variable (with wrong camel case conventions,
e.g. argList=>arg_list) didn't include the surrounding method's parameter list. I found the same bug in a couple of other local name related quickfixes (nested block vars, block var reuse) and fixed it there too.
Then I noticed that these fixes (interactive quickfixes which invoke instant-renaming mode) didn't have unit tests of the fixes themselves, only the part which finds the problem. Now that I have preview capability, I modified the GSF infrastructure to also support applying quickfixes for hints that implement the PreviewableFix interface; in that case, it will use the generated preview to generate the golden file. Clean up the introduce method/variable tests to use this mechanism instead of the old way which would use a variable to side effect set the target name.
Tor Norbye <tor@netbeans.org> [Mon, 15 Sep 2008 13:04:18 -0700] rev 16753
Modify GSF hints infrastructure to always (unless turned off by overriding checkAllHintOffsets() to return false) run the hints multiple times. It will both run the hints normally, as well as run the hints on an empty document (AFTER the parse tree has been constructed). This will ensure that the hints always properly check that the offsets computed from the AST don't go outside the current document bounds. (This is a source of errors where users delete a lot of text from the document, and the hints run on a parse tree that still reflects the old positions).