Created attachment 9711 [details]
Stack trace

Also check in 3.5, probably the same.

Defect: UIManager.getColor("Table.selectionBackground") returns null.
The fix would be easy. Jesse, is it so urgent to integrate to nb3.5?

I just filed JDK bug 4844816 for another problem in GTK L&F.  They
will need to fix this - it's an incompatibility between JTable and 
the GTK L&F - either JTable should test to see if UIManager.getColor()
is returning null for this arg, or (better), the L&F should use 
UIManager.  I'd say file a JDK bug.  

I filed a bugreport 4844870 as a blocker this issue.

Jirko, if you need a temporary workaround for this issue, add it
after the code in Main that calls NbTheme, smth like:

Object o = UIManager.get ("Table.whatever");
if (o == null) UIManager.put (new ColorUIResource (whatever));

That will at least get it working so GTK L&F is usable on 3.5 (yes,
they need to fix it for the release).

Created attachment 9846 [details]
a workaround Gtk bug 4844870

Created attachment 9847 [details]
jar to lib/patches

Not sure why this was marked as being a cross-platform bug, unless a
similar problems happens on other operating systems...?

The attached workaround suppresses the thrown exception when
Tool|Options dialog is being opened. It doesn't fix the reported jdk's
 bug 4844870 but only its known symptom in TreeTableView. I'm not sure
if it's enough to integrate to nb code line

reproducible on Sol8/CDE

A comment from the JDK engineer working on the fix of #4844870:

----------------------%<-------------------------
I can fix the setSelectionForeground/setSelectionBackground color, but
then you guys are invoking setGridColor(null) from TreeTableView which
will throw an  IllegalArgumentException:

java.lang.IllegalArgumentException: New color is null
	at javax.swing.JTable.setGridColor(Unknown Source)
	at
org.openide.explorer.view.TreeTableView.initializeTreeTable(TreeTableView.java:292)

Which means you guys will need to change your code.
Is that acceptable?
----------------------%<-------------------------

Jirka, could you please re-evaluate?

A fix for #4844870 should be provided in final version of 1.4.2.
Nevertheless, the "workaround" proposed by Jiri should be integrated
into trunk, because the current code is depending on an assumption
which is not guaranteed by Swing.

No longer urgent for 3.5. P2 -> P3.

Proposed checks were added to all calling UIManger for colors.
Integrated to maintrunk.

Date: 03/04/15 07:03:53
Revision  Path
1.23      openide/src/org/openide/explorer/view/NodeRenderer.java
1.21      openide/src/org/openide/explorer/view/TableSheet.java
1.10      openide/src/org/openide/explorer/view/TableSheetCell.java
1.25      openide/src/org/openide/explorer/view/TreeTable.java
1.46      openide/src/org/openide/explorer/view/TreeTableView.java

*** Issue 33381 has been marked as a duplicate of this issue. ***

*** Issue 33381 has been marked as a duplicate of this issue. ***

closed