This seems to be triggered by having a modal (i.e. blocking input to other windows) dialog launch another modal dialog.

It sounds as if Johannes is already on the case, but I'm afraid I can't reproduce this on my Mac.  What versions of Mac OS and Java is this seen on?

(In reply to comment #2)
> It sounds as if Johannes is already on the case, but I'm afraid I can't
> reproduce this on my Mac.  What versions of Mac OS and Java is this seen on?
> 
I could not reproduce it for Shape-based averaging. But it still fails for Quantile Based normalization (v1.v).
Here are the details of the machine incriminated:
MacPro, running MacOSX 10.5.6.
Java is:
Java properties applets can read:
  java.version: 1.5.0_16
  java.vendor: Apple Inc.
  mrj.version: 1050.1.5.0_16-284
  os.name: Mac OS X
  os.version: 10.5.6
  os.arch: x86_64
  file.separator: /
  path.separator: :
  line.separator: <lf>

I tried to find a Mac with Mac OS later than 10.4 today, but apparently I don't have a login on any computer with 10.5 or later in the lab.  However, just in case the fix is as simple as making the original dialogs non-modal, as Johannes suggests might be the case, I've pushed a branch modeless-fix-for-mac-os of VIB:

  http://pacific.mpi-cbg.de/cgi-bin/gitweb.cgi?p=VIB.git;a=commit;h=cfffe3433011c445e4aa6c73342233e46b08d60d;hb=refs/heads/VIB

Jean-Yves, could you try that out and see if it fixes the bug?

many thanks,
mark

> http://pacific.mpi-cbg.de/cgi-bin/gitweb.cgi?p=VIB.git;a=commit;h=cfffe3433011c445e4aa6c73342233e46b08d60d;hb=refs/heads/VIB
> 
> Jean-Yves, could you try that out and see if it fixes the bug?
> 
Hey
Unfortunately, the symptoms stay the same. 
Can someone confirm?

Oh well.  I'll have access to an appropriate Mac late this afternoon, and I'll have another go then.

Created attachment 6
thread states on plugin hang

I don't properly understand this yet, but here's some more information.  With VIB at d23b0e8d5775602c92 (i.e. using GenericDialog as before) the plugin seems to be waiting forever on AWTTreeLock.  (I've attached the kill -3 dump of the thread states.)  I think this is something to do with the ScrollPane used in vib.app.gui.FileGroupDialog, since with the following patch (just adding the List directly, rather than in a ScrollPane) it doesn't happen:

iff --git a/vib/app/gui/FileGroupDialog.java b/vib/app/gui/FileGroupDialog.java
index 47f00fe..dc75049 100644
--- a/vib/app/gui/FileGroupDialog.java
+++ b/vib/app/gui/FileGroupDialog.java
@@ -76,15 +76,12 @@ public class FileGroupDialog extends Panel
                gridbag.setConstraints(name, c);
                this.add(name);
 
-               ScrollPane scroll = new ScrollPane();
-               scroll.add(list);
-               scroll.setPreferredSize(new Dimension(300,100));
                c.gridx = 0;
                c.gridy++;
                c.fill = GridBagConstraints.BOTH;
                c.weightx = c.weighty = 0.5;
-               gridbag.setConstraints(scroll, c);
-               this.add(scroll);
+               gridbag.setConstraints(list, c);
+               this.add(list);
 
                Panel buttons = new Panel(new GridLayout(3,1));
                add = new Button("Add to files");

(In reply to comment #8)
> I don't properly understand this yet, but here's some more information.  With
> VIB at d23b0e8d5775602c92 (i.e. using GenericDialog as before) the plugin seems
> to be waiting forever on AWTTreeLock.  (I've attached the kill -3 dump of the
> thread states.)  I think this is something to do with the ScrollPane used in
> vib.app.gui.FileGroupDialog, since with the following patch (just adding the
> List directly, rather than in a ScrollPane) it doesn't happen:
> 

I could not apply the patch. 
Can you push it to the modeless branch as before?

(In reply to comment #9)
> (In reply to comment #8)
> > I don't properly understand this yet, but here's some more information.  With
> > VIB at d23b0e8d5775602c92 (i.e. using GenericDialog as before) the plugin seems
> > to be waiting forever on AWTTreeLock.  (I've attached the kill -3 dump of the
> > thread states.)  I think this is something to do with the ScrollPane used in
> > vib.app.gui.FileGroupDialog, since with the following patch (just adding the
> > List directly, rather than in a ScrollPane) it doesn't happen:
> > 
> 
> I could not apply the patch. 
> Can you push it to the modeless branch as before?
> 

I don't think the modeless branch is relevant any more, so I've pushed it to a new branch, called "scrollpane-non-fix":

  http://pacific.mpi-cbg.de/cgi-bin/gitweb.cgi?p=VIB.git;a=commitdiff;h=6b5139ee43075844072aaffaa99766efc9a5d67b

Ok this fix works.
The plugin does not crash Fiji, but I could not test it, because I do not know what it does :)

Note that it's not really a fix, since the dialog is unusable for more than a few images without the scroll bars.

I have more urgent things to deal with, unfortunately, so if someone had time (and an available Mac OS X 10.5 machine( in order to investigate why there's the hang in first place that would be great.  Otherwise I'll just disable them on Mac OS 10.5 (with a warning dialog explaining) for this release.

We can remove them from staged-plugin, and see if the updated will be able to put them back when they are ready.
They are anyway not documented.

(In reply to comment #13)
> We can remove them from staged-plugin, and see if the updated will be able to
> put them back when they are ready.
> They are anyway not documented.

Shape_Based_Averaging should be removed for this release, but for different reasons, as discussed in IRC.

Quantile_Based_Normalization *is* documented, however:

  http://homepages.inf.ed.ac.uk/s9808248/imagej/quantile-normalization/

... which has been pointed to by this page for some time.

  http://pacific.mpi-cbg.de/wiki/index.php/Quantile_Based_Normalization

Please don't just remove plugins that people use because they don't work on particular versions of Mac OS.  What I suggested in comment 12 will be fine until I have more time to look at this.  (That'll certainly be after I've come back from holiday, on the 12th of October.)

Oops sorry. Then i did not get what you meant by disabling...

I've just pushed something that does that selective disabling to the VIB branch of VIB:

  http://pacific.mpi-cbg.de/cgi-bin/gitweb.cgi?p=VIB.git;a=commitdiff;h=c8982b4d4c7fad8b9fc85bc0b278396f30cfd601

I'll look at this more when I'm back, and hopefully feeling better.

(In reply to comment #16)
> I've just pushed something that does that selective disabling to the VIB branch

The selective disabling work. It is not in new release tough.

I just stumbled upon this bug because I tried to run "Quantile Based Normalization" on a Mac, and it is disabled.
Since the report has been a while ago, I wondered if the cause of this bug maybe disappeared in the meantime.

Any new information?

Hi Jan,

I'm travelling today, but I'll try to get access to a Mac tomorrow to see whether this bug still exists.

regards,
mark