Ubuntu empty window bug

[Michael Doube]

Hi all

I have struck a bug in today's daily update (1.44f10), which results in
failure to display the contents of a stack, then ImageJ becoming
unresponsive.  I think it appeared around 1.44f9 or 1.44f10.

It appears to only affect tiff stacks as I can drag and drop a directory
of dicom slices as normal.  It also only appears for me on Ubuntu
(10.04, Sun Java 1.6.0_20, 64-bit). Win XP (Sun JDK 1.6.0_21, 32-bit) is
unaffected.

Has anyone else seen this?

Michael

[dscho]

Hi,

On Tue, 17 Aug 2010, Michael Doube wrote:

> I have struck a bug in today's daily update (1.44f10), which results in
> failure to display the contents of a stack, then ImageJ becoming unresponsive.
> I think it appeared around 1.44f9 or 1.44f10.

It appeared in 1.44f9 (I used Git's bisect tool to find out). The problem
was introduced due to an apparent bug fix. You can work around it with
this patch.

-- snipsnap --
diff --git a/ij/ImagePlus.java b/ij/ImagePlus.java
index 3a57cde..4736103 100644
--- a/ij/ImagePlus.java
+++ b/ij/ImagePlus.java
@@ -572,7 +572,7 @@ public class ImagePlus implements ImageObserver, Measurements {
  if (isDisplayedHyperStack()) setOpenAsHyperStack(true);
  win = new StackWindow(this);   // replaces this window
  setPosition(1, 1, 1);
- } else if (dimensionsChanged || sliderChange)
+ } else if (dimensionsChanged /*|| sliderChange*/)
  win.updateImage(this);
  else
  repaintWindow();

[dscho]

Hi,

On Tue, 17 Aug 2010, Johannes Schindelin wrote:

> On Tue, 17 Aug 2010, Michael Doube wrote:
>
> > I have struck a bug in today's daily update (1.44f10), which results
> > in failure to display the contents of a stack, then ImageJ becoming
> > unresponsive. I think it appeared around 1.44f9 or 1.44f10.
>
> It appeared in 1.44f9 (I used Git's bisect tool to find out). The
> problem was introduced due to an apparent bug fix. You can work around
> it with this patch.

Actually, I take it back. It seems to be a deadlock caused sometimes due
to unfortunate timing between threads, and my patch just papered over the
real issue, even if it fixes the symptom some times.

Ciao,
Johannes

[dscho]

Hi,

On Tue, 17 Aug 2010, Johannes Schindelin wrote:

> On Tue, 17 Aug 2010, Johannes Schindelin wrote:
>
> > On Tue, 17 Aug 2010, Michael Doube wrote:
> >
> > > I have struck a bug in today's daily update (1.44f10), which results
> > > in failure to display the contents of a stack, then ImageJ becoming
> > > unresponsive. I think it appeared around 1.44f9 or 1.44f10.
> >
> > It appeared in 1.44f9 (I used Git's bisect tool to find out). The
> > problem was introduced due to an apparent bug fix. You can work around
> > it with this patch.
>
> Actually, I take it back. It seems to be a deadlock caused sometimes due
> to unfortunate timing between threads, and my patch just papered over
> the real issue, even if it fixes the symptom some times.

... [show rest of quote]

I should have known right away when I saw the deadlock. The new
addScrollbars() method is synchronized. It needs to be unsynchronized.

Ciao,
Johannes

[Rasband, Wayne (NIH/NIMH) [E]]

On Aug 17, 2010, at 7:31 AM, Michael Doube wrote:

> Hi all
>
> I have struck a bug in today's daily update (1.44f10), which results in
> failure to display the contents of a stack, then ImageJ becoming
> unresponsive.  I think it appeared around 1.44f9 or 1.44f10.
>
> It appears to only affect tiff stacks as I can drag and drop a directory
> of dicom slices as normal.  It also only appears for me on Ubuntu
> (10.04, Sun Java 1.6.0_20, 64-bit). Win XP (Sun JDK 1.6.0_21, 32-bit) is
> unaffected.
>
> Has anyone else seen this?

This bug is fixed in the 1.44f11 daily build.

-wayne