Bug 126 – Fiji quits unexpectedly when windows are tiled

Bug 126 - Fiji quits unexpectedly when windows are tiled


Summary:	Fiji quits unexpectedly when windows are tiled

Status:	RESOLVED LATER

Product:	Fiji
Classification:	Unclassified
Component:	ImageJ1
Version:	unspecified
Hardware:	Macintosh Mac OS

Importance:	P4 normal
Assigned To:	ImageJ Bugs Mailing List

URL:

Depends on:
Blocks:

Reported:	2009-12-29 00:22 CST by Patrick Dummer
Modified:	2011-03-23 12:27 CDT
CC List:	1 user (show)

See Also:

Description Patrick Dummer 2009-12-29 00:22:41 CST

about 30% of the time I attempt to tile the open windows in fiji, the program quits unexpectedly and I have to relaunch. I typically have no more than 7 windows open at a time and change the view mode to tile immediately upon open, so I haven't lost any unsaved data yet, nor do I know if I would were this glitch to occur after some editing. I'm currently running ImageJA 1.43m on a Macbook with OS X 10.5.8, 2GHz Intel Processor, and 2 GB Memory.

Comment 1 G. Landini 2009-12-29 11:41:51 CST

It does not happen here under linux:
Load the sample mri.tif stack
Stack to Images
Tile windows 
All seems OK, I have done this several times. Maybe a MacOS-only problem?

Comment 2 Christophe Leterrier 2009-12-29 12:09:50 CST

(In reply to comment #1)
> It does not happen here under linux:
> Load the sample mri.tif stack
> Stack to Images
> Tile windows 
> All seems OK, I have done this several times. Maybe a MacOS-only problem?
> 

It definitely is an OSX problem. The same problem is seen in ImageJ. It does not happen all the time and does not seem to be very deterministic. It is here for a very long time (at least the four years since I started using ImageJ), with all OS X versions and all Java updates. I stopped using the Tile Windows command for a long time now because of this problem. It reminds me of the crash when using Import Sequence or Export as Image Sequence that also occurs under OSX only and is difficult to reproduce.

Comment 3 Jean-Yves Tinevez 2009-12-29 13:12:37 CST

If I can add to it:

Under MacOSX, doing it with the MRI-stack does not generate the bug, even if I do it a lot of time. (Stack-to-images, then Tile).

However, it systematically crashes when I do it with the T1-head, which is a much larger stack. 

Maybe it is a memory management issue?

What to do with this bug then? If we can't solve it?

Comment 4 G. Landini 2009-12-29 15:06:28 CST

> What to do with this bug then? If we can't solve it?

WONTFIX? but it would be good if it can be resolved.

Has anybody checked memory use before the crashing?

Does anything come up when running n Debug mode? (Edit>Options>Misc>Debug_mode) 

Has anybody checked memory use before the crashing?

Is this "number of images" problem, rather than image data size?

Wayne develops mainly in a Mac I think, has anybody asked him whether he has seen this earlier?

Comment 5 Johannes Schindelin 2010-02-03 17:20:57 CST

Is it possible that this is the same issue as http://pacific.mpi-cbg.de/cgi-bin/bugzilla/﷒0﷓ and http://pacific.mpi-cbg.de/cgi-bin/bugzilla/﷒1﷓ i.e. a MacOSX-specific bug that shows a race condition in the AWT component?

Comment 6 Johannes Schindelin 2011-03-23 12:27:51 CDT

I had some success with installing OpenJDK (which is a bit involved, though) on some user's machine, which happened to "fix" the problem.

For the moment, I cannot do anything else than marking this "Later", since we cannot do anything about bugs deep in Apple's Java.

There is some hope in the far distant future, though: http://openjdk.java.net/projects/macosx-port/ purports to be an Apple-backed port of OpenJDK that will eventually support both X11 and Aqua.