Re: Image resizing in app -> Sudden death
Re: Image resizing in app -> Sudden death
- Subject: Re: Image resizing in app -> Sudden death
- From: Hugi Thordarson <email@hidden>
- Date: Thu, 12 May 2005 15:23:35 +0000
*sigh* We restarted all WO services on the machine, and now, suddenly,
all the AWT stuff works very nicely. I guess that will have to do -
until the next time it fails. *sigh*
I believe the bug you're mentioning below is a 5.2.4 issue, at least I
haven't seen it in 5.2.3.
Cheers,
Hugi
On 12.5.2005, at 14:23, Mike Schrag wrote:
I need to file a radar on this actually -- I just did a small test
case:
public class Test {
public static void main(String[] _args) {
java.awt.Dimension dim = new java.awt.Dimension(10,
10);
System.out.println("Done.");
}
}
When you run that "headful", it works fine (and you see a small app
launch in the gui). If you run it java.awt.headless=true it has
problems:
Schragtop:~ localmschrag$ java Test
Done.
Schragtop:~ localmschrag$ java -Djava.awt.headless=true Test
** ATTENTION ***
Please Attach Debugger : No Autorelease pool exists
Press any key to continue...
On May 12, 2005, at 9:35 AM, Hugi Thordarson wrote:
Thanks for the reply, Mike!
I suspected it might be something happening deep down in the JVM guts
- every tiny operation that involves AWT seems to cause the crash.
I'm still going to attempt to figure out what the heck the problem
is, since this code has been running problem free on this server for
some time and I need to resolve this fast, so I don't have time to
implement ImageMagick just now.
Sometimes, just sometimes, I really, really hate java.
Thanks!
- Hugi
On 12.5.2005, at 13:08, Mike Schrag wrote:
I ran into this same problem when using AWT-referencing-code (which
fires up Toolkit). It basically results in a total VM crash
because it fails inside of a native method. Annoying
java.awt.headless is supposed to exist to prevent this problem, but
for whatever reason it didn't behave as expected. Ultimately I
never got the performance out of ImageIO that I needed anyway, so I
switched to custom JNI using the ImageMagick API's (which,
incidentally, was between 10 and 100x faster performing some of the
same operations as ImageIO). It was REALLY frustrating -- I know
your pain. At the end of my case, I only depended on
java.awt.Dimension (which ALSO fires up Toolkit), which I just wrote
my own version of. It wasn't cheaper (time-wise) to go this route
then to try and diagnose exactly what was going on under the covers
in the VM. I'm fairly confident there IS a workaround, because I
would be shocked that it just doesn't work at all, but it wasn't
immediately obvious.
One option you have, unless you're TRULY headless, is to launch as
root instead of appserver, which can then always get to the
windowserver. But that, of course, brings with it other problems.
ms
On May 12, 2005, at 7:46 AM, Hugi Thordarson wrote:
Good morning everyone!
I have a problem with one of my apps. When this application tries
to perform image resizing, it dies suddenly and horribly, without
giving any error messages in the application log (with debugging
enabled). The application instance immediately terminates and then
restarts.
The app is running as the "appserver" user and the
-Djava.awt.headless property is set to true. Running WO 5.2.3 on OS
X server 10.3.8. Classes involved in the resize operation inclide
BufferedImage and ImageIO.
The image resizing code resides in a framework and is running fine
on other servers, so I really don't know what to think. I just
wanted to see if anyone had experienced something similar, since I
suspect this might become a long winded bug hunt.
Cheers,
Hugi
// Hugi Thordarson
// Atvinnulaus
// http://www.karlmenn.is/
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
email@hidden
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden