Hi Dave,
The following bug filed against Java3D1.2 will be fixed in Java3D1.2.1 Beta:
4370843 memory leak due to incorrect transform targets update
The root cause is that old Shape3DRetained references are not cleanup
properly when a Branch containing a TransformGroup is detached from the scene
graph.
If you send a simple test program to "[EMAIL PROTECTED]",
we can verify this fix for you.
Thanks for reporting this problem.
Lawrence Chow
Java3D Team
Sun Microsystems
>MIME-Version: 1.0
>Content-Transfer-Encoding: 7bit
>X-Priority: 3
>X-MSMail-Priority: Normal
>X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2615.200
>Date: Wed, 6 Sep 2000 07:15:46 -0400
>From: David <[EMAIL PROTECTED]>
>Subject: [JAVA3D] Memory Leak?
>To: [EMAIL PROTECTED]
>
>It looks like my shapes are not being released to garbage collection after a
>detach:
>
>TRACE 6671:
> javax.media.j3d.Shape3DRetained.setLive(Shape3DRetained.java:1102)
> javax.media.j3d.GroupRetained.doSetLive(GroupRetained.java:1475)
> javax.media.j3d.BranchGroupRetained.setLive(BranchGroupRetained.java:101)
> javax.media.j3d.GroupRetained.checkSetLive(GroupRetained.java:1009)
> javax.media.j3d.GroupRetained.checkSetLive(GroupRetained.java:869)
> javax.media.j3d.GroupRetained.doAddChild(GroupRetained.java:431)
>
>
>SITES BEGIN (ordered by live bytes) Wed Sep 06 07:02:48 2000
> percent live alloc'ed stack class
> rank self accum bytes objs bytes objs trace name
> 1 7.01% 7.01% 591892 157 591892 157 1 [I
> 2 3.33% 10.34% 281304 3907 281304 3907 6671
>javax.media.j3d.Shape3DRetained
>
>
>Notice that there are as may live objects as were ever alocated? This means
>that never, in the course of my run, did one Shape3DRetained get released.
>I am dropping my reference to them after detaching, so I would expect them
>to go away. I know there is a bug with Shapes not being released with
>shared Appearance (4182781 Shape3D memory not freed when unassigned if
>associated appearance object isalive ). But I assumed that was a shared
>Appearance() object, not a sub object. All my shapes share material and
>texture objects, which is necessary and correct that they do so. Don't I
>just detach branchgroups and let them die? or do I need to make them
>"unalive" somehow?
>
>By the way... when is the next version of Java3d going to be made available?
>
>Dave Yazel
>
>===========================================================================
>To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
>of the message "signoff JAVA3D-INTEREST". For general help, send email to
>[EMAIL PROTECTED] and include in the body of the message "help".
-lawrence
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff JAVA3D-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
