[OpenJDK 2D-Dev] JPEGImageWriter memory leak, regression due to fix for 800791

[Andrew Brygin]

Hello Mario,

  I am sorry for the delay.

  This problem is now tracked  as a bug 8020983:
  http://bugs.sun.com/view_bug.do?bug_id=8020983

  This problem can be avoided by an explicit call for reset() or
  dispose() on an instance of jpeg writer.

Thanks,
Andrew

On 7/20/2013 9:22 PM, Mario Torre wrote:
>
> Hi Phil, everybody,
>
> This request is one month old now, can somebody please comment on 
> that? The silence is starting to be a bit embarrassing to be honest.
>
> Cheers,
> Mario
>
> Il giorno 11/lug/2013 13:03, "Mario Torre" <neugens at redhat.com 
> <mailto:neugens at redhat.com>> ha scritto:
>
>     On Fri, 2013-07-05 at 15:17 +0200, Mario Torre wrote:
>     > 2013/7/2 Adam Domurad <adomurad at redhat.com
>     <mailto:adomurad at redhat.com>>:
>     > > Hi all, please see the attached Test8007918Regression.java.
>     > >
>     > > There seems to be a regression caused by this patch:
>     > > http://hg.openjdk.java.net/jdk7u/jdk7u-dev/jdk/rev/90c9f1577a0b
>     > >
>     > > As the small testcase demonstrates, creating a JPEGImageWriter
>     and calling
>     > > setOutput causes it to live permanently. This is due to JNI global
>     > > (non-weak) reference in sun/awt/image/jpeg/imageioJPEG.c.
>     > >
>     > > Is it acceptable to simply make this reference weak ? This
>     should bring
>     > > object lifetimes back to how they were before this patch.
>     Attached is a
>     > > patch that, from my tests, fixes the issue.
>     > >
>     > > Thanks & happy hacking,
>     > > -Adam
>     >
>     > Hi all,
>     >
>     > Any news on that one?
>
>     Hi all,
>
>     Can somebody please comment on this bug?
>
>     Thanks,
>     Mario
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/2d-dev/attachments/20130722/80c5ed8f/attachment.html>