Re: Error loading GIF image using JAI in FOP Trunk (was: Error w hile processing a PDF - OK)

2005-08-30 Thread Manuel Mall
Stephen,

thanks for pointing this out (again). I'll post a patch to fix this.

Manuel

On Wed, 31 Aug 2005 07:02 am, Stephen Denne wrote:
> Regarding a fop-users email mentioning:
> > com/sun/media/jai/codec/FileCacheSeekableStream
>
> Reminded me of something I encountered recently:
>
> In my brief experience (of a modified version of FOP 0.20.5 that I
> inherited), if the href is to a file, then its input stream is being
> given to JAI, JAI is reading from the input stream, and writing it to
> another file in the temp directory, which isn't deleted, and uses up
> a file handle. (long running server processes don't exit, so
> deleteOnExit isn't very useful.)
>
> In my environment, all hrefs are to files, so I chose this
> alternative in org.apache.fop.image.JAIImage loadImage():
>
>   RenderedOp imageOp = JAI.create("fileload", this.m_href.getFile());
>
> and I also added
>
> imageOp.dispose();
>
> immediately after setting imageData.
>
> I see that trunk has rearranged things a bit, but it still appears to
> have the same leak.
>
> Stephen Denne


RE: Error loading GIF image using JAI in FOP Trunk (was: Error w hile processing a PDF - OK)

2005-08-30 Thread Stephen Denne
Regarding a fop-users email mentioning:

> com/sun/media/jai/codec/FileCacheSeekableStream

Reminded me of something I encountered recently:

In my brief experience (of a modified version of FOP 0.20.5 that I
inherited), if the href is to a file, then its input stream is being given
to JAI, JAI is reading from the input stream, and writing it to another file
in the temp directory, which isn't deleted, and uses up a file handle. (long
running server processes don't exit, so deleteOnExit isn't very useful.)

In my environment, all hrefs are to files, so I chose this alternative in
org.apache.fop.image.JAIImage loadImage():

RenderedOp imageOp = JAI.create("fileload", this.m_href.getFile());

and I also added 

imageOp.dispose(); 

immediately after setting imageData.

I see that trunk has rearranged things a bit, but it still appears to have
the same leak.

Stephen Denne