Linux Acroread memory link with Context

Linux Acroread memory link with Context

[Matthew Baker]

Hi,

Last week I posted a message about Linux Acroread's memory usage growing
as I browse through an image-intensive document I'm working on.  I've
played with this a bit more and have found the following:

1. Even if I have a document with a small number of images, flipping back 
   and forth between pages containing big images eventually takes up more 
   and more swap space.  So it seems images are not being deallocated
   after use, or something along those lines.

2. Converting the document to PostScript with Acrobat (Linux version) then
   converting the PostScript file back to PDF with Distiller (3.0, Win)
   elliminates this problem, ie the resulting file doesn't result in the
   memory leak when viewed under Linux. 

3. Acroread 3.0 and 4.0 beta both exhibit this problem.

4. I don't have any problems under Windows.  I haven't tried other Unix
   platforms yet.

So, to me it seems that it is either a Context or pdftex bug since the
same images converted to PDF with Distiller don't cause problems.  Or if
not a bug, something which triggers a bug in Linux Acroread that Distiller
PDF files don't trigger.

For a test document, all I'm using is several pages of \startcombination
... \stopcombination with 9 images in each. Images are 256x384 pixels
each, so they are not in any way unreasonably big.  I monitor memory usage
by watching swap space in xosview.

Any ideas why this might happen?  Next step for me is to try something
with plain pdftex and includegraphics commands to see if that causes the
same problem.

Cheers,
Matthew

--
Dr. Matthew Baker           matthew.baker@gmd.de
GMD - FIT.MMK               http://fit.gmd.de/hci/pages/matthew.baker.html

Re: Linux Acroread memory link with Context

[Hans Hagen]

Matthew Baker wrote:

> So, to me it seems that it is either a Context or pdftex bug since the
> same images converted to PDF with Distiller don't cause problems.  Or if
> not a bug, something which triggers a bug in Linux Acroread that Distiller
> PDF files don't trigger.
> 
> For a test document, all I'm using is several pages of \startcombination
> ... \stopcombination with 9 images in each. Images are 256x384 pixels
> each, so they are not in any way unreasonably big.  I monitor memory usage
> by watching swap space in xosview.
> 
> Any ideas why this might happen?  Next step for me is to try something
> with plain pdftex and includegraphics commands to see if that causes the
> same problem.

This really sounds strange, because we have documents with hundreds of
illustrations and large bitmaps.

Technically context/pdftex uses an xobject, which is the appropriate
structure. You can set \pdfcompresslevel=0 and take a look at the file.
You can try to follow the object pointers. 

I'll ask Thanh too, 

Hans

-----------------------------------------------------------------
                                          Hans Hagen | PRAGMA ADE
              Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
      tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.nl
-----------------------------------------------------------------