Bug 202092

I cannot reproduce it.
When I use the umlaute.pdf from the Debian bug
http://bugs.debian.org/cgi-bin/bugreport.cgi/umlaute.pdf?bug=381386;msg=10;att=1
it shows in Adobe Reader the line
"öäüßÖÄÜ"
and the same when I display it with Ghostscript
gs -r90 umlaute.pdf
and when I print it to a PCL5e printer using
gs -sDEVICE=ljet4 -r600 -sOutputFile=/dev/lp0 umlaute.pdf
the printout looks correctly.

I use ghostscript-library-8.15.2rc1 and
rpm -q --changelog ghostscript-library | head
shows the last canges which are
-------------------------------------------------------------------------
* Tue Jun 13 2006 - werner@suse.de
- Do not stop execution on unknown page device features (#161004)

* Thu Mar 09 2006 - werner@suse.de
- Use the pdf_sec.ps of the new GPL ghostscript 8.50 to be able
  to read newest encrypted pdf files.

* Wed Mar 01 2006 - werner@suse.de
- Add new krgb-1.2 patch for gdevijs to fix buffer overflow
-------------------------------------------------------------------------

Do you use the same ghostscript-library?

If yes, provide a PostScript test file as attachment to this bug
which doesn't print correctly for you and tell us exactly which
printer model and which printer driver you use. For the latter,
provide the PPD file for your printer from /etc/cups/ppd/ as
second attachment to this bug.

It is no critical problem because there is no real
"Crash, loss of data, corruption of data, severe memory leak", see
https://bugzilla.novell.com/page.cgi?id=fields.html#bug_severity

Created attachment 97328 [details]
Not printing correctly

If you print this you will notice the missing dots over the Ü in "Über Ihren Besuch..."

Created attachment 97329 [details]
my ppd file

This is the ppd for my HP-Deskjet 6540

I can reproduce it partially with the attachment in comment #2:

When I view it with "gv" I don't see the dots over the Ü in "Über ..."
but when I set the magnification in gv to 8 or 10, I see that the dots
are there but they look like only two tiny dashes.
I can see this even better when I set the mouse cursor over the "Ü"
and then use maximum magnification via the middle mouse button.
It looks as if there is a too small bounding box for the "Ü" so that most
of the dots is cut on the top.
It depends on the printer driver and printer hardware if the tiny dashes
are actually printed or not (my printer prints them).

A workaround for printing:
As the Adobe Reader shows the dots, use the Adobe Reader to print it.
You can even use the Adobe Reader to convert it to printable PostScript
on the command line (see "rpm -qi acroread").
acroread -toPostScript -size a4 <input.pdf | lp -d queuename

Matthias:
Which application did you use to create your PDF?
Can you also create a PostScript file with the same application?
If yes, does the PostScript file print correctly?
Does it print correctly when you print directly from the application?

As Adober Reader shows and prints it correctly, I assume that the font
in the PDF is not broken and that it is really a bug in Ghostscript.
Therefore I reassign it to our Ghostscript maintainer.
Werner, see also the debian bug in comment #0.

(In reply to comment #4)

> When I view it with "gv" I don't see the dots over the Ü in "Über ..."
> but when I set the magnification in gv to 8 or 10, I see that the dots
> are there but they look like only two tiny dashes.

Yes, sometimes they are printed as two tiny dashes.

> A workaround for printing:
> As the Adobe Reader shows the dots, use the Adobe Reader to print it.

Just tried to print it with adobe reader but the dots are missing again.

> You can even use the Adobe Reader to convert it to printable PostScript
> on the command line (see "rpm -qi acroread").
> acroread -toPostScript -size a4 <input.pdf | lp -d queuename

Is this something I must do to print the dots?

> Matthias:
> Which application did you use to create your PDF?

OpenOffice.org

> Can you also create a PostScript file with the same application?

No, sadly not.

> Does it print correctly when you print directly from the application?

No. I only created a PDF for you convenience as I didn't know if you have the Bitstream Vera fonts installed.
 
> As Adober Reader shows and prints it correctly, I assume that the font
> in the PDF is not broken and that it is really a bug in Ghostscript.
> Therefore I reassign it to our Ghostscript maintainer.
> Werner, see also the debian bug in comment #0.

Thank you in advance and hopefully this gets fixed as soon as possible as this bug can hit you everytime with every so important document.

Matthias:

I used your PDF from attachment #97328 [details] as "input.pdf" to print it
correctly with Adobe Reader (both via using the usual graphical
user interface of the Adobe Reader and via the command line).
I have no idea why your Adobe Reader doesn't print it correctly.

You can print directly from OOo.
You can print from OOo into a file and then the resulting file
should be a PostScript file.

Johannes? Does this also happen with the ghostscript espgs-8.15.2 of SLES10-SP1?

It seems this is the matching STR:
http://www.cups.org/str.php?L1911
Michael's comment on Aug 25 indicates that it is not yet fixed.

Nevertheless I am about to test ghostscript-library from SLES10 SP1.

Werner:
The same problem with ghostscript-library from SLES10 SP1
(i.e. only two tiny dashes insead of the dots).

To be safe that I really used the right ghostscript-library
rpm -q --changelog ghostscript-library | head
----------------------------------------------------------------------------
* Mon Aug 14 2006 - werner@suse.de
- Ghostscript update to version 8.15.2 (last change does not hurt
  due dynamic font path generation) (bug #190870)
- This version includes e.g. fix for unknown pages devive (#161004)

* Fri Aug 11 2006 - mfabian@suse.de
- move URW symlink to /usr/share/fonts (because of X11R7).

* Wed Jul 12 2006 - werner@suse.de
- Do never build in parallel, this will not work
----------------------------------------------------------------------------

Hmmm ... then I'd like to know why the debian people do not have
submitted their fix?  Is this fiy only a dirty hack breaking other
stuff or are their to lasy to submit?

The only problem now is that the debian version includes a epsgs post-8.15.2
which includes a patch for CJKV support breaking umlaute. Now one of the
patches of debian reverts this change that is that the version seems
to be identical with our version.

Which ghostscript version do you use?

ghostscript-x11-8.15.2rc1-20

OK, just tried the debian patches and it does not work.  Same behaviour
as without.  The debian patches uses the last update to espgs post-8.15.2
and the reverts the CJK patch because this one seems to broke umlaute ...
AFAIS your problem seems to be caused by an other source.

Now I've tried without reverting the CJK patch and now _all_ accents or dots
of the umlaute are gone.  All characters using overlay technique for showing
dots or accents do not have any dots nor accents.
IMHO your hint http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=381386
has nothing todo with your problem.

(In reply to comment #14)
> IMHO your hint http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=381386
> has nothing todo with your problem.

May be. But the problem needs a solution nevertheless. 

I'm only the maintainer of ghostscript and fixing a SIGSEGV
wihtin a driver is not the same as hunting a wrong fine tune
within the rendering engine of ghostscript.

So what will I have to do to get this fixed?

Please open a Software Trouble Report at http://www.cups.org/espgs/str.php

Created attachment 100875 [details]
PostScript sample: some accented letters: accents and M/W trimmed

I have the same problem with Czech accents. Also large letters are trimmed from right. I attached a sample. Simply run gs on the file. Some accents look like short horizontal lines, while they should look like ^ only upside down.

I observed that if I replace 'gs' from ghostscript-library-8.15.2rc2 with either GNU ghostscript 8.54 or espgs-8.15.3, the output/display is correct.

Before that I tried espgs-8.15.2 from factory, but that didn't help.

oh, I forgot to mention that "gs -dNOCACHE print.ps" also seems to produce correct output (with ghostscript-library-8.15.2rc2).

OK it works.

Andreas? I'll need a SWAMP ID for this bug and bug #209879
... only the problem of bug #203287 isn't solved by espgs 8.15.3.
With this the update for SLES10-SP1 would be dead.

We will release an update with fixes for  bug #202092,  bug #209879 and bug #203287.

bug #203287 is still worked on, but I expect a solution is not to far away.

Werner, ask Anja for a SWAMPID in 203287, when you have a fix.

Submitted

released