Printing a transparent gif header result in a completely black header.

Is this symptom still reproducible in 8.10 or 9.04?

Still confirmed under intrepid. However, while it didn't print in the past, it now results in ugly completely black header, which is still a bug. I updated bug title and description.

Can you attach the error_log after doing "cupsctl LogLevel=debug" and then printing such a job? Can you print into a PostScript file from GNU Cash to see whether the image comes out black already from the application?

Till Kamppeter : When printing a postscript file, the bug does not happen. It only happens when printing with my HP printer so far. I removed error_log, did "sudo cupsctl LogLevel=debug" and printed with my printer. I attach the clean error_log file from /var/log/cups .

The error_log shopws that GNU Cash is sending PDF to the printing system and this gets processed at first by pdftopdf and then by foomatic-rip. So there are theree possibilities for what makes the GIF images black (or at least the transparent parts of them):

- GNU Cash already
- The pdftopdf filter
- Ghostscript and the HPIJS printer driver (they get called by foomatic-rip).

The first step is to find out what GNU Cash sends to the printing system.

To find out this, disable your print queue with

cupsdisable <queue name>

This prevents jobs which you send from being processed and removed from the queue,

Send a job from GNU Cash with your GIF image to the print queue. It will stay waiting in the

/var/spool/cups/

directory, a file whose name begins with a "d". Copy the file into your home directory (note that normal users cannot read these files):

sudo cp /var/spool/cups/d* .
sudo chown <your user name> d*
mv d* gnucash.pdf

Clean up your print queue with

cancel -a
cupsenable <queue name>

Now have a look at the file with evince or Adobe Reader. Is the image already black? If so, GNU Cash (or the GUI libraries which it is using) has a bug.

If not, the problem is in the filters. Do

cupsfilter -m application/vnd.cups-pdf -p /etc/cups/ppd/<your print queue>.ppd gnucash.pdf > pdftopdfout.pdf 2>pdftopdf.log

This runs only the pdftopdf filter. Display pdftopdfout.pdf with evince or Adobe Reader. How does the image look like. If it is black, pdftopdf has a bug, otherwise the bug is in Ghostscript or the driver. Try to display it also with Ghostscript, ("gs pdftopdfout.pdf"). If the image is black only then, the bug is in Ghostscript.

Set up a print queue for your printer with Gutenprint as the driver (Use the HP DeskJet 990C as the model if Gutenprint is not listed as a driver for your printer). Does the image come out correctly then. If so, the HPIJS driver or the IJS interface of Ghostscript is faulty. To check whether it is not Ghostscript's IJS interface you can install the packages foomatic-db-gutenprint and ijsgutenprint and set up a queue with a "Foomatic/gutenprint-ijs" driver. If the image comes out correctly, the bug is in HPIJS, otherwise in Ghostscript's IJS interface.

Here's the result!

/var/spool/cups/d* OK
pdftopdfout.pdf OK
gs pdftopdfout.pdf BUG! (The header is only black at that point).

I changed the driver of my printer to HP Deskjet 990C Gutenprint and reproduced the same steps from the beginning, and the result is exactly the same. I didn't try "Foomatic/gutenprint-ijs" driver because I didn't find it in the list (I already installed the packages, I'm probably missing the category where these printers are classified).

Can you please attach /var/spool/cups/d* and pdftopdfout.pdf?

Please do an actual printout with the HP Deskjet 990C/Gutenprint queue. Does the GIF come out black?

Did you install both the foomatic-db-gutenprint and ijsgutenprint packages? In system-config-printer you choose "HP", "DeskJet 990C", and then "HP DeskJet 990C Foomatic/gutenprint-ijs.5.2". Do also an actual print out with this queue and tell whether the GIF comes out correctly.

Ok I already did that (changing driver of my printer to "HP DeskJet 990C Foomatic/gutenprint-ijs.5.2" and follow the steps you wrote) and the result is :

/var/spool/cups/d* OK
pdftopdfout.pdf OK
gs pdftopdfout.pdf BUG! (The header is only black at that point).

I'm attaching the requested files, Thanks for your instructions.

Thank you for the file. evince (=Poppler) displays it correctly and gv (=Ghostscript which ships with Intrepid) not. I have also taken a Subversion snapshot of the newest development state of Ghostscript. Also here the GIF comes out black (I used the "png16m" output device). So current Ghostscript still has the bug and the bug is independent of the output device/driver. It is also not dependent on Ubuntu's patches. Iy is an upstream bug.

It also does not depend on whether pdftopdf has processed the file. I get the same results with both your d00217-001 and pdftopdfout.pdf.

Bug reported upstream:

http://bugs.ghostscript.com/show_bug.cgi?id=690189

In case that it might be useful information, I was also able to reproduce the bug by printing Firefox png icon in /usr/share/pixmaps directly from eog.

I have a series of similar entries in /var/log/messages. These were generated when printing from seamonkey.

# grep cups /var/log/messages
May 29 08:40:27 asus kernel: [254244.927576] type=1503 audit(1243611627.586:40): operation="inode_permission" requested_mask="::r" denied_mask="::r" fsuid=7 name="/var/run/samba/gencache.tdb" pid=19270 profile="/usr/sbin/cupsd"
May 29 08:40:27 asus kernel: [254244.927615] type=1503 audit(1243611627.586:41): operation="inode_permission" requested_mask="::r" denied_mask="::r" fsuid=7 name="/var/run/samba/gencache.tdb" pid=19270 profile="/usr/sbin/cupsd"
May 29 08:40:27 asus kernel: [254245.158900] type=1503 audit(1243611627.818:42): operation="inode_permission" requested_mask="::r" denied_mask="::r" fsuid=7 name="/var/run/samba/unexpected.tdb" pid=19270 profile="/usr/sbin/cupsd"
May 29 08:40:27 asus kernel: [254245.249089] type=1503 audit(1243611627.911:43): operation="inode_permission" requested_mask="::r" denied_mask="::r" fsuid=7 name="/var/run/samba/unexpected.tdb" pid=19270 profile="/usr/sbin/cupsd"
May 29 08:40:28 asus kernel: [254245.339282] type=1503 audit(1243611627.998:44): operation="inode_permission" requested_mask="::r" denied_mask="::r" fsuid=7 name="/var/run/samba/unexpected.tdb" pid=19270 profile="/usr/sbin/cupsd"
May 29 08:40:28 asus kernel: [254245.339330] type=1503 audit(1243611627.998:45): operation="inode_permission" requested_mask="::r" denied_mask="::r" fsuid=7 name="/var/run/samba/gencache.tdb" pid=19270 profile="/usr/sbin/cupsd"
May 29 12:33:45 asus kernel: [268242.875303] type=1503 audit(1243625625.535:46): operation="inode_permission" requested_mask="::r" denied_mask="::r" fsuid=7 name="/var/run/samba/gencache.tdb" pid=24312 profile="/usr/sbin/cupsd"
May 29 12:33:45 asus kernel: [268242.875343] type=1503 audit(1243625625.535:47): operation="inode_permission" requested_mask="::r" denied_mask="::r" fsuid=7 name="/var/run/samba/gencache.tdb" pid=24312 profile="/usr/sbin/cupsd"
May 29 12:33:45 asus kernel: [268242.997639] type=1503 audit(1243625625.658:48): operation="inode_permission" requested_mask="::r" denied_mask="::r" fsuid=7 name="/var/run/samba/unexpected.tdb" pid=24312 profile="/usr/sbin/cupsd"
May 29 12:33:45 asus kernel: [268243.087797] type=1503 audit(1243625625.746:49): operation="inode_permission" requested_mask="::r" denied_mask="::r" fsuid=7 name="/var/run/samba/unexpected.tdb" pid=24312 profile="/usr/sbin/cupsd"
May 29 12:33:45 asus kernel: [268243.177967] type=1503 audit(1243625625.838:50): operation="inode_permission" requested_mask="::r" denied_mask="::r" fsuid=7 name="/var/run/samba/unexpected.tdb" pid=24312 profile="/usr/sbin/cupsd"
May 29 12:33:45 asus kernel: [268243.178025] type=1503 audit(1243625625.838:51): operation="inode_permission" requested_mask="::r" denied_mask="::r" fsuid=7 name="/var/run/samba/gencache.tdb" pid=24312 profile="/usr/sbin/cupsd"

Bug 362641 is probably the same problem.

This bug is fixed in the upcoming Ghostscript 8.70 which will be released next week, so the bug will be fixed in Karmic.

This bug was fixed in the package ghostscript - 8.70.dfsg.1~rc1-0ubuntu1

---------------
ghostscript (8.70.dfsg.1~rc1-0ubuntu1) karmic; urgency=low

  * New upstream release
     o Fixes many transparency problems: color space conversion, mask contexts,
       patterns, ...
     o Fixes in font handling, especially when generating PDF
     o Improvements in robustness, correctness, and performance
     o New generic Esc/Page drivers: "eplmono", "eplcolor"
     o New "cdnj500" driver for many HP DesignJet printers
     o License is now GPLv2 or later and not GPLv2-only any more
     o Merged all patches of the Ubuntu and Debian packages upstream
     o Fixes LP: #196009
  * debian/patches/33_bad-params-to-xinitimage-on-large-bitmaps.dpatch,
    debian/patches/35_bitcmyk-blank-output.dpatch,
    debian/patches/37_fix-segfault-in-cups-raster-output-device.dpatch,
    debian/patches/38_CVE-2009-0583_0584.dpatch,
    debian/patches/40_pdfwrite-numcopies.dpatch,
    debian/patches/41_CVE-2009-0196.dpatch,
    debian/patches/42_CVE-2009-0792.dpatch,
    debian/patches/43_add-cdnj500-driver.dpatch,
    debian/patches/45_cups-device-pagesize-margins-duplex-fixes.dpatch,
    debian/patches/47_ps2write-segfault-fix.dpatch,
    debian/patches/50_ps2write-do-not-advertize-dsc-conformance.dpatch,
    debian/patches/53_fix-pstoraster-for-call-with-input-filename.dpatch:
    Removed patches backported from upstream.
  * debian/copyright: License change.
  * debian/ghostscript.links: s/8.64/8.70/

 -- Till Kamppeter <email address hidden> Wed, 29 Jul 2009 09:47:54 +0200