Heavy printing regression in Jaunty: SRU?

Till Kamppeter till.kamppeter at gmail.com
Thu Jun 4 13:29:27 BST 2009

I had some discussion on IRC whether or not to issue an SRU for Jaunty 
on a big problem with printing.



During the Jaunty release cycle I have switched the pdftops CUPS filter 
(filter which turns PDF print job data into PostScript) from using 
Poppler to using Ghostscript becuase the Poppler solution did not 
support documents which contain pages of different sizes. As the Poppler 
developers were not willing to fix this problem, I decided on switching 
to Ghostcript and Jaunty got released that way. See

   A3 pdf file is cropped and printed on A4 paper
   Poppler's pdftops does not support multiple-page-size output, use
   Ghostscript for the CUPS filter instead

Unfortunately, this caused many regressions, especially for users of 
PostScript printers, which makes the support of PostScript printers by 
Jaunty very bad, whereas this class of printers is considered the best 
supported by Linux. Also the cups-pdf print-into-PDF-file facility and 
many third-party (manufacturer) drivers got broken by that.

The problems are jobs not getting printed due to huge temporary files 
filling up the hard disk, extremely slow printing, corrupted font 
characters, no searchable text, incompatibility with PostScript 
interpreters in printers, ...

Regressions are reported as these bugs (there are more than these bugs 
and some of these bugs have also duplicates):

   Cannot print documents to Laserjet 4350, via network

   Spurious lines on print outs

   intrepid: printing very slow

   [jaunty] cups-pdf no longer embeds fonts in pdf file

   No output when printing in Ubuntu 9.04

   printer wont print although ubuntu says that job is compleated, also
   driver wont update

In all these bugs I have asked the users to replace their pdftops CUPS 
filters by a test script which uses Poppler and not Ghostscript. All 
users who answered told that the test script solved their problems.

Simply returning to Poppler would reintroduce bug 310575/bug 329991 
(documents with pages of different size), but in the mean time I have 
found a simple fix for Poppler to solve this bug, see


The debdiff for our Poppler package is attached to


and the fixed Poppler package will soon get uploaded into Karmic. I will 
let the reporter of bug 310575 test the fix.

So we can fix all the regressions with simple fixes in two packages, the 
mentioned Poppler patch and a patch to CUPS to use Poppler again and to 
call Poppler's pdftops the right way. This means, with the new 
"-origpagesizes" when the CUPS option "fitplot" is not given, never 
using "-noembtt" as this causes missing text characters in some cases 
(see bug 362186 and bug 381788) and not emitting PostScript Level 3 (see
bug 277404). As the fix in Poppler prevents the reappearing of bug 
310575/bug 329991 there is no known regression when returning to Poppler.

So I am thinking about applying the two small changes to Poppler and 
CUPS as an SRU for Jaunty. WDYT?


More information about the ubuntu-devel mailing list