Arch Linux

Please provide your pdf creation workflow. What packages do you use. Please pacman -Syu && pacman -Q cups cups-pdf ghostscript cups-pdf poppler.

Please attach all cups.logs with debug mode.

If it's caused by a bug in pdf2ps - the filter is part of ghostscript pkg. You may want to get in contact with Till Kampeter being the printer guru and responsible for most upstream developement.

I'm also seeing this problem. The error and cups-pdf logs are attached.

I have some really old pdfs I have printed with cups-pdf that do not show this problem, the differences I see are:
- fonts are embedded in the good pdf
- the producer and creator fields of the pdfs are different. The good pdf has "Producer: GPL Ghostscript 8.71" and "Creator: cairo 1.8.10 (http://cairographics.org)", the bad pdf has "Producer: GPL Ghostscript 9.06" and "Creator: GPL Ghostscript 906 (ps2write)".

How to reproduce (to get a bad pdf):
Open any page (any page of the cups web interface will do) with a browser (chomium will do), print it to the virtual printer associated with cups-pdf.

A good pdf would be similar to what can be obtained by using Print -> Print to file with any program that supports printing to pdf directly.

Output of pacman -Q cups cups-pdf ghostscript cups-pdf poppler:
cups 1.6.1-6
cups-pdf 2.6.1-1
ghostscript 9.06-1
cups-pdf 2.6.1-1
poppler 0.20.5-1

It won't get fixed magically until someone who's affected will report it somewhere upstream.

I'll try to contact the cups-pdf(1) dev sometime next week. I'll drop a comment here after I get some feedback.

(1) cups-pdf seems to be the one to blame since printing to pdf with cups directly works as expected (jut not as convenient and secure as with cups-pdf).

Any word on this Mauros? Or somewhere where I can follow it up upstream? I'm having the same problem and the fix didn't work for me. Though I do agree, cups-pdf is most certainly the problem

Upstream is looking into the problem. As far as I know there is no bugtracker (I couldn't find any bugtraker) and I have contacted upstream by email.

In the emails we have exchanged, it was determined that cups-pdf works as intended if called manually/directly and fed a postscript file. I the last email I received I was told this:

"I just wanted to let you know I am still on it but could not find the culprit yet. I think I will set up a number of virtual machines with different distributions so I can figure out when the error happens and when it does not."

Until the problem is found and fixed, I have found 2 workarounds, none of which is ideal though.

1) Change the default postscript printing renderer, this will make cups-pdf work a little better, at least type 1 fonts will get embedded but text might not be searchable. Do the change with:
lpadmin -p <CUPS-PDF printer name> -o pdftops-renderer-default=pdftops

2) Use a different way to print to pdf, this one is cumbersome to use and might not be safe in a multiuser environment.
To use this, edit cups configuration file and add a line with 'FileDevice Yes', then restart cups.
Add a new printer, select any option from the "Other Network Printers". In the "Connection" field input "file:/var/spool/pdfoutput.pdf".
In the "Make" list select "Generic" and in the "Model" list select "Generic PDF Printer (en)".

You can now print to the new printer and get a pdf with type 1 fonts embedded. However the pdf will get overwritten every time you print a new pdf, also you will have to change the permissions on the file so everyone can read it, otherwise only root can read the file.

Alright. Great, thanks for the information. I'll probably figure some work around, I have a few ways I can circumvent it without toying with things.

Let me know if I can be of any use by also having the same bug, perhaps I can provide info to upstream if they need it.

Either way, thanks much for chasing things down.

http://bugs.ghostscript.com/ - Ghostscript tracker
https://bugs.linuxfoundation.org/ - for cups-filters choose product "OpenPrinting", component "cups-filters"
for cups-pdf ping the upstream dev.

Status after latest updates? Any upstream report we can follow?