Re: Why is Firefox printing in Linux so slow?

From: Enrique Perez-Terron (enrio_at_online.no)
Date: 10/13/05 

Date: Thu, 13 Oct 2005 23:32:10 +0200

On Thu, 13 Oct 2005 16:50:45 +0200, <usenet.1.massysett@spamgourmet.com> wrote:

> I am wondering why printing in Firefox in Linux on a PostScript printer
> takes so long.
>
> Here is my setup: SuSE 10.0 using CUPS; Brother HL-5150D laser printer
> (which is a native PostScript printer) hooked up via USB; recent
> Firefox (either 1.0.6 or 1.0.7); PPD file from the Brother CD that
> comes with the printer.
>
> When I used this printer from Windows XP to print pages from Firefox,
> pages would start emerging from the printer within seconds after I
> clicked "print." This was true with both the Brother PCL Windows driver
> and the Brother PostScript Windows driver. After printing started, the
> pages would continue to emerge at top speed until all the pages were
> printed.
>
> Then I tried the new SuSE. Firefox printing is extremely slow. Even
[...]

You show quite convincingly that there is something strange with the
postscript that FF generates or with how the printer executes that postscript.

I suppose you would have to take the question to a forum specialised in
postscript matters, and perhaps even provide some of the contents
of the postscript file.

I am afraid that you will even have to do a lot of work reducing the
postscript file (yes, print-to-file output) to a minimum. That is not
easy, because postscript is a programming language and each postscript
files start with a fairly large prolog of "function definitions" which
essentially defines a new language that the rest of the program is
written in. That makes the job of removing parts of the file much more
difficult.

Just in case, let's consider a few other possibilities. Could it
be that FF somehow manages to make CUPS believe that the file is
not Postscript, but needs some awkward filtering before it is printed?
What does the first line of the file look like?
I would check in /var/log/cups/error-log, search for lines containing
"Started ". You should see the filters that are run, and the backend.
Comparing this when using OpenOffice should tell if there is
anything here.

Hm, I just fired up FF, and printed out a page from Washington Post.
It took virtually no time. On my system, I have a cheap Lexmark printer
that depends on Ghostscript for the interpretation of the postscript.
Don't expect debugging this will be easy, as you will probably be
debugging the postscript engine inside your printer. You might
be better off if you can figure out how to do as Peter suggests.
Define the printer's data type to be something else but Postscript.
You could define a separate print queue that uses that definition
and use that queue with FF while keeping the regular queue for all
other purposes. The special queue needs a different PPD file.

That ppd file needs a few lines like this:

   *cupsVersion: 1.1
   *cupsModelNumber: "0"
   *cupsManualCopies: False
   *cupsFilter: "application/vnd.cups-raster 100 rastertolxx74"

in addition to the more ordinary printer description.
The Interesting line here is the cupsFilter line, I don't
know how important the others are.

This line assigns a mime-like data-type to the data that the
last filter needs, and then specifies the name of an executable
filter in /usr/lib/cups/filter, that will do the last format
converson. The number 100 is a relative cost measure of the
conversion, used by CUPS to pick the cheapest path from any
given data type to one that the printer supports.

If your printer understands some of the langauges supported
by ghostscript (run gs -help and look at "available
devices"), you can write a script that uses ghostscript to
convert postscript to that language. Then the filter's
input data type will be application/postscript.

By the way, CUPS also provides a version of the /usr/bin/lpr program,
so using lpr still uses cups.

-Enrique

Relevant Pages

  • Re: Application does not print while queues seem to work fine
    ... though I did *not* bring down the queue I printed to. ... In the app there is no way to ... always sending PostScript to the printer queue. ... SMIT, but this did not work, while printing an ASCII file the very same ...
    (comp.unix.aix)
  • Re: Printing thru terminal servers to a Mac X computer
    ... And Ghostscript ... Postscript printer but since the attached printer is a Deskjet, ... never getting shared by RDC in the first place. ... ahead of time with a print queue specifically for this user's IP address ...
    (microsoft.public.win2000.macintosh)
  • Re: Postscript queue only outputs single line
    ... Why have a PostScript queue if its not PostScript files that you're printing ... You can go into smit chpq and try setting the print type to ... PostScript only queue? ...
    (comp.unix.aix)
  • Re: mac os x - connecting to a print queue that has parentheses in the name
    ... Mario (Postscript) ... When I set up the printer as LPD/LPR on Mac OS X 10.x, I put in Mario ... This results in the print jobs not going to the correct queue. ...
    (comp.sys.mac.printing)
  • Re: Samsung CLP500
    ... > was that the queue expects a stream of postscript, ... > Either try using a Postscript driver under Windows, or try and set up the ... This RAW queue: will that allow me to use the ... Samsung driver under win2k and then send the windoz print instructions ...
    (alt.os.linux.suse)