1.1. History

This is the fourth generation of the Printing HOWTO. The history of the PHT may be chronicled thusly:

1. I wrote the printing-howto in 1992 in response to too many printing questions in comp.os.linux, and posted it. This predated the HOWTO project by a few months and was the first FAQlet called a `howto'. This edition was in plain ascii.

2. After joining the HOWTO project, the Printing-HOWTO was merged with an Lpd FAQ by Brian McCauley <B.A.McCauley@bham.ac.uk>; we continued to co-author the PHT for two years or so. At some point we incorporated the work of Karl Auer <Karl.Auer@anu.edu.au>. This generation of the PHT was in TeXinfo, and available in PS, HTML, Ascii, and Info.

3. After letting the PHT rot and decay for over a year, and an unsuccessful attempt at getting someone else to maintain it, this rewrite happened. This generation of the PHT is written in SGML using the LinuxDoc DTD and the SGML-Tools-1 package. Beginning with version 3.27, it incorporates a summary of a companion printer support database; before 3.27 there was never a printer compatibility list in this HOWTO (!).

4. In mid-January, 2000, I found out about the PDQ print "spooler". PDQ provides a printing mechanism so much better than lpd ever did that I spent several hours playing with it, rewrote parts of this HOWTO, and bumped the version number of the document to 4.

5. In mid-2000, I moved my printing website to www.linuxprinting.org, and began offering more powerful configuration tools there. I also converted the HOWTO to DocBook, and initiated coverage of CUPS, LPRng, and GPR/libppd.

1.2. Copyright

This document is Copyright (c) 1992-2000 by Grant Taylor. Feel free to copy and redistribute this document according to the terms of the GNU General Public License, revision 2 or later.

3.1. With PDQ

Most systems today ship with lpd, so this section won't apply. That said, I now recommend that people install and use PDQ in most cases instead of (or in addition to) lpd. PDQ just has much better support for printer options and such.

With PDQ, instead of the lpr command, you use the command pdq or xpdq. Both work much like the traditional lpr in that they will print the files you specify, or stdin if no files are given.

3.2. With LPD and the lpr command

If you've already got lpd setup to print to your printer, or your system administrator already did so, or your vendor did so for you, then all you need to do is learn how to use the lpr command. The Printing Usage HOWTO covers this, and a few other queue manipulation commands you should probably know. Or just read the lpr(1) man page.

In a nutshell, you specify the queue name with -P, and specify a filename to print a file, or nothing to print from stdin. Driver options are traditionally not controllable from lpr, but various systems accept certain options with -o, -Z, or -J.

If, however, you have a new system or new printer, then you'll have to set up printing services one way or another before you can print. Read on!

3.3. GUI Printing Tools

Most spooling systems alone offer only a rather basic command-line interface. Rather than use lpr directly, you may wish to obtain and use a front-end interface. These generally let you fiddle with various printing options (the printer, paper types, collation, n-up, etc) in an easy-to-use graphical way. Some may have other features, as well.

4.1. The lp device (kernels <=2.1.32)

The Linux kernel (<=2.1.32), assuming you have compiled in or loaded the lp device (the output of cat /proc/devices should include the device lp if it is loaded), provides one or more of /dev/lp0, /dev/lp1, and /dev/lp2. These are NOT assigned dynamically, rather, each corresponds to a specific hardware I/O address. This means that your first printer may be lp0 or lp1 depending on your hardware. Just try both.

A few users have reported that their bidirectional lp ports aren't detected if they use an older unidirectional printer cable. Check that you've got a decent cable.

One cannot run the plip and lp drivers at the same time on any given port (under 2.0, anyway). You can, however, have one or the other driver loaded at any given time either manually, or by kerneld with version 2.x (and later 1.3.x) kernels. By carefully setting the interrupts and such, you can supposedly run plip on one port and lp on the other. One person did so by editing the drivers; I eagerly await a success report of someone doing so with only a clever command line.

There is a little utility called tunelp floating about with which you, as root, can tune the Linux 2.0 lp device's interrupt usage, polling rate, and other options.

When the lp driver is built into the kernel, the kernel will accept an lp= option to set interrupts and io addresses:

When the lp driver is built in to the kernel, you may use the LILO/LOADLIN command line to set the port addresses and interrupts that the driver will use. Syntax: lp=port0[,irq0[,port1[,irq1[,port2[,irq2]]]]] For example: lp=0x378,0 or lp=0x278,5,0x378,7 ** Note that if this feature is used, you must specify *all* the ports you want considered, there are no defaults. You can disable a built-in driver with lp=0.

When loaded as a module, it is possible to specify io addresses and interrupt lines on the insmod command line (or in /etc/conf.modules so as to affect kerneld) using the usual module argument syntax. The parameters are io=port0,port1,port2 and irq=irq0,irq1,irq2. Read ye the man page for insmod for more information on this.

**For those of you who (like me) can never find the standard port numbers when you need them, they are as in the second example above. The other port (lp0) is at 0x3bc. I've no idea what interrupt it usually uses.

The source code for the Linux 2.0 parallel port driver is in /usr/src/linux/drivers/char/lp.c.

4.2. The parport device (kernels >= 2.1.33)

Beginning with kernel 2.1.33 (and available as a patch for kernel 2.0.30), the lp device is merely a client of the new parport device. The addition of the parport device corrects a number of the problems that plague the old lp device driver - it can share the port with other drivers, it dynamically assigns available parallel ports to device numbers rather than enforcing a fixed correspondence between I/O addresses and port numbers, and so forth.

The advent of the parport device has enabled a whole flock of new parallel-port drivers for things like Zip drives, Backpack CD-ROMs and disks, and so forth. Some of these are also available in versions for 2.0 kernels; look around on the web.

The main difference that you will notice, so far as printing goes, is that parport-based kernels dynamically assign lp devices to parallel ports. So what was lp1 under Linux 2.0 may well be lp0 under Linux 2.2. Be sure to check this if you upgrade from an lp-driver kernel to a parport-driver kernel.

The most popular problems with this device seems to stem from misconfiguration:

You can also read the parport documentation in your kernel sources, or look at the parport web site.

4.3. Serial devices

Serial devices are usually called something like /dev/ttyS1 under Linux. The utility stty will allow you to interactively view or set the settings for a serial port; setserial will allow you to control a few extended attributes and configure IRQs and I/O addresses for non-standard ports. Further discussion of serial ports under Linux may be found in the Serial-HOWTO.

When using a slow serial printer with flow control, you may find that some of your print jobs get truncated. This may be due to the serial port, whose default behavior is to purge any untransmitted characters from its buffer 30 seconds after the port device is closed. The buffer can hold up to 4096 characters, and if your printer uses flow control and is slow enough that it can't accept all the data from the buffer within 30 seconds after printing software has closed the serial port, the tail end of the buffer's contents will be lost. If the command cat file > /dev/ttyS2 produces complete printouts for short files but truncated ones for longer files, you may have this condition.

The 30 second interval can be adjusted through the "closing_wait" commandline option of setserial (version 2.12 and later). A machine's serial ports are usually initialized by a call to setserial in the rc.serial boot file. The call for the printing serial port can be modified to set the closing_wait at the same time as it sets that port's other parameters.

4.4. USB Devices

I don't have any USB devices to play with, so all I can offer are pointers. Once set up, you end up with the device file /dev/usb/lp0, much as you do with parallel ports, which will work fine in printcap or as a PDQ local-port device.

USB is documented at the Linux USB Website.

5.1. Postscript

As for what printers do work with Linux, the best choice is to buy a printer with native PostScript support in firmware. Nearly all Unix software that produces printable output produces it in PostScript, so obviously it'd be nice to get a printer that supports PostScript directly. Unfortunately, PostScript support is scarce outside the laser printer domain, and is sometimes a costly add-on.

Unix software, and the publishing industry in general, have standardized upon Postscript as the printer control language of choice. This happened for several reasons:

5.2. Non-Postscript

Failing the (larger) budget necessary to buy a Postscript printer, you can use any printer supported by Ghostscript, the free Postscript interpreter used in lieu of actual printer Postscript support. Note that most Linux distributions can only ship a somewhat outdated version of Ghostscript due to the license. Fortunately, there is usually a prepackaged up to date Ghostscript made available in each distribution's contrib area.

Adobe now has a new printer language called "PrintGear". I think it's a greatly simplified binary format language with some Postscript heritage but no Postscript compatibility. And I haven't heard of Ghostscript supporting it. But some PrintGear printers seem to support another language like PCL, and these printers will work with Linux (iff the PCL is implemented in the printer and not in a Windows driver).

Similarly, Adobe offers a host-based Postscript implementation called PressReady. This works much like Ghostscript does to provide Postscript support for a non-Postscript printer, but has the disadvantage that it runs only on Windows.

5.3. What printers work?

If you want to buy a printer, you can look in several places to see if it will work. The cooperatively maintained Printing HOWTO printer database aims to be a comprehensive listing of the state of Linux printer support. A summary of it is below; be sure to check online for more details and information on what driver(s) to use.

Ghostscript's printer compatibility page has a list of some working printers, as well as links to other pages.

Dejanews contains hundreds of "it works" and "it doesn't work" testimonials. Try all three, and when you're done, check that your printer is present and correct in the database, so that it will be listed properly in this document in the future.

If you're lazy, I keep a short list of suggested printers on my website. These center around color inkjets and low-cost laser devices; fully compatible mid-range and high-end devices are much easier to find. You can even help support this document and the website by buying from buy.com or outpost.com through me.

5.4. How to buy a printer

It's a bit difficult to select a printer these days; there are many models to choose from. Here are some shopping tips:

6.1. LPD

LPD, the original BSD Unix Line Printer Daemon, has been the standard on Unix for years. It is available for every style of Unix, and offers a rather minimal feature set derived from the needs of timesharing-era computing. Despite this somewhat peculiar history, it is still useful today as a basic print spooler. To be really useful with modern printer, a good deal of extra work is needed in the form of companion filter scripts and front-end programs. But these exist, and it does all work.

LPD is also the name given to the network printing protocol by RFC 1179. This network protocol is spoken not only by the LPD daemon itself, but by essentially every networked print server, networked printer, and every other print spooler out there; LPD is the least common denominator of standards-based network printing.

LPRng (see Section 6.3) is a far better implementation of the basic LPD design than the regular one; if you must use LPD, consider using LPRng instead. There is far less voodoo involved in making it do what you want, and what voodoo there is is well documented.

There are a large number of LPD sources floating around in the world. Arguably, some strain of BSD Unix is probably the official owner, but everyone implements changes willy-nilly, and they all cross-pollinate in unknown ways, such that it is difficult to say with certainty exactly which LPD you might have. Of the readily available LPDs, VA Linux offers one with a few minor modifications that make the user interface much more flexible. The SourceForge LPD supports command-line option specification with a -o flag; options are then passed through to filters. This is similar to the features offered by a number of traditional Unix vendors, and similar to (although incompatible with) LPRng's -z option mechanism.

6.2. PDQ

PDQ is a non-daemon-centric print system which has a built-in, and sensible, driver configuration syntax. This includes the ability to declare printing options, and a GUI or command line tool for users to specify these options with; users get a nice dialog box in which to specify resolution, duplexing, paper type, etc (see Figure 7).

Running all of the filters as the user has a number of advantages: the security problems possible from Postscript are mostly gone, multi-file LaTeX jobs can be printed effectively as dvi files, and so forth.

This is what I now use; I've written driver spec files for my printers, and there are several included with the distribution, so there are plenty of examples to base yours on. I've also written a few tools to automate driver spec generation to help the rest of you.

PDQ is not without flaws: most notably it processes the entire job before sending it to the printer. This means that, for large jobs, PDQ may simply be impractical—you can end up with hundreds of megs being copied back and forth on your disk. Even worse, for slow drivers like the better quality inkjet drivers, the job will not start printing until Ghostscript and the driver have finished processing. This may be many minutes after submission.

If you have many users, many printers, or anything else complex going on, I recommend using PDQ as a front-end to LPD-protocol based network printing (you can print via the lpd protocol to the local machine). In most such situations, rather than using the traditional BSD lpd as the back-end, I recommend LPRng:

6.3. LPRng

Some Linux vendors (including Caldera) provide LPRng, a far less ancient LPD print spooling implementation. LPRng is far easier to administer for large installations (read: more than one printer, any serial printers, or any peculiar non-lpd network printers) and has a less frightfully haphazard codebase than does stock lpd. It can even honestly claim to be secure - there are no SUID binaries, and it supports authentication via PGP or Kerberos.

LPRng also includes some example setups for common network printers - HP LaserJets, mainly, that include some accounting abilities. If you'd like more information on LPRng, check out the LPRng Web Page. LPRng uses more or less the same basic filter model as does BSD lpd, so the LPD support offered by my website applies to LPRng as well. This can help you effectively use free software drivers for many printers.

LPRng is distributed under either the GPL or an Artistic license.

6.4. PPR

PPR is a Postscript-centric spooler which includes a rudimentary Postscript parsing ability from which it derives several nice features. It includes good accounting capabilities, good support for Appletalk, SMB, and LPD clients, and much better error handling than lpd. PPR, like every other spooler here, can call Ghostscript to handle non-Postscript printers.

I only recently found out about PPR; I don't know of anyone who has tried it. It was written by, and is in use at, Trinity College. The license is BSD-style; free for all use but credit is due.

According to the documentation, it's somewhat experimental. Malformed Postscript jobs won't print; instead they bounce, and it's up to the user to fix the Postscript. This may make it unsuitable for some environments, although most users generate Postscript with a small handful of well-characterized Postscript generators, so it probably wouldn't be that big an issue.

6.5. CUPS

One interesting newcomer on the scene is CUPS, an implementation of the Internet Printing Protocol (IPP), an HTTP-like RFC standard replacement protocol for the venerable (and klunky) LPD protocol. The implementation of CUPS has been driven by Michael Sweet of Easy Software Products; CUPS is distributed under the GPL.

I've finally done some work with CUPS, and it does work as advertised. There are a number of very good features in it, including sensible option handling; web, gui, and command-line interfaces; and a mime-based filtering system with strong support for Postscript. Since it is so new, however, it does have a number of quirks, and it is hard to recommend for large or secure installations at this time (as of version 1.1). It is a fine solution, however, for smaller installations or especially larger installatons with trusted users.

Like other systems, CUPS can be used with most existing drivers. Unfortunately, it's a bit tricky to configure an arbitrary driver for use with CUPS—at least if you want all the options to work—so it's best to find a preexisting PPD file and filter script to make your driver go. There are at least four sets of drivers which you can use with CUPS:

The third-party program XPP (see Figure 4) offers a very nice graphical interface to the user functionality of CUPS, including an marvelous interface to print-time options (shown in Figure 5). For information on using XPP, see Section 3.3.2.

7.1. PDQ

Pdq stands for "Print, Don't Queue", and the way it works reflects this design. The following sequence of events happens when you use PDQ to print:

• You run pdq or xpdq, specifying a file.

• You specify a printer.

• You specify the settings for the various options and arguments defined in the printer's PDQ driver file (duplex, copies, print quality, and so forth).

• PDQ analyzes the contents of what you printed, and follows the instructions in the PDQ driver file which tell it how to process your data for this printer with your options.

• PDQ sends the processed data to the printer according to the interface defined for that printer (straight to /dev/lp0, or to an LPD daemon on the network, over the network to an Apple or Microsoft system, or even to a fax machine).

• If PDQ can't send the data to the printer right away, it spawns a background process to wait and try again until it succeeds or hits a time limit.

7.2. LPD

Lpd stands for Line Printer Daemon, and refers in different contexts to both the daemon and the whole collection of programs which run print spooling. These are:

So how does it fit together? The following things happen:

1. At boot time, lpd is run. It waits for connections and manages printer queues.

2. A user submits a job with the lpr command or, alternatively, with an lpr front-end like GPR, PDQ, etc. Lpr contacts lpd over the network and submits both the user's data file (containing the print data) and a control file (containing user options).

3. When the printer becomes available, the main lpd spawns a child lpd to handle the print job.

4. The child lpd executes the appropriate filter(s) (as specified in the if attribute in /etc/printcap) for this job and sends the resulting data on to the printer.

The lp system was originally designed when most printers were line printers - that is, people mostly printed plain ascii. By placing all sorts of magic in the if filter, modern printing needs can be met with lpd (well, more or less; many other systems do a better job).

There are many programs useful for writing LPD filters. Among them are:

8.1. Configuring PDQ

PDQ can be configured by either the superuser or by a joeuser. Root's changes are made to /etc/printrc, and affect everyone, while joeuser can only modify his personal .printrc. Everything applies to both types of configuration.

If PDQ is not available prepackaged for your distribution, you should obtain the source distribution from the PDQ web page and compile it yourself. It is an easy compile, but you must first be sure to have installed the various GTK development library packages, the C library development package, the gcc compiler, make, and possibly a few other development things.

8.2. Configuring LPD

Most Linux systems ship with LPD. This section describes a very basic setup for LPD; further sections detail the creation of complex filters and network configuration.

8.3. Large Installations

Large installations, by which I mean networks including more than two printers or hosts, have special needs. Below are some tips. For really large environments, merely distributing printcap/filter information becomes a difficult problem; the Cisco Enterprise Print System addresses this and is probably either a good starting point or a nearly complete solution, depending on your needs. Medium to large environments can be well supported by native LPRng features.

• Each printer should have a single point of control, where an administrator can pause, reorder, or redirect the queue. To implement this, have everyone printing to a local server, which will then queue jobs and direct them to the proper printer. For large campuses or distributed networks, have one server per building or other suitable network subset.

• Use LPRng, at least on servers; the BSD LPD is too buggy for "real" use. So is CUPS, at least right now in mid-2000. But don't take my word for it—you should test a number of spoolers and see which suits you best.

• Client systems should not have unique printing configurations. To implement this, use LPRng's extended printcap syntax so that you have one printcap to use everywhere. CEPS provides for this by building atop a lightweight distributed database instead of traditional printcap files.

• Print queues should not be named for make or model; name print queues for something sensible like location (floor2_nw) or capability (color_transparency). Three years from now, when a printer breaks, you will be able to replace it with a different make or model without causing confusion.

• Operate a web page which shows detailed information on each printer, including location, capabilities, etc. Consider having it show the queue and include a button to remove jobs from the queue. Complex networked environments are unmanagable for users without proper documentation.

• On Unix systems, use PDQ or the like to allow selection of print job attributes such as duplex or paper size, and to force users to run all Ghostscript processing under the proper user ID. If you have all Postscript printers (as is best), you can also select from the GPR or XPP front-ends; both are prettier.

• On Windows and Apple systems, use either the platform-specific drivers everywhere (Samba supports the Windows automagical driver-download mechanism) or, better, use generic Postscript drivers everywhere. Do not mix and match; primitive word processors often produce different output when the installed printer driver changes; users cannot deal with output that vaires depending on the particular client/printer pair.

• If at all possible, buy a large-volume printer for large-volume printing. If on a budget, use LPRng's multiple printers/one queue facility and assign a babysitter; printers are complex mechanical devices that will often jam and run out of paper in such configurations.

• Do not feel that printers must be plugged into workstations; Ethernet "print servers" now cost under $100. The ability to locate printers anywhere you can network is a big improvement over forced location near a host; locate printers in sensible, central locations.

• Use any SNMP trap or other monitoring/alert facility available to you - someone should be tasked with running around and fixing printers with no ink or paper. Npadmin (see Section 11.10.1) can be used to do some management operations with SNMP printers.

8.4. Accounting

Regular LPD provides very little to help you with accounting. You can specify the name of an accounting file in the af printcap attribute, but this is merely passed as an argument to your if filter. It's up to you to make your if filter write entries to the accounting file, and up to you to process the accounting file later (the traditional format is mainly useful for line printers, and is nontrivial to parse in Perl, so there's no reason to preserve it). Also, if you're using my lpdomatic program as your filter, you'll need to make changes, since it depends on being given a configuration file as the ``accounting'' file name.

Ghostscript provides a PageCount operator that you can use to count the number of pages in each job; basically you just tack a few lines of postscript onto the end of the job to write an accounting file entry; for the best example of this see the file unix-lpr.sh in the Ghostscript source distribution.

Note that the unix-lpr implementation of accounting writes to a file from the Ghostscript interpreter, and is thus incompatible with the recommended -dSAFER option. A better solution might be to query the printer with a PJL command after each job, or to write a postscript snippet that prints the pagecount on stdout, where it can be captured without having to write to a file.

The LPRng print spooler includes an HP-specific sample implementation of accounting; I assume that it queries the printer with PJL. This technique should work for most PJL, Postscript, or SNMP printers with which you have two-way communications.

If you have a networked printer that supports SNMP, you can use the npadmin program to query a pagecount after each job. This should work properly for all print jobs. See Section 11.10.1 for more information on npadmin.

9.1. Red Hat

Red Hat has a GUI printer administration tool called printtool which can add remote printers and printers on local devices. It lets you choose a ghostscript-supported printer type and Unix device file to print to, then installs a print queue in /etc/printcap and uses a filter program from the rhs-printfilters package to support postscript and other common input types. This solution works fairly well, and is trivial to setup for common cases.

Where Red Hat fails is when you have a printer which isn't supported by their standard Ghostscript (which is GNU rather than Aladdin Ghostscript, and which supports fewer printers). Check in the printer compatibility list above (or online) if you find that you can't print properly with the stock Red Hat software. If your printer isn't supported by Red Hat's tools, you may need to install a contributed verison of Aladdin Ghostscript, and will probably also be better off if you use the lpdomatic or apsfilter packages, which know all about the printers supported by late-model Ghostscripts, and others besides.

In future versions of Red Hat the printtool will be reimplemented to support a larger list of printers and with the intent to support an eventual rhs-printfilters replacement (the current filter has difficulty with many common printers like some non-PCL DeskJets and most Lexmarks). Some VA Linux-developed PPD features may be incorporated, as well.

9.2. Debian

Debian offers a choice between plain LPD, LPRng, or CUPS; LPRng or CUPS are probably the better choices. I believe Debian also offers a choice of printer configuration tools; apsfilter version 5 or later is probably your best bet, since that verison adds support for LPRng and Ghostscript's uniprint driver scheme. Red Hat's printtool is also supported, for those who like GUI administration tools.

9.3. SuSE

The printing system on SuSE Linux is based on apsfilter, with some enhancements; SuSE's apsfilter will recognize all common file formats (including HTML, if html2ps is installed). There are two ways to setup printers on SuSE systems:

• YaST will let you configure "PostScript", "DeskJet" and "Other printers", supported by Ghostscript drivers; it's also possible to setup HP's GDI printers (DeskJet 710/720, 820, 1000, via the "ppa" package). YaST will provide /etc/printcap entries for every printer ("raw", "ascii", "auto" and "color", if the printer to configure is a color printer). YaST will create spool directories and it will arrange apsfilterrc files, where you're able to fine tune some settings (Ghostscript preloads, paper size, paper orientation, resolution, printer escape sequences, etc.). With YaST it's also possible to setup network printers (TCP/IP, Samba, or Novell Netware Printer).

• In addition SuSE includes the regular SETUP program from the original apsfilter package (with some enhancements); run lprsetup to invoke this configuration script. Once you get accustomed to its GUI, you'll be able to configure local and network printers.

The SuSE installation manual explains both of these setup procedures.

Wolf Rogner reported some difficulties with SuSE. Apparently the following bugs may bite:

• Apsfilter's regular SETUP script is a bit broken, as are the KDE setup tools. Use YaST. [ Ed: does this still apply? It's been some time sice Wolf's report. ]

• For networked printers that need to be fed from Ghostscript, you'll need to first uncomment the line REMOTE_PRINTER="remote" in /etc/apsfilterrc. Then run YaST to configure the printer and, under Network configurations, set up a remote printer queue.

• YaST's setup doesn't allow color laser printers, so configure a mono printer and then change mono to color everwhere in the printcap entry. You may have to rename the spool directory, too.

9.4. Caldera

Caldera ships LPRng. I have no idea what sort of setup tools they offer.

9.5. Corel

Corel is Debian-based, so all the Debian facts above should still apply. In addition, they've written their own setup tool, based on the sysAPS library which in turn uses my database. They've certainly done so as part of WordPerfect.

Corel operates a printing support newsgroup named corelsupport.linux.printing. The bulk of the traffic appears to be WordPerfect and Corel Linux related.

9.6. Mandrake

As of version 7.2b1, Mandrake ships with CUPS standard. The program QtCUPS is used to provide a clean GUI administration interface. Till went to some trouble to include as many drivers as possible, and they ship CUPS PPD files build with my own foomatic interface code.

I think Earlier Mandrake versions shipped with the Red Hat printtool.

9.7. Other Distributions

Please send me info on what other distributions do!

10.1. Invoking Ghostscript

Typically, Ghostscript will be run by whatever filter you settle upon (I recommend apsfilter or my own lpdomatic if your vendor didn't supply anything that suits you), but for debugging purposes it is often handy to run it directly.

gs -help will give a brief listing of options and available drivers (note that this list is the list of drivers compiled in, not the master list of all available drivers).

You might run gs for testing purposes like: `gs <options> -q -dSAFER -sOutputFile=/dev/lp1 test.ps'.

10.2. Ghostscript output tuning

There are a number of things one can do if Ghostscript's output is not satisfactory (actually, you can do anything you darn well please, since you have the source).

Some of these options, and others are described in the Ghostscript User Guide (the file Use.htm in the Ghostscript distribution; possibly installed under /usr/doc or /usr/share/doc on your system) are all excellent candidates for driver options in your filter system or PDQ driver declaration.

11.1. Printing to a Unix/lpd host

To allow remote machines to print to your printer using the LPD protocol, you must list the machines in /etc/hosts.equiv or /etc/hosts.lpd. (Note that hosts.equiv has a host of other effects; be sure you know what you are doing if you list any machine there). You can allow only certain users on the other machines to print to your printer by usign the rs attribute; read the lpd man page for information on this.

11.2. Printing to a Windows or Samba printer

There is a Printing to Windows mini-HOWTO out there which has more info than there is here.

11.3. Printing to a NetWare Printer

The ncpfs suite includes a utility called nprint which provides the same functionality as smbprint but for NetWare. You can get ncpfs from Metalab. From the LSM entry for version 0.16:

" With ncpfs you can mount volumes of your netware server under Linux. You can also print to netware print queues and spool netware print queues to the Linux printing system. You need kernel 1.2.x or 1.3.54 and above. ncpfs does NOT work with any 1.3.x kernel below 1.3.54. "

11.4. Printing to an EtherTalk (Apple) printer

The netatalk package includes something like nprint and smbclient. Others have documented the procedure for printing to and from an Apple network far better than I ever will; see the Linux Netatalk-HOWTO.

11.5. Printing to a networked printer

Many printers come with an ethernet interface which you can print to directly, typically using the LPD protocol. You should follow the instructions that came with your printer or its network adaptor, but in general, such printers are "running" lpd, and provide one or more queues which you can print to. An HP, for example, might work with a printcap like:

lj-5|remote-hplj:\ :sh:\ :sd=/var/spool/lpd/lj-5:\ :rm=printer.name.com:\ :rp=raw:

or, using the PDQ bsd-lpd interface arguments of REMOTE_HOST=printer.name.com and QUEUE=raw.

HP Laserjet printers with JetDirect interfaces generally support two built in lpd queues - "raw" which accepts PCL (and possibly Postscript) and "text" which accepts straight ascii (and copes automatically with the staircase effect). If you've got a JetDirect Plus3 three-port box, the queues are named "raw1", "text2", and so forth.

Note that the ISS company has identified an assortment of denial of service attacks which hang HP Jetdirect interfaces. Most of these have been addressed beginning in Fall 98. These sorts of problems are common in embedded code; few appliance-style devices should be exposed to general Internet traffic.

In a large scale environment, especially a large environment where some printers do not support PostScript, it may be useful to establish a dedicated print server to which all machines print and on which all ghostscript jobs are run. This will allow the queue to be paused or reordered using the topq and lprm commands.

This also allows your Linux box to act as a spool server for the printer so that your network users can complete their print jobs quickly and get on with things without waiting for the printer to print any other job that someone else has sent. This is suggested too if you have unfixable older HP Jetdirects; it reduces the likelihood of the printers wedging.

To do this, set up a queue on your linux box that points at the ethernet equipped HP LJ (as above). Now set up all the clients on your LAN to point at the Linux queue (eg lj-5 in the example above).

Some HP network printers apparently don't heed the banner page setting sent by clients; you can turn off their internally generated banner page by telnetting to the printer, hitting return twice, typing "banner: 0" followed by "quit". There are other settings you can change this way, as well; type "?" to see a list.

The full range of settings can be controlled with HP's webJetAdmin software. This package runs as a daemon, and accepts http requests on a designated port. It serves up forms and Java applets which can control HP printers on the network. In theory, it can also control Unix print queues, but it does so using the rexec service, which is completely unsecure. I don't advise using that feature.

11.6. Running an if for remote printers with old LPDs

One oddity of older versions of lpd is that the if is not run for remote printers. (Versions after 0.43 or so have the change originated on FreeBSD such that the if is always run). If you find that you need to run an if for a remote printer, and it isn't working with your lpr, you can do so by setting up a double queue and requeueing the job. As an example, consider this printcap:

lj-5:\ :lp=/dev/null:sh:\ :sd=/var/spool/lpd/lj-5:\ :if=/usr/lib/lpd/filter-lj-5: lj-5-remote:sh:rm=printer.name.com:\ :rp=raw:sd=/var/spool/lpd/lj-5-raw:

in light of this filter-lj-5 script:

#!/bin/sh gs <options> -q -dSAFER -sOutputFile=- - | \ lpr -Plj-5-remote -U$5

The -U option to lpr only works if lpr is run as daemon, and it sets the submitter's name for the job in the resubmitted queue correctly. You should probably use a more robust method of getting the username, since in some cases it is not argument 5. See the man page for printcap.

11.7. From Windows.

Printing from a Windows (or presumably, OS/2) client to a Linux server is directly supported over SMB through the use of the SAMBA package, which also supports file sharing of your Linux filesystem to Windows clients.

Samba includes fairly complete documentation, and there is a good Samba FAQ which covers it, too. You can either configure a magic filter on the Linux box and print PostScript to it, or run around installing printer-specific drivers on all the Windows machines and having a queue for them with no filters at all. Relying on the Windows drivers may in some cases produce better output, but is a bit more of an administrative hassle if there are many Windows boxen. So try Postscript first. Modern versions of Samba should support the automagical driver download mechanism offered by Windows NT servers to deal with this problem.

With PDQ, you should configure Samba to run the pdq command with appropriate arguments instead of the lpr command that it defaults to running. I believe that Samba will run pdq as the proper user, so it should work well this way. There are several Samba options that you should adjust to do this:

11.8. From an Apple.

Netatalk supports printing from Apple clients over EtherTalk. See the Netatalk HOWTO Page for more information.

11.9. From Netware.

The ncpfs package includes a daemon named pserver which can be used to privide service to a NetWare print queue. From what I understand, this system requires a Bindery-based NetWare, ie 2.x, 3.x, or 4.x with bindery access enabled.

For more information on ncpfs and it's pserver program, see the ncpfs FTP site.

11.10. Networked Printer Administration

Most networked printers support some method of remote administration. Often there are easy-to-use web pages for configuration. More usefully, there is often support for SNMP management. Typically you can find out interesting information on printer status like ink and paper levels, print volumes, and so forth, and you can usually change certain settings. SNMP printer control, and a number of other printing-related things, are being standardized by the IEEE's Printer Working Group

12.1. The Ghostscript Windows redirector

There is now a Windows printer driver available (called mswinpr2) that will run a print job through Ghostscript before finally printing it. (Rather like an if filter in Unix's LPD). There is also a new Ghostscript driver which will print using Windows GDI calls. Taken all together, this allows a Windows machine to print PostScript to a Windows-only printer through the vendor's driver.

If you get that working, you can then follow the instructions above for printing to a Windows printer over the network from Linux to let Unix (and other Windows, Mac, etc) hosts print to your lemon printer.

12.2. HP Winprinters

Some HP printers use "Printing Performance Architecture" (marketingspeak for "we were too cheap to implement PCL"). This is supported in a roundabout way via the pbm2ppa translator written by Tim Norman. Basically, you use ghostscript to render PostScript into a bitmapped image in pbm format and then use pbm2ppa to translate this into a printer-specific ppa format bitmap ready to be dumped to the printer. This program may also come in ghostscript driver format by now.

The ppa software can be had from the ppa home page; pbm2ppa supports some models of the HP 720, 820, and 1000; read the documentation that comes with the package for more details on ppa printer support.

12.3. Lexmark Winprinters

Most of the cheap Lexmark inkjets use a proprietary language and are therefore Winprinters. However, Henryk Paluch has written a program which can print on a Lexmark 7000. Hopefully he'll be able to figure out color and expand support to other Lexmark inkjets. See here for more info.

Similarly, there are now drivers for the 5700, 1000, 1100, 2070, 3200, and others. See the supported printers listing above, and my web site, for more information on obtaining these drivers.

13.1. Using a faxmodem

There are a number of fax programs out there that will let you fax and receive documents. One of the most powerful is Sam Leffler's HylaFAX. It supports all sorts of things from multiple modems to broadcasting.

SuSE ships a Java HylaFax client which allegedly works on any Java platform (including Windows and Linux). There are also non-Java fax clients for most platforms; Linux can almost certainly handle your network faxing needs.

Also available, and a better choice for smaller installations, is efax, a simple program which sends and receives faxes. The getty program mgetty can receive faxes using efax (and do voicemail or interactive logins).

13.2. Using the Remote Printing Service

There is an experimental service offered that lets you send an email message containing something you'd like printed such that it will appear on a fax machine elsewhere. Nice formats like postscript are supported, so even though global coverage is spotty, this can still be a very useful service. For more information on printing via the remote printing service, see the Remote Printing WWW Site.

13.3. Commercial Faxing Services

A number of companies operate web-based faxing services. EFax, in particular, offers free inbound faxes (to your own dedicated fax number, no less) via email, and fax transmission for a fee. Other companies offer similar services.

14.1. Markup languages

Most markup languages are more suitable for large or repetitive projects, where you want the computer to control the layout of the text to make things uniform.

14.2. WYSIWYG Word Processors

There is no shortage of WYSIWYG word processing software. Several complete office suites are available, including one that's free for personal use (StarOffice).

Other vendors should feel free to drop me a line with your offerings.

15.1. Ghostscript and Photos

Ghostscript has some difficulties rendering color photographs through most drivers. The problems are several:

• Many drivers have poorly tuned color support. Often the colors don't match the Windows driver output or the screen. OTOH, all drivers, and Ghostscript as a whole, have readily adjustable color support; the "Gamma" settings (see Section 10.2.2) are one thing to play with, and there are others documented in Ghostscript's Use.htm documentation file.

• I'm only aware of one Ghostscript driver with support for 6 and 7 color printing; it's in beta at the moment and supports most Epson Stylus Photo models. It is rumoured to produce better color than the Windows driver (!). The Ghostscript driver core itself provides no support for non CMYK or RGB colors; arguably, some work to put that there is needed.

• Ghostscript often ends up dithering coarsely, or generating printouts with artifacts like banding. The dithering can usually be corrected; see Section 10.2.3, and read the documentation for your driver.

That said, the obvious solution for now is to use non-Ghostscript software for printing photos, and indeed, such things do exist. The main contender is the print plugin in the Gimp, which supports pixel-for-pixel printing on Epson Styluses and Postscript printers (with basic PPD support). That Epson Stylus portion of that driver is available for Ghostcript, as well, as the stp driver. Also possible to use for this purpose are the assorted external pnm-to-foo programs used to print on printers like the cheap Lexmarks; these print attempt to print pixmaps pixel-for-pixel.

The best solution, of course, is to buy a Postscript printer; such printers can usually be completely controlled from available free software, and will print to the full capability of the printer.

15.2. Paper

Color inkjets are extremely dependent on the paper for good output. The expensive glossy coated inkjet papers will allow you to produce near-photographic output, while plain uncoated paper will often produce muddy colors and fuzzy details. Nonglossy coated inkjet papers will produce results in between, and are probably best for final prints of text, as well. Stiffer glossy coated "photo" papers will produce similar output to lighter-weight glossy papers, but will feel like a regular photo.

15.3. Printer Settings

For photo output on most color inkjets, you should use the most highly interlaced (and slowest) print mode; otherwise solid regions may have banding or weak colors. Generally with Ghostscript this is what will happen when you pick the highest resolution. With Postscript printers, you may need to add a snippet to the prologue based on the settings available in the PPD file. The Gimp's PPD support doesn't include (printer-specific) print quality settings, but I added one in an ugly way for my own use; contact me if you'd like that. If you use PDQ or CUPS, you can easily control all the printer settings you need. VA Linux's libppd and the GPR front-end can also add these options for Postscript printers.

15.4. Print Durability

Color inkjet printouts usually fade after a few years, especially if exposed to lots of light and air; this is a function of the ink. Printers with ink-only consumables like the Epsons and Canons can buy archival inks, which are less prone to this problem. Newer printers often use pigment-based inks, which don't fade as much as the older dye-based ink did. No inkjet output is really particularly good for long-term archival use. Write the bits to a CD-R and store that instead.

15.5. Shareware and Commercial Software

There's a program called xwtools which supports photo printing with all the bells and whistles on an assortment of Epson, HP, and Canon printers. Unfortunately, it was written under NDA, so comes without source. Unless you use it for the Epson Stylus Color 300 on Linux x86, it costs E15 for personal use; commercial pricing is unknown.

The ESP Print Pro package from Easy Software supports some printers which might otherwise be unsupported. These drivers are not reported to be very well-tuned for photos, but they do work.

16.1. PostScript

Ghostscript has an X11 driver best used under the management of the PostScript previewer gv. The latest versions of these programs should be able to view PDF files, as well. Note that gv has replaced the older previewer "Ghostview"; the new user interface is mch prettier and featureful that ghostview's plain old Athena GUI.

16.2. TeX dvi

TeX DeVice Independent files may be previewed under X11 with xdvi. Modern versions of xdvi call ghostscript to render PostScript specials.

A VT100 driver exists as well. It's called dgvt. Tmview works with Linux and svgalib, if that's all you can do.

16.3. Adobe PDF

Adobe's Acrobat Reader is available for Linux; just download it from the Adobe web site.

You can also use xpdf, which is free software, and I believe gv supports viewing PDF files with gs under X11.

17.1. Setting up in printcap

Lpd provides five attributes which you can set in /etc/printcap to control all the settings of the serial port a printer is on. Read the printcap man page and note the meanings of br#, fc#, xc#, fs# and xs#. The last four of these attributes are bitmaps indicating the settings for use the port. The br# atrribute is simply the baud rate, ie `br#9600'.

It is very easy to translate from stty settings to printcap flag settings. If you need to, see the man page for stty now.

Use stty to set up the printer port so that you can cat a file to it and have it print correctly. Here's what `stty -a' looks like for my printer port:

dina:/usr/users/andy/work/lpd/lpd# stty -a < /dev/ttyS2 speed 9600 baud; rows 0; columns 0; line = 0; intr = ^C; quit = ^\; erase = ^?; kill = ^U; eof = ^D; eol = <undef>; eol2 = <undef>; start = ^Q; stop = ^S; susp = ^Z; rprnt = ^R; werase = ^W; lnext = ^V; min = 1; time = 0; -parenb -parodd cs8 hupcl -cstopb cread -clocal -crtscts -ignbrk -brkint -ignpar -parmrk -inpck -istrip -inlcr -igncr -icrnl ixon -ixoff -iuclc -ixany -imaxbel -opost -olcuc -ocrnl -onlcr -onocr -onlret -ofill -ofdel nl0 cr0 tab0 bs0 vt0 ff0 -isig -icanon -iexten -echo -echoe -echok -echonl -noflsh -xcase -tostop -echoprt -echoctl -echoke

The only changes between this and the way the port is initialized at bootup are -clocal, -crtscts, and ixon. Your port may well be different depending on how your printer does flow control.

You actually use stty in a somewhat odd way. Since stty operates on the terminal connected to it's standard input, you use it to manipulate a given serial port by using the `<' character as above.

Once you have your stty settings right, so that `cat file > /dev/ttyS2' (in my case) sends the file to the printer, look at the file /usr/src/linux/include/asm-i386/termbits.h. This contains a lot of #defines and a few structs (You may wish to cat this file to the printer (you do have that working, right?) and use it as scratch paper). Go to the section that starts out

/* c_cflag bit meaning */ #define CBAUD 0000017

This section lists the meaning of the fc# and fs# bits. You will notice that the names there (after the baud rates) match up with one of the lines of stty output. Didn't I say this was going to be easy?

Note which of those settings are preceded with a - in your stty output. Sum up all those numbers (they are octal). This represents the bits you want to clear, so the result is your fc# capability. Of course, remember that you will be setting bits directly after you clear, so you can just use `fc#0177777' (I do).

Now do the same for those settings (listed in this section) which do not have a - before them in your stty output. In my example the important ones are CS8 (0000060), HUPCL (0002000), and CREAD (0000200). Also note the flags for your baud rate (mine is 0000015). Add those all up, and in my example you get 0002275. This goes in your fs# capability (`fs#02275' works fine in my example).

Do the same with set and clear for the next section of the include file, "c_lflag bits". In my case I didn't have to set anything, so I just use `xc#0157777' and `xs#0'.

17.2. Older serial printers that drop characters

Jon Luckey points out that some older serial printers with ten-cent serial interfaces and small buffers really mean stop when they say so with flow control. He found that disabling the FIFO in his Linux box's 16550 serial port with setserial corrected the problem of dropped characters (you apparently just specify the uart type as an 8250 to do this).