There were old good times when one could install magicfilter on his GNU/Linux system and send a file to a printer and the printer printed the file as expected. Since I started to use CUPS a few years ago it’s no longer true. Whenever I send some PostScript or PDF file to my printer, I’m full of tension about what the printer will produce. Sometimes I receive output looking like printed at 50 dpi, sometimes the page content is scaled by factor 2 horizontally and/or vertically, sometimes characters kernings are broken. Whether it is an e-mail, a Web page or a PostScript document, I’m never sure that I can print it usably. I often start printing by printing a test page, because what I can see in gv may not be what I receive on the paper.
I can understand that new drivers need debugging, which is the case of printing images with Guttenprint for example. I appreciate the hard work of people who try to provide their best, equipped just with incomplete specifications (if any specifications are available at all) and with testing and reverse engineering reports from voluntary testers. But I really don’t understand why people invent new great things which simply can’t work as before, which produce the same broken result on any printer and won’t get fixed for several years.
If you laugh now and point to other operating systems, calm down. I’ve recently tried to install Epson Stylus Photo R220 drivers on a Windows system and tried to perform nozzle check from it. When I perform nozzle check using the printer’s buttons or with Guttenprint on Linux, the printer just prints the testing pattern. Not so with the Epson Windows drivers, which add (without warning) an extra action before printing the pattern: an excessive head cleaning which wastes about 6% (!!) of the complete ink amount of full cartridges. Now I can understand why so many people complain that Epson printers waste a lot of ink.