The ( Pre ) History of Color in Rokicki ' s dv i ps

In t h s paper I give an abbreviated history of the current scheme for using color with Rokicki's dvips program up to the end of 1993. The real story begins in early 1990, when a local user asked if I could add to the fledgling FoilTEX project support for color to take advantage of our new color printers. This started a major effort, in collaboration with Tom Rokicki, to find an efficient and simple method for specifying color in TEX documents. Introduction our new color printers. (At the time, I was not aware The \speci a1 command enables you to make use of special equipment that might be available to you, e.g., for printing books in glorious T~XIIicolor. D. E. Knuth As the quote above indicates, the grand wizard himself expected that color could (would?) be incorporated into TEX. He expected that this would be done through the use of \special commands to the dvi driver. In spite of t h s , not much was done with color for many years. Even SLITEX, where color is very desirable, was written to handle color in a rather cumbersome way. In t h s paper, I will describe the efforts that went into the design and development of the current color support in Tom Roluclu's dvi ps program. I consider this the prehistory of true color support because only some of the real color issues were addressed (and many of these were done via simple hacks). Before we go into dvi ps's method, let me set the stage. The availability of color PostScript printers created a need for a better method to handle color. In stepped a number of people, including Leslie Lamport who wrote co lor . s t y and Timothy Van Zandt who wrote PStricks. These all use literal PostScript commands passed to the dvi driver and then to the output PostScript file to create color effects. Unfortunately, there are problems with the use of literal PostScript. Namely, since each page is generally a self-contained graphics object, color effects on one page would not readily pass over to the next. Furthermore, effects at the end of a current page might trickle into the footnotes or page footer. This forces the use of these color utilities to be limited to very small parts of documents, e.g., single boxes. On the positive side, most dvi-to-Postscript drivers handle these lands of literal PostScript \speci a1 s, so usability/portability was not an issue. My character enters the story in early 1990, when a local user asked if I could add to the fledgling FoilTEX project support for color to take advantage of of the two packages mentioned above.) Using dvi 2ps and dvi a1 w, I massaged some primitive color support into these programs but certain obstacles came immediately to light. For example, in dvi a1 w, large characters and all rules are placed immediately on the first pass of the page, and then the graphcs environment is set up for the main characters. This is efficient for memory use but not for consistent color. If one tries to set a large square root sign in color, the opening check mark is fine but the long rule above the enclosed formula will always be black. Similar splits of colors occur for large brackets. Finally, I came across Rokicki's dvi ps and determined that this is very well suited for color. Some of the reasons for this are stated below. This started a collaboration with Tom about how one could achieve the desired effects. In the next section I discuss the relevant issues. Later I talk about the first real attempts at getting at the problem. Finally, we describe the current system in some detail and discuss some of the limitations.