History

In my first year as a physics student at Bielefeld University in 1992, I met Johannes Schlosser. What began as a study group for physics problems soon developed into a great friendship and an intense and fruitful collaboration on exploring what computers could do.

At this time a computer would be in ``text mode'' as standard and only for special problems would a ``graphics mode'' be activated. At this time only Apple computers broke with this paradigm and Microsoft was only a small company making the Microsoft DOS as well as a few text programs. A graphics resolution of 600:300 was a great resolution and color was a luxury we could not afford. We each bought an 8068 IBM compatible computer and began checking out what we could do with it. Switching the monochrome Hercules graphics card into graphics mode was a major challenge. We started using Turbo Pascal as our programming language of choice with a few Assembler inline additions.

But seeing the graphics on the screen is rather fleeting, so we also wanted to be able to print it. There was the option of a screen print, but that meant limiting the print output to the low resolution of the screen, even though I had invested in an NEC P6c, a color dot printer. So we wanted to make sure that we could print out our graphics with maximal resolution and in color on our printer. But how can you transfer a picture from the screen to the printer and increase the resolution? The solution we found was to have two sets of basic graphic routines: one would write to the screen, the other would write to a set of memory which was a bitmap of what we would then write to the printer. We realized that if you wrote your program so that all sizes would be relative to the X and Y dimensions of your medium, they would look right, no matter if they would be printed on a screen or printed to the printer.

On this level, we managed to program a graphical user interface which was supposed to be easy to adapt for new programs. If you wanted to print something you could use the same routines that you had used to create your graphics on the screen. It came as a total revelation to us when, a few years later, Peter Seroka introduced us to the X windows system common on Unix workstations. Being able to open windows on other computers seemed like a little miracle to us. A little earlier Linus Torwald had developed the Linux kernel and now the first working versions of the Linux operating systems became available and they were free! Another revelation there. So with the help of Linux (version 0.92, I believe) we could transform our computers (80286 IBM compatibles at this time) into Unix workstations.

This signaled the change from Pascal to C (which only our friend Tim Scheideler was using at the time) and X windows for our graphics needs. Previously all printing was different for all printers, but the people of Adobe Inc. had developed PostScript, a portable graphics language. And with the help of ghostscript you could (and still can) print any postscript file on any printer. So we decided to now use PostScript files as the output which could then be printed everywhere.

At this time both Johannes and myself were working on our Diplomarbeit (masters thesis) at Bielefeld University. Not surprisingly, we were both doing computational projects and were heavily relying on graphical output. At this point we had actually developed two separate graphics libraries. Johannes had true three-dimensional capabilities whereas I had developed basic menus, the printer support, and coordinate systems. Because Johannes was the much more organized of the two of us we used the structure of this graphics library and added my extensions into his code to create what remains to this day the basic foundation of the library: a set of basic two dimensional graphics routines which interface either xlib of the X windows system or PostScript routines, a library for the representation of three dimensional scenarios. This was supported by a set of menu functions that dealt with mouse interactions.

After we successfully used the graphics libraries for our Diplomarbeit Johannes went on to a rapidly advancing career at IBM Deutschland whereas I went on to graduate school at Oxford. For my graduate work I picked a numerical project again, but this time I dealt with complex fluids which needed a number of different representations that our graphics library did not yet support. I included support for density plots and contour lines, but whenever I wrote a new program I still copied lots of old code from the last programs to reuse them in the new programs. And then I realized that anytime I copied code from other programs, it meant that it should be part of the library and not part of the user code. It took quite a few numbers of iterations to reduce the amount of code the user had to write to use the graphics code.

In 1998, at the end of my D.Phil. (Ph.D.), I considered briefly founding a company for simulating complex fluids and using the graphics as a part of this commercial enterprise. But I soon realized that my heart was with science and I did not enjoy the prospects of meeting with venture capitalists to fund my software company (and that was at the beginning of the Internet bubble). Instead, I continued my academic career with postdoctoral positions at MIT and Edinburgh University. At each of these places I realized that scientists as a whole don't tent to have easy access to the data in their computer simulations and that the approach I had taken was quite unique.

Now I have a faculty position at North Dakota State University and I want to give you the ability to steer your computer simulations and to monitor them closely. Considering how much Johannes and I have benefited from the free software project we have decided to release our library under the GNU public license to make sure that it is available for free to anyone.