Free Software Magazine and the TeX Users Group (TUG) both like to publish interviews. Recently, Gianluca Pignalberi of Free Software Magazine and Dave Walden of TUG both approached Frank Mittelbach about interviewing him. Rather than doing two separate interviews, Mittelbach, Pignalberi, and Walden decided on a combined interview in keeping with the mutual interests already shared by Free Software Magazine and TUG.

DW: Frank, please start by telling us a bit about yourself and how you got involved with LaTeX.

FM: I have lived with my family in Mainz (Germany) since the early eighties, i.e., by now the larger part of my life. Besides my primary hobby (typography), which can effectively be called my second job, I enjoy playing good board games, listening to jazz music, and reading (primarily English literature). Professionally I work for Electronic Data Systems where these days I’m responsible for concepts and implementation for remote monitoring and management of distributed systems and networks.

While I was studying Mathematics and Computer Science at the Gutenberg-University Mainz, I was first introduced to TeX and later LaTeX and, eventually, this got me interested in typesetting and in particular in research on algorithms for automated high quality typesetting.

Figure 1: Frank Mittelbach

During my student days in the eighties, a friend brought back a source tape from Stanford University containing something like TeX 1.1 and fascinating news about the quality of that program (back then we did our theses using a typewriter and either hand-pasting symbols in or, in case of some sophisticated IBM typewriter, changing the “ball” every couple of seconds). He tried to implement that program on the Multics system we had at the university and in fact succeeded—it probably was the first if not the only implementation of TeX on this operating system.

This way I got introduced to TeX and AmSTeX and typed my first paper, achieving beautiful results. The only catch was that back then the Stanford tape only contained Almost Computer Modern fonts in 200 dpi resolution (and no METAFONT) and the only graphical printing device available to us had a resolution of 72 dpi. So the output we got was of the size of the formula in figure 2 (or bigger).

Figure 2: oversized output

Wonderful to plaster the walls with, but not necessarily suitable for handing in your thesis. As a result, my friend finally had to type his diploma thesis in the traditional way, despite his efforts.

Sometime afterwards I was asked by the department to install a commercial TeX installation on our shiny new PCs and to give a series of lectures to students and professors on how to use it. And that distribution came with LaTeX 2.08 and a loose-bound copy of the manual (which later became Leslie Lamport’s book on LaTeX). LaTeX compared to plain TeX looked very good to me, but alas, when trying to produce any document with it, it died while loading the “article document style” due a lack of memory on those PCs. So my introduction to LaTeX stopped after I read the manual, and I was forced to develop my own TeX macro package that implemented similar concepts while requiring less memory. A year later it became possible to actually use LaTeX at the department, and we could retire my macro package.

But this initial exercise gave me a good insight into the inner workings and concepts of a system like LaTeX and enabled me later to constructively criticize certain aspects of LaTeX—something that eventually led to Leslie passing on the development and maintenance of LaTeX to me.

GP: Many of our readers are familiar with LaTeX, but for those who aren’t, can you introduce LaTeX to them?

FM: LaTeX is a batch-oriented typesetting system that uses the typesetting engine TeX or one of its variants (ε-TeX, pdfTeX, Omega).

The TeX program itself (developed by Professor Donald Knuth in the early eighties) is a programmable low-level typesetting engine, whose concepts and algorithms provide micro-typography (1) knowledge of highest quality in these days when this knowledge is slowly declining due to the fact that more and more authors are forced to become their own designer and typesetter without proper training. TeX is especially known for its excellent paragraph breaking algorithm and for its math formula typesetting capabilities, both of which are unsurpassed even though the program and its algorithms have been freely available for more than twenty years.

LaTeX is a macro package written for the TeX engine which allows the user to step back from the low-level formatting capabilities of TeX by providing higher-level interfaces that give the author the ability to mark up the text with logical markup rather than procedural markup (e.g., specifying that something is a list or a section, rather than stating that something should be set in a bold typeface with a little space above and below). The actual transformation of a LaTeX source into a typeset document is done with the help of “style sheets” and configuration adjustments that allow even radical changes to the design and layout in a consistent manner without touching or changing the source (2).

Historically speaking, LaTeX was largely influenced by a system called Scribe (by Brian Reid). In turn, LaTeX’s concept of logical markup was quite influential on HTML and various SGML/XML DTDs, as were its approaches for turning such logical markup into visual representation.

One of the differences between LaTeX and many other similar approaches is that the LaTeX language is in fact a community development: new packages that augment (or modify) LaTeX’s markup and typesetting functionalities are constantly appearing, so that these days LaTeX offers typesetting solutions for nearly every subject domain—as diverse as game typesetting (such as chess, go, or crossword puzzles), chemical formulas, or music. Another important difference is that, although LaTeX brought the concept of logical markup to a larger audience, it also provides ways to fine-tune the results (essentially providing interfaces to procedural markup), acknowledging the fact that no automated transformation of logical markup into a visual representation is able to automatically resolve all problems produced by the physical restrictions of the output format (e.g., line width or page size). While in certain applications such fine tuning adds no value (like database content publishing where full automation is required), it is crucial for typesetting high quality books and journal articles.

GP: How many people are officially part of the LaTeX Project? And how would you define the “LaTeX project”?

FM: The LaTeX Project Team is a fairly small (slowly changing) group of people who look after the LaTeX kernel and a small number of core packages that provide a stable basis for a huge number of constantly evolving packages and add-ons. Providing and guarding a stable core is (although not necessarily popular with everyone) an important part in keeping LaTeX alive as a language for document exchange. Current and past members of the team include Javier Bezos, Johannes Braams, David Carlisle, Michael Downes, Denys Duchier, Robin Fairbains, Morten Høgholm, Alan Jeffrey, Thomas Lotze, Chris Rowley, Rainer Schöpf, and Martin Schröder with varying degrees of involvement.

Figure 3: The LaTeX Project home page

Historically, the project took over maintenance and development of LaTeX 2.09 from Leslie Lamport in 1991. At one time the system was split into several incompatible variants that often prohibited successful processing by LaTeX at one site of documents created by LaTeX from a different site, even though system independence was originally one of the important goals of LaTeX as a documentation language for the scientific community. Another goal for the team was to address apparent deficiencies in the concepts of LaTeX 2.09. The project team addressed both issues in the early nineties with LaTeX 2e, which provided a stable and consolidated platform offering further development possibilities outside the kernel code.