Bart Childs

(completed 2010-05-19);

Bart Childs was extensively involved in the early distribution of TeX and early governance of TUG.



Dave Walden, interviewer:     Please tell me about yourself.

Bart Childs, interviewee:     My complete name is Selma Bart Childs. I was born in 1938 in Magnolia, Arkansas, and have been called Bart all my life except on a few occasions like when I was Childs, Selma B in the National Guard and other rare occasions. I asked where the name Bart came from and my sweet Mother said, I had heard it and I liked it. The origin of Selma is a longer story.

My folks were from the area of western Arkansas and eastern Oklahoma. My Father wrote in his first college theme

The first thing I remember is when I was four years old,
I was sitting in my Aunt Laura's lap at my Father's funeral.
She kept saying, ``Poor Mellie, what is she going to do with all these children?''
Five already and a sixth one on the way.

Well, Mellie's farm was next to her brother Selma's. Uncle Selma was the Father figure for my Dad and I was given his name which was an honor and a pain. There are several definitions for it and the one I often repeat is ``It is an Arabic girl's name which means fair.'' They were so far back in the hills that I don't think they knew it was a feminine name.

It has led to many incidents including inordinate delays in Security Clearances, being listed in a Who's Who in American Women, etc. but no fights. So now you know why you might find me as Bart Childs, S. Bart Childs, or Selma Bart Childs.

My great grandfather fought in the Civil War. He was wounded and died before it was over. About 125 years later, a cousin was frustrated when the Daughters of The Confederacy kept denying her daughter membership. My Mother helped her at the Little Rock Federal Cemetery and they found that he fought for the Union Army. The subject of which side he fought for was suppressed in the family for 125 years.

Regardless, both sets of my grandparents encouraged education which led to my parents being educators. My Father was a Professor of Agriculture at Southern Arkansas University (now) and my mother taught all levels in a number of public schools. My parents corrected my use of the English language even after I had been a full Professor for 30 years.

I was about a top ten-percent student in high school. I was not dedicated as an undergraduate and believe I was admitted to graduate school on a clerical error. Suddenly, I was in an environment with several excellent students and found the joy of real scholarship.

I earned three degrees in Civil Engineering from Oklahoma State University. The crux of my dissertation was published in the Quarterly of Applied Mathematics. I owe a lot to those wonderful comrades in graduate school and several professors who believed in my potential.

I finished my Master of Science in the summer of 1960 and taught a lot for the next three years while finishing my PhD coursework. I had no funding for the summer of 1963 and received an invitation to participate in a short course at the University of Houston. This was sponsored by the NSF and was for giving engineering educators the opportunity to learn about this new thing, the computer. Actually, I had applied and been rejected. They wanted permanent faculty and I was a lecturer, a graduate student in reality. A few slots were not filled and Elliott Organick, the director, invited me wanting to fill all the slots.

It was an eye opening experience. Elliott Organick, Bruce Arden, and others introduced the concepts of data structures, compilers, and operating systems while teaching us the MAD language, using the MESS executive. After I got back to Stillwater, I was in a quandary. I wanted to be a computer scientist! My advisor, Robert W. Little, recommended I get the ``union card'' and then find a computer job or look for a post doc in computing.

In 1965 I finished, took a job as an Assistant Professor of Mechanical Engineering at the University of Houston, and moved to that dynamic city. My job was to work on integrating computing into the mechanical engineering curriculum and each semester I would teach a class at NASA. I had a large number of NASA engineers and scientists in those classes and directed theses of many.

I had a great six years of that. In my fifth year, I took a sabbatical in the Maths department of the University of Glasgow with Ian Sneddon. What a wonderful experience. After returning, I married Shirley Marshall, a Houstonian of Czech background.

We moved to the University of Louisville in 1971 where I became a (too) young Chairman of the Applied Mathematics and Computer Science Department. Our daughter Meredith came along late in 1972. I lacked sufficient diplomacy to remain there and became a professor of computing science at Texas A&M University in 1974.

The summer of 1978 was another great time of change in my life. I started playing soccer and Shirley decided it was time for me to learn how to dance. Due to her Czech upbringing, she thought that everybody knows how to dance. She learned that was not true. She was an excellent dancer, I learned to be a decent polka, waltz, and C&W dancer.

We hosted the TUG meeting at Texas A&M University in 1990. We had an executive committee meeting before the conference started and followed it with a desert dinner of local specialties that Shirley fixed. As part of the entertainment we had Mexican food, a BBQ dinner, and a polka dance. We had a dozen local couples come out for a free dinner and be a partner for everybody that wanted to try. David Ness said that as he walked down the hall to an executive committee meeting he heard two comments: one said ``this was a great meeting and I loved the polka dance'' while a second said ``this was a great meeting but I could have done without the darned dance.''

I lost my Shirley to breast cancer in 2004.

DW:     Please tell me about your introduction to TeX.

BC:     I mentioned that with my first instruction on computing included that there was a lot of background behind computing. Before the next year was out, I learned about Don Knuth's Volume 1. I went to the library and was duly impressed. I bought my own copy and in my spare time perused it when I felt I could spend more than a few minutes. I just did not have the right background to pick it up and read it like a novel. After all, it was not done using Literate Programming.

Early in my career at Texas A&M, the 1977–78 school year to be exact, I was given the task to write the specifications for a computer system to be purchased for laboratory work in the Department of Industrial Engineering. I received the assignment because of my insistence the laboratory computer should be interactive and have some word processing software.

It was reasonably successful and the word processing software was done by the OEM that sold us the system, so we had the sources. Before long, I was the leader in making extensions to it. The primary output device was a Diablo spin writer. Two of the earliest features I added were somewhat proportional spacing and drawing integral signs by putting the printer in graphics mode and a lots of periods later I had a crude integral sign.

Norman Naugle and I had taught at NASA at the same time and he served on a number of the committees of my students. He came back to his alma mater before I arrived. We met again at the campus creamery and I told him that I wanted him to come see a word processor I was supporting. He was excited and told me that I would have to listen to him at least as long as he listened to me. We agreed.

A few days later he showed up as I was finishing a session with my graduate students. He listened dutifully and asked about many extensions that would be appropriate. Of course, nearly all were missing. He gave me a copy of Don's TeX and METAFONT book to study for a few days.

It did not take long for me to realize the orders of magnitude of differences between the best I could ever do and what Don had already done. So, I called Norman and said what can we do to get the software.

He quickly formulated the plan to use my access to plenty of good graduate students and beginning availability of minicomputers, although only 16-bit ones, and we would work together to port TeX in SAIL to some system. Norman purchased the TeX sources.

We had some proof of concept success and so I got the money to send a graduate student to a TeX Users Group meeting in Palo Alto. He came back and said that Dr. Knuth is throwing it away and writing it again. It stunned us and then with two or three times through it we began to understand what Literate Programming was as defined by the new version of TeX, in Pascal WEB.

By the time we got the new sources, Data General had given us a 32-bit virtual memory system, an MV-8000. The conversion was unbelievable. It was one of the few compilers that TeX did not find an error in. Well, not quite. One revision had an error resulting from a change but DG had found it by the time we identified the source that had caused it.

I started going to the meetings after that. David Kellerman approached Norman to be a candidate for TUG President. Norman declined and suggested me. David came to me after checking with the committee.

DW:     At readers can look up your various reports during the time (1983–1992) when you were active on the TUG board, as an officer, as president (1985–1989), and as a site coordinator. However, perhaps you can describe here some of the significant events and the feelings within the TeX community at the time for me and our readers.

BC:     I mentioned that our first introduction to TeX was the original version which was written in SAIL. SAIL was an Algol extension and commonly available on DEC-10 and 20 machines. These machines were fairly common at the leading computer science research universities. These were 36-bit machines with a limited address space virtual memory system (quite an oxymoron). DEC announced that their future emphasis would not include these systems. Don has written about his decision to write a new version of TeX and the reasons included: his openness in declaring it be public domain engendered a larger interested public than he expected, the limitations he placed in the data structures were restricting many users, and the plans for DEC's 36-bit machines pointed to a need for a different source language and broad support by an array of computers. Don surveyed many and concluded that Pascal was the “second favorite” language of a majority of users that would meet the necessary functionality for hosting the new TeX. Just a few years later it would almost certainly have led to a decision for C.

I wrote earlier that we were about to embark on a project of translating TeX in SAIL to (probably) Pascal (maybe PL/I). We had made some small successful steps and then learned about the new version in Pascal/WEB and started studying the sources waiting for the delivery of our new machine. The Pascal compiler did not come for a few months and when it came I was ready, more or less. Pascal was a new language for me but Don's documentation of the process, the vanilla versions of the tools that would be recreated once the change files were created, and the completeness of the system made it a straightforward task. Most of the delays were due to my misunderstandings and lack of knowledge of Pascal. We had a working and correct TeX well before we even had a laser printer. This was possible because of the completeness of Don's trip.tex.

Norman Naugle and I asked many sources for help, with little success. When we finally struck paydirt with QMS of Mobile, Alabama, we started being able to see the beautiful output. I am not sure of the details of how we got output for the first few months. We must have used a VAX in another department, probably ELEN. I/we created drivers for a number of printers and a few previewers (more later on this). Most of these were based on Don's dvitype which was created to be a template of a dvi reader, i.e. a printer driver.

We registered our lab with TUG as the distribution site for Data General MV systems and I was the site coordinator. The growth of open systems as well as the growth of the power in personal computers and tight coupling of the processor and graphics displays was the death knell for the proprietary super-minis. The site coordinator title lasted only a few years in reality.

DW:     What was the role of a site coordinator?

BC:     The role of a site coordinator was varied from one to the next. As a Data General coordinator, we only supported the virtual memory systems. We delivered a tape and sufficient instructions on how to install it. Most DG sites did not have the compiler so customization required their coming back to us. In several cases, we developed a new driver for printers they already owned. Usually they would buy a printer we already supported. In some cases we assisted the customer in installing a driver for a typesetter that came from a commercial company.

Our support was for one compatible line of computers. The University of Washington (Pierre MacKay and Rick Furuta) support was for unix systems. They were for a broad range of hardware and versions of unix.

I could have been considered as the site coordinator for Cray supercomputers. It was felt at Los Alamos Labs that their scientists should be able to use the same interface for TeX as for their number crunching. Surprisingly (to me), Cray had a good Pascal compiler and after being stymied by one of the rare “errors in TeX” we had a clean port. It was in use at Los Alamos and Livermore. The output was then previewed on SUN and Apollo workstations.

Site coordinators were members of the TUG Board. That was an exciting place to be for discussions of successes, failures, problems, and solutions. Like TUG as a whole, the members were interested in helping.

DW:     Our interview of Tomas Rokicki also mentions Norman Naugle (Tomas got his start with TeX at TAMU). Did you know Tomas there?

BC:     Tom was one of our best products from Texas A&M, alas he was not in my department, Computer Science, or Norman's, Mathematics. Tom was from the Dallas area and had a summer job at E-Systems before coming to TAMU as a freshman Electrical Engineering student. One of his mentors was Brian Oldham who had been a colleague of Norman in graduate school. He made sure that Tom and Norman got together. Norman used Tom as a student worker whenever he could, as he had done with a number of other bright youngsters.

I wrote the QMS driver in Pascal/WEB and then Norman hired Tom to make a VAX version. As I remember it, Tom had little interest in the WEB version. I believe that as soon as he had it working he rewrote it in C.

During this same time, the PC was becoming usable as far as only a few crashes per hour, sufficient memory and disk capacity, and compilers. Norman hired Tom to write a dvi previewer called Cdvi. It gave the masses the quick view of his/her typesetting.

For most of us, the limited memory and disk capacity was a problem that we just lived with. Tom's experience with the drivers and previewer along with knowledge gained in a study of discrete mathematics led to a really bright idea. He created the concept of the pk file, a packet pixel file. He wrote a program to create some statistics about the pixel files. He noted that some high proportion of the lines of pixels were identical to the previous (I think this was for the computer modern 10 point fonts at 300 dpi.) I think that high proportion was 73 percent. With this knowledge he was able to pack this more efficiently than generic packing, as like Zipping. He rewrote the driver and packed the files on the VAX. The results were significant. He did this as he graduated and was ready to leave Aggieland to attend graduate school at Stanford.

Don arranged for Tom to come a few months early and contribute to the TeX project. At a project meeting, Tom proposed the PK concept. Don responded with “write a proposal.” Tom did and Don returned it to him on the first day of that summer's TUG meeting, at Stanford. Tom showed the document to Norman and me. Never have I seen so much green ink on a few sheets of paper. Tom was really proud and deservedly so. It was full of compliments and many suggestions to make the process better! Norman was equally proud.

Tom created the popular dvips driver and the predecessor of the web2c system to convert Pascal/WEB code into C.

DW:     I know you will be giving a presentation on Literate Programming at the TUG2010 convention, but perhaps you can say something about LP here.

BC:     Don once made a comment that he thought Literate Programming may be the most significant lasting contribution of the TeX project. The quickness with which the Literate Programming column in the CACM was canceled was surprising to me. I will reflect on my thoughts in this vein in the paper at TUG2010. A quick glance at a few of these are:

DW:     Please say something about your life with TeX since your time on the TUG board and as a site coordinator.

BC:     I taught a senior seminar for several years. I required that all writing assignments be done in LaTeX and furnished templates for a proposal for a research paper, a resume, a book report, and an evaluation of each of the required 16 classes and the course as a whole. These were submitted electronically and in the later years I also required posting a WEB resume. These projects were required by shortly after midterm. The last quarter of the semester was spent in exit interviews and evaluations of the book report, research paper; and course evaluations were done in an exit interview.

The course evaluations and exit interviews often including comments like:

At the end of the interview, we stood up, shook hands, and I said “Graduate, work hard, and send money to your alma mater.” The long term feedback has been on the positive side although many obviously felt that it is not necessary to know that there are en-dashes, em-dashes, display mathematics, description items, and so on. I am sure there is not that many that are TeXies, other than those that go to graduate school.

DW:     Is TeX/LaTeX still used significantly at Texas A&M?

BC:     Many graduate students are instructed by their advisors that TeX/LaTeX is required. I know of several book projects that are being done in TeX/LaTeX. I get an occasional call requesting some help with nagging details.

Also, I still write a few literate programs. My handwriting has deteriorated and so I write a lot of my communications in LaTeX. Another common document is a recipe and the styles for them are lacking, the problem is NIH (Not Invented Here or NIBM, by me?).

DW:     Was Literate Programming (web or otherwise) an effective tool with your students?

BC:     It was effective as a pedagogical tool. See my paper at the 16th TUG meeting. I doubt that it would be effective for someone new to the process and the tools without a significant time for preparation.

The data in that paper came from teaching CS/1 honors classes using Pascal/WEB and our web-mode. About half the class was non-majors, but still honors students. We were surprised at first when we realized our majors were outperformed by the non-majors. We concluded that the more background the students had the more they wanted/expected an easy A. We have heard of other anecdotal evidence from other instances where the first course was one used by many students in secondary school.

DW:     Thank you for taking the time to participate in this interview. I look forward to meeting you in person this coming summer at the TUG2010 conference in San Francisco.

BC:     Thank you! I too look forward to meeting you. Thank you for your service to TUG and the computing professions.

Web site regeneration of May 24, 2010; TUG home page; search; contact webmaster.