Seibel Peter. Coders at Work
Подождите немного. Документ загружается.
C H A P T E R
9
Guy Steele
Guy Steele is a true programming polyglot. When I asked him what languages he
has used seriously he came up with this list: COBOL, Fortran, IBM 1130 assembly,
PDP-10 machine language, APL, C, C++, Bliss, GNAL, Common Lisp, Scheme,
Maclisp, S-1 Lisp, *Lisp, C*, Java, JavaScript, Tcl, Haskell, FOCAL, BASIC, TECO,
and TeX. “Those would be the main ones, I guess,” he added.
He had a hand in the creation of both of the major surviving general-purpose Lisp
dialects: Common Lisp and Scheme. He served on the standards bodies that
defined Common Lisp, Fortran, C, ECMAScript, and Scheme and was recruited by
Bill Joy to help write the official language specification for Java. He is now at work
designing Fortress, a new language for high-performance scientific computing.
Steele’s academic career included an AB from Harvard and an SM and PhD from
MIT. While at MIT he collaborated with Gerald Sussman on a series of papers
now known as “The Lambda Papers,” which included the original definition of the
Scheme programming language. He has also been a chronicler of hacker culture
as one of the original compilers of the Jargon File and editor of the book version,
The Hacker’s Dictionary (subsequently updated and expanded by Eric S. Raymond
as The New Hacker’s Dictionary). And he played an important role in the birth of
Emacs and was one of the first programmers to port Donald Knuth’s program
TeX.
Download at Boykma.Com
Guy Steele
326
Steele is a Fellow of the Association for Computing Machinery and the American
Academy of Arts and Sciences and a member of the U.S. National Academy of
Engineering. He won the ACM’s Grace Murray Hopper Award in 1988 and Dr.
Dobb’s Excellence in Programming Award in 2005.
In this interview he talks about designing software and the relation between writing
and programming, and he gives one of the best explanations I’ve ever heard of the
value—and limitations—of formal proofs of correctness.
Seibel: How did you get involved in programming?
Steele: Well, when I was in elementary school I remember being fascinated by
science and math and I read books such as Irving Adler’s House of Numbers; it
was one of my favorites. And I liked kiddie science fiction like, say, the Danny
Dunn series and that kind of thing. So I had this general interest in science and
math. In reading everything I could about science and math, I read a little bit
about these newfangled computers that were coming up as well.
Seibel: And when would that have been?
Steele: I was in elementary school from 1960 through ’66. But I think the
real turning point was when I got to Boston Latin School—it would have
been in the equivalent of the ninth grade. A friend asked me, “Have you
heard about the new computer in the basement?” I thought this was the
newest story after the one about the fourth-floor swimming pool and the
school only has three stories. But he said, “No really, it exists.”
It turns out that T. Vincent Learson had arranged for an IBM 1130
minicomputer to be in the basement of the Boston Latin School. He was an
alum and a very generous one apparently. My friend proceeded to show me
a Fortran program of about five lines and I was immediately fascinated.
Then I went to one of our math teachers and asked if I could have some
books to study. He gave me some books and thought they’d keep me busy
for a month but they actually lasted a weekend. I taught myself Fortran over
Thanksgiving weekend—so it was a long weekend—of 1968. After that I
was completely hooked.
Download at Boykma.Com
Guy Steele
327
My Latin School friends and I had a fascination with IBM because of the IBM
1130 and we took to visiting the IBM office downtown every couple of
months and talking with the people there and occasionally ordering
publications with what little money we had.
There was also a bookstore downtown that had books about exotic
languages like PL/I, and we’d occasionally buy books there. So, through Latin
School we got to know IBM equipment. We just had the 1130 but we
drooled over System 360. We read about it and didn’t really have access to
one.
Then I became involved at MIT in the spring of 1969 in the High School
Studies Program. This was great—go on Saturday mornings and have college
students teach you all this cool stuff. I took courses in group theory and
computer programming and I forget what all else. I became rather heavily
involved and therefore got to feel on very familiar turf at MIT. Through the
High School Studies Program, we had access to both IBM 1130s and DEC
PDP-10s. So that’s how we got to know the Digital Equipment line.
As high-school students we became aware of the DEC office in Central
Square that tended to cater to the MIT students. They didn’t blink when
high-school students walked in and asked for reference manuals. It was
great. When I was a junior or senior at Latin School, a friend of mine and I
typed up a proposal to DEC to implement an APL for the PDP-8. And they
took the proposal seriously. They looked at it for a week and then they
said, “Well, we don’t think this is a good idea, but thanks for the offer.”
Seibel: What was the first interesting program you wrote?
Steele: Well, I learned Fortran first so I think things really got interesting
when I started to learn IBM 1130 assembly language. And I guess the earliest
interesting program I can remember was something that would generate
keyword-in-context indexes. IBM provided these so-called quick indexes for
their manuals that, given a keyword, you could look it up in the index and it
was alphabetized according to keyword but on either side of the keyword
you would see several other words of context surrounding that word.
Seibel: Context from where the word appeared?
Download at Boykma.Com
Guy Steele
328
Steele: Where the word had appeared in the original publication. So down
the middle column were the words that had been alphabetized and sticking
out on either side would be chunks of context. So I thought I’d tackle that
problem, doing it on the 1130. And considering that the 1130 had only
4,000 words of memory, it was clear that I was going to have to keep the
records on disk. So I took this opportunity to learn about efficient out-of-
core sorting techniques. What was interesting about this program was not
so much the generation of the KWIC index but implementing a multiway
out-of-core merge sort. And it was reasonably effective. Unfortunately my
in-core sort was bubble sort because I wasn’t that sophisticated. I should
have also done a merge sort in core but I hadn’t quite figured that out at the
time.
Seibel: How long do you think it was from when you first realized there
really was a computer in the basement to when you were writing this
program? Was it months? Weeks?
Steele: It would have been within the first two years. I don’t remember
whether it was in the first year. I learned Fortran in the fall of ’68. And I
remember that APL was my third language so I must have learned the
assembly language somewhere around Christmastime or shortly after. I
know I learned APL in the spring of ’69 because that’s when the Spring Joint
Computer Conference came to Boston.
IBM had an exhibit on the show floor touting all manner of IBM products
but the APL 360 in particular, and I hung around that booth. At the end of
the trade show they were about to throw the stack of paper away that had
been used for demos on the Selectric terminal. At just the right moment I
walked up and asked, “Are you going to throw that away?” And the lady
looked at me, puzzled, and said, “Here, it’s yours,” as if she was giving me
this big Christmas present. Which she was.
Seibel: What was on that paper?
Steele: It was fan-fold paper from a Selectric terminal on which they had
been demonstrating APL for the past two days. Just little programming
examples of whatever they had happened to type in. And from that and the
brochure they had on the trade-show floor, I taught myself APL.
Download at Boykma.Com
Guy Steele
329
Seibel: So you were very comfortable at MIT but you ended up going to
Harvard and working at MIT. What happened?
Steele: By the time I was applying to colleges I applied to MIT, Harvard,
and Princeton and really wanted to go to MIT. I got accepted at all three.
The headmaster of Boston Latin School was Wilfred L. O’Leary, an old-
school classicist, a wonderful gentleman. He called up my parents and said,
“Do you realize that your son is actually considering going to Tech when he
has an acceptance at Harvard?!” So he twisted their arms and they twisted
my arm and I decided to go to Harvard after all.
Then my parents were on my case to get a summer job and not just sit
around the house—you know, the classic syndrome. I knew I was interested
in computing and didn’t want to flip burgers. So I interviewed for
keypunching jobs, figuring that was something I’d be reasonably qualified to
do. But nobody wanted to hire me, in part because I wasn’t 18 yet. I didn’t
figure that out until later. They just listened to my story and said, “Don’t call
us, we’ll call you.”
Then around the beginning of July I heard that Bill Martin at MIT was looking
for Lisp programmers. I thought, “Aha, I know Lisp.” I’d hung around MIT so
much and had obtained copies of Lisp documentation from the Artificial
Intelligence Lab and I would sneak into the labs and play with the
computers. The doors were open in those days—the Vietnam protests had
not yet happened, which is what caused them to put locks on the doors.
And I had spent my senior year implementing my own Lisp for the IBM
1130.
So I showed up at Bill Martin’s office, this skinny kid out of nowhere, and
poked my head in and said, “I hear you’re looking for Lisp programmers.”
And he didn’t laugh at me. He just looked at me and said, “Well, you have
to take my Lisp quiz.” “OK. How about now?” So I sat down and spent two
hours working on a list of questions and puzzles. When I was done I gave
him the papers and he spent ten minutes looking them over and said,
“You’re hired.”
Seibel: Was Lisp one of the things you had actually studied in this High
School Studies Program?
Download at Boykma.Com
Guy Steele
330
Steele: A little bit, though it was more Fortran and some other things.
Seibel: Did you have any important mentors when you were starting out?
Steele: At Latin School I’d primarily have to credit the math teachers with
encouraging me just the right amount. In the ninth grade Ralph Wellings,
who lent me those books over the Thanksgiving weekend, struck a deal
with me. He said, “I notice you’ve been getting 100 percent on all your math
quizzes.” He said, “I’ll let you spend four math classes a week in the
computer room if on the fifth day you take the quiz and get 100 percent. If
you ever get less than 100 percent then the deal is over.” So that was
incentive. I proceeded to ace quizzes for the rest of the year—I studied
math especially hard and that gave me access to the computer. Even better,
the next year my math teacher would not offer the same deal, which was
appropriate because I did not know the math for that year. So they judged it
about right. So I had good teachers that gave me what I needed to learn all
kinds of things.
Seibel: And then, as you got more involved in computers, were there
particular folks who helped you along the way?
Steele: Well, certainly Bill Martin, who hired me. And Joel Moses, who was
in charge of the Macsyma project into which I was hired at MIT.
Seibel: And you ended up working on that project throughout college?
Steele: Yes, I was an employee of MIT all the time I was at Harvard. It was
a full-time job in the summers and it became a afternoon job during the
school year. I’d do my best to arrange my classes to be in the mornings at
Harvard, then I could take the T down to MIT and get in two or three
hours of programming before heading home.
Seibel: And that was all working on Macsyma in Lisp?
Steele: Yeah. My specific job was to be the maintainer of the Maclisp
interpreter. JonL White had been in charge of both the interpreter and the
compiler and he became pretty much the compiler guru, and I took care of
the interpreter, and it was a pretty good split. So JonL White was a mentor
of mine. All the people on the Macsyma project kind of took me under their
Download at Boykma.Com
Guy Steele
331
wing. I also got to know some of the people in the AI Lab. So by the time I
applied to MIT for graduate school it was pretty easy to get accepted
because they already knew me and what I was doing.
Seibel: Did you get your undergrad degree in computer science?
Steele: Yeah. I set out to be a pure math major and arranged my courses
appropriately and then discovered that I had no intuition whatsoever for
infinite dimensional Banach spaces. That’s what did me in. Fortunately, just
out of interest, I had taken enough computer courses on the side that I was
well-positioned to make the switch in major. To be precise, what I switched
to was an applied math major. Computer science was part of applied math,
and applied math was part of the engineering department at Harvard.
Seibel: At Harvard what kinds of machines were you dealing with?
Steele: DEC PDP-10s. There was a PDP-10 on campus, but I think that was
mostly used for the graduate work. Undergraduates had access to teletype
terminals to a commercial system that Harvard was renting or leasing or
something.
Seibel: Is there anything you would do differently about how you went
about learning to program? Is there anything you wish you had done earlier?
Steele: It’s not as if I set out with a particular goal in mind. I have no
regrets about the particular path I took. Looking back I think I was the
fortunate beneficiary of a number of interesting coincidences or blessings.
This experience of being, in effect, both at MIT and Harvard at the same
time I now realize was a very unusual experience. I could run back and forth
and say, “The professor at the other end of the river says this.” And this
one will say, “Oh, he’s full of it; here’s the way you should think about it.”
That gave me a very broad education, very quickly.
Having access to MIT as a high schooler was another relatively unusual
thing. And to be allowed to play with million-dollar computers when I was
15, back when a million dollars was real money. So no, I certainly don’t have
any complaints or regrets or wishes that I had done anything differently. I
also tend to be a laid-back kind of guy and to take things as they come.
Download at Boykma.Com
Guy Steele
332
Seibel: What has changed the most in the way you think about
programming now, vs. then? Other than learning that bubble sort is not the
greatest sorting technique.
Steele: I guess to me the biggest change is that nowadays you can’t possibly
know everything that’s going on in the computer. There are things that are
absolutely out of your control because it’s impossible to know everything
about all the software. Back in the ’70s a computer had only 4,000 words of
memory. It was possible to do a core dump and inspect every word to see if
it was what you expected. It was reasonable to read the source listings of
the operating system and see how that worked. And I did that—I studied
the disk routines and the card-reader routines and wrote variants of my
own. I felt as if I understood how the entire IBM 1130 worked. Or at least
as much as I cared to know. You just can’t do that anymore.
Seibel: Were there books that were important to you when you were
learning to program?
Steele: In the ’70s, absolutely: Knuth, The Art of Computer Programming.
Seibel: Did you read those cover-to-cover?
Steele: Pretty close to cover-to-cover, yes. I worked as many exercises as I
felt I was capable of tackling. Some called for higher math or other things I
didn’t understand, and I’d sort of gloss or skip over those. But the first two
volumes and much of the third I read pretty carefully. The Aho, Hopcroft,
and Ullman algorithms book—that’s where I learned how to do sorting for
real, I think. I’d have to step across to my library to try to remember other
ones. I’m a pack rat—I’ve saved all these books. But those are the ones that
I would cite off the top of my head.And books about Lisp. The Triple-I Lisp
book edited by Berkeley and Bobrow: kind of a scatter-shot collection of
papers, but I learned a lot of interesting stuff from that. And then I started
reading SIGPLAN Notices and Communications of the ACM. Back in those days
CACM had real technical content and was well worth reading.
I should mention two things. First, I did science fairs when I was at Latin
School and I made a point of doing projects about computer science. One of
the judges one year said, “Have you considered becoming a student
Download at Boykma.Com
Guy Steele
333
member of ACM?” I don’t know his name. But I have been very thankful
ever since. That was a good thing for me.
And when I got to Harvard, if I had a spare hour to kill in the morning I
would go over to Lamont Library and I would do one of two things: I would
read my way backwards through Scientific American and I would read my way
forward, from the beginning, in Communications of the ACM. So I was, in
particular, trying to pick up all of Martin Gardner’s columns on
mathematical games. And I just read whatever interested me out of CACM.
In 1972 there was only about 15 years of that journal, so it didn’t take that
long to plow through them all.
Seibel: It also must have been easier then than it would today in the sense
that the same way you could understand whole systems, one person could
hope to understand the whole field.
Steele: Yeah, you could hope to understand the whole field. There were
lots of one-page articles. You know: “Here’s a clever new hashing
technique.” I read a lot.
Seibel: I often find older papers are hard to get into since they’re tied to
the particulars of old hardware or languages.
Steele: Well, necessity is the mother of invention—an idea arises because
it’s needed in a particular context. Then a little later it’s recognized that that
idea is the important thing. And then you need to strip away the context so
the idea can be seen and that takes some years. “Here’s a clever technique
for reversing the bits of a word,” and they give something in 7090 assembly
language. And there’s an interesting mathematical idea there but they
haven’t quite abstracted yet.
Seibel: I guess that’s Knuth’s job, right?
Steele: Knuth and people like him, absolutely.
Seibel: Presumably people who study computer science in school get
guided through all that stuff. But there are also a lot of programmers who
came into it without formal training, learning on the job. Do you have any
advice for how to tackle that problem? Where do you start and how do you
Download at Boykma.Com
Guy Steele
334
get to the point where you can actually read these technical papers and
understand them? Should you start at the beginning of the ACM and try to
get up to the present?
Steele: Well, first of all, let me say that that exercise of reading through
CACM from early on wasn’t my plan to become a great computer scientist
by reading everything there was in the literature. I read it because I was
interested in stuff and felt internally motivated to tackle that particular set
of material. So I guess there are two things: one is having the internal
motivation to want to read this stuff because you’re interested or because
you think it will improve your skills.
Then there is the problem of how do you find the good stuff? And of course
the view of what is the good stuff changes from decade to decade. Stuff that
was considered the really good stuff this year may be kind of dated in ten
years. But I guess you go to a mentor who’s been through it and say, what
do you think was the good stuff? For me the good stuff was Knuth; Aho,
Hopcroft, and Ullman. Gerald Weinberg on The Psychology of Computer
Programming, which I think is still very readable today. Fred Brooks’s
Mythical Man-Month gave me some insights.
In those days I haunted the computer-science book section of the MIT
bookstore and just made a point of going through there once a month and
browsing through the bookshelves. Of course now you walk into a
bookstore and there’s a computer section that’s ten times as big, but most
of it is about how to do C or Java this year. But there will be a smaller
section of books about the theoretical background, algorithms, that kind of
thing.
Seibel: There’s another kind of reading, which I know you think is
important—reading code. How do you find your way into a big pile of code
you didn’t write?
Steele: If it’s a piece of software that I know how to use and just don’t
know how the insides work, I will often pick a particular command or
interaction and trace it through.
Seibel: The execution path?
Download at Boykma.Com