Seibel Peter. Coders at Work
Подождите немного. Документ загружается.
Dan Ingalls
395
actually do use a code repository now, and there are other guys on the
team who are doing more application stuff and a little bit in the kernel. And
I see that it’s nice having a shared repository that works—that’s great. The
next step for the Lively Kernel is to integrate it with the repository—you
can change stuff in the Lively Kernel but it’s only that running version; it
doesn’t get pushed out to the repository and become a part of the future.
That’s the next step we need to do.
Seibel: Have you ever done pair programming?
Ingalls: I’m trying to think of examples. Usually I’ve been on my own or
working on a separate part. There are lots of projects on which I’ve worked
with others, and many intense pair debugging sessions.
Seibel: Are there any techniques for managing that kind of collaboration?
When everybody goes off to each do their own work, there’s always the
potential of things not quite fitting together.
Ingalls: Either you agree on some interface, or often I’ll build some
framework that’s incomplete but that works for one example and then it’s
obvious to someone else where their stuff would fit in. Or they’ve done
that and it’s obvious to me where my stuff fits in. It’s usually been in that
concrete way, rather than any kind of spec because we’ve usually been
working on something that nobody wrote down. It’s all kind of been
following the needs of the moment.
Seibel: You’ve worked at a lot of levels from down in the bits of BitBlt and
writing microcode up to fairly high-level stuff in Smalltalk. How much do
programmers need to know about the various levels of software and
hardware they’re dealing with?
Ingalls: That’s a good question. To think outside the box, you’ve got to be
a little outside the box. If there’s something to be taken advantage of that
you don’t see in your normal exposure to a language, then you’ve got to
have some intuition outside and some understanding of that and the ability
to work in a system that has control over that.
In the context of language design, maybe you’re going to want to work with
the processors that are out there so maybe you don’t need to know a lot
Download at Boykma.Com
Dan Ingalls
396
about them, other than, for decent performance, how caching works and
that kind of stuff. I think you have to stand back and say, “What boundaries
is this going to cross?”
Seibel: Leaving aside how much you need to know eventually, when it
comes to learning to program, some people say you should start with a
high-level language and learn certain universal concepts. Other people say
you need to start with assembly and work your way up so you understand
what’s going on. Do you buy that?
Ingalls: No, I don’t think I necessarily buy that. That’s how I learned it and I
was fascinated with it. And I think there will always be people who are
fascinated with this level or that. But I don’t think there’s any one way, just
like there’s any one way to do art.
I think there are other things that are equally exciting and probably more
appropriate nowadays still to be explored. Lord, we were thinking of doing
artificial intelligence a quarter of a century ago. The machines are
immeasurably faster and we’re doing almost nothing in that space—we’re
still doing very close to Fortran. Prolog has been out there for a long time;
there’s all sorts of stuff that can be done with logic programs. If you ought
to learn about assembly and find out how that works, you ought to be doing
some immersion in things that are more outside the box and that are really
a possible piece of the future.
So I’m not saying, “not assembly language.” I’m saying, you should learn
some of these other powerful techniques so when you think about how to
go forward, you can take advantage of them. In terms of a place to start,
immediate gratification has always worked for me. When I want to teach
somebody Smalltalk, I usually start with a little dialog: “What interests you
most? Are you interested in how to play with text or things you can do with
numbers or things you can do with music, or things you can do with
graphics?” And start with any one of those.
There’s all kinds of fun stuff you can do taking text apart and putting it
together. There’s all kinds of fun stuff you can do with numbers and
different bases and floating-point and fixed. And the same thing with music:
you can start with notes and put them together in melodies and chords.
And with graphics: superpositions and rotations. Any one of those is a rich
Download at Boykma.Com
Dan Ingalls
397
space to explore. I really think different people think differently that way.
Similarly, if you’re going to teach somebody to program a computer, maybe
you work with expression evaluation, maybe you work with logic
programming. Maybe you do some stuff in user interface. People will light up
in one area and that’s where they should go deep.
Seibel: As I understand it, the original purpose of Smalltalk was to teach a
kind of programming literacy. Is that something everyone should have, just
the way everyone is expected to be able to read and write and do a little bit
of math? Should everyone have some ability to program a computer just
because it’s a useful way of thinking?
Ingalls: It’s very hard for me to say that anyone should do anything, because
I’ve met people that I think are better than me in this way or that way that
know nothing about programming. In terms of literacy, the stuff that’s under
there is logic and math and yes, people should be able to think logically. But
I’d never say that somebody should know how to program, I don’t think.
There’s stuff we do in everyday life that’s like programming. You need to
know about procedures with steps, that kind of thing.
Computers incorporate some powerful ideas and can bring some powerful
ideas to life. The wonderful thing about computers is they bring
mathematics to life. So they can be a great tool that way. Now, my feeling
about the powerful ideas that are necessary to lead a good life, it’s not clear
how many of them are in this space.
Seibel: Seymour Papert wrote in Mindstorms about debugging as an
important element of an intellectual toolkit—the idea that the name of the
game is not to get the right answer but to get an answer and then debug it.
Ingalls: Oh, absolutely! People should learn to think clearly and to
question. And to me it’s very basic. If you grow up in a family where when
the cupboard door doesn’t close right, somebody opens it up and looks at
the hinge and sees that a screw is loose and therefore it’s hanging this way
vs. if they say, “Oh, the door doesn’t work right; call somebody”—there’s a
difference there. To me you don’t need any involvement with computers to
have that experience of what you see isn’t right, what do you do? Inquire.
Look. And then if you see the problem, how do you fix it? To me that’s so
basic and human and comes so much from parent to child.
Download at Boykma.Com
Dan Ingalls
398
Computers are certainly a medium for doing that. But they’re just
computers. There’s a lot of that that will transfer, but to me it’s really big
and basic and human, so it’s not like we’re going to enlighten the world just
by teaching them computers.
Seibel: Do you remember the first interesting program you wrote?
Ingalls: Oh, let’s see. In each programming experience, there was
something that was out of the box. When I discovered VisiCalc, I wrote a
spreadsheet to translate English to pig Latin in VisiCalc. That was, to me,
interesting, because it used the spreadsheet metaphor as a parallel
programming approach. Parsing text that way was fun and enlightening.
Seibel: So VisiCalc had primitives for taking apart strings?
Ingalls: Yeah, you could take apart strings. Maybe what I actually had was
Lotus 1-2-3, not VisiCalc, because I’m not sure that VisiCalc had string
primitives. I got one of those little Poqet PCs—it was the first truly
handheld PC. It ran on two penlights, and I put 1-2-3 on it, and I had a plane
flight across the country and I thought, “What can I do in this time?”
Seibel: That must’ve been well after you learned to program, because
there obviously weren’t Poqet PCs when you were starting out.
Ingalls: That was later on. The really interesting thing that I did in Fortran
was I got a hold of Val Schorre’s paper on META II—just a wonderful, really
simple compiler compiler and wrote an implementation of that in Fortran.
So all of a sudden this meant you could have other languages in a Fortran-
only environment. That was the most interesting thing I did in Fortran,
because it used Fortran to escape from the world of Fortran.
Seibel: That seems to be a bit of a theme: the pig-Latin spreadsheet, then
this, and a hash table in COBOL for your profiler—is there a part of you
that just likes cutting against the grain?
Ingalls: I don’t feel like it’s against the grain, but whenever I have a
computing environment to play with, I like to try new things in it. That’s
what’s been so much fun about doing the Smalltalk systems I did. You start
from pretty much scratch and your job is to figure out what to assemble
Download at Boykma.Com
Dan Ingalls
399
and get working first that will help you to do the next step and build stuff up
from there.
In these cases, it was a matter of getting outside the box. That’s a way of
asserting that you’ve mastered the thing—if you can do some things that
you wouldn’t have thought you could do in it.
Seibel: Can you identify any big changes in the way you think about
programming now?
Ingalls: That’s a good question. One thing is we’ve got lots of computer
cycles to spend. So I’m comfortable now, as the pejorative saying goes,
pissing away cycles to get something done cleanly. But the basic thing hasn’t
changed for me at all, which is trying to get clear about what the kernel or
kernels are that I have to work with and the goal that I’m trying to achieve.
It’s changed a little bit in that I’m no longer the quantum mechanic in the
group that I work in. I’m more at a higher level—that just means that I’m
spending a little bit more time on goals and politics than on code. It’s got
more to do with creating a context. Until this later phase of my life, I’ve
been lucky to be in an existing context where I didn’t have to create it. But
every once in a while I still get down and write serious code.
Seibel: I was looking at one of your papers from the ’70s about your
Fortran profiler. In the preamble you were very enthusiastic about how that
tool changed your programming from figuring out what you were going to
write, writing it, and debugging it, to figuring out what you were going to
write, writing a really simple version, profiling it, then optimizing it. Do you
still work that way?
Ingalls: I definitely go for effects first, bits on the screen or whatever it is,
because that’s so motivating, and you often learn something new about what
you had wanted to do just by getting a chance to glimpse where it’s going.
Then if profiling is what’s needed, you profile. Or it may turn out you
weren’t doing the right thing—that wasn’t quite what you wanted to do, so
you changed the goals, or you change how you’re doing it. But when it’s a
matter of performance, I still operate the same way. We did a really nice
profiler in Smalltalk, and then Squeak, that gave you good feedback that way.
Download at Boykma.Com
Dan Ingalls
400
Part of that is about performance but part of it is about just structure and
architecture. You may find that some things are hardly ever used and maybe
you could just get rid of them by doing things differently. It’s just a different
perspective on the thing.
Seibel: It seems that virtually every programmer has a copy of Knuth’s The
Art of Computer Programming. Some people have it and it’s just on their shelf.
Some people have it and use it as a reference. And some people have it and
really read it cover-to-cover. You actually studied with Knuth at Stanford;
how much of it have you read?
Ingalls: I loved working with Don and I taught his MIX course for one
quarter at Stanford, too, which was also enlightening. I think I’m pretty
different from Don, but what I like about him is he’s got a great
mathematical mind but also loves to go deep into the bits—the pragmatic
parts of things. I, too, love to get into the pragmatic side of things but I’m
not as rigorous as Don.
I was trained in physics and to me the problems I work on, or how I work
on them, is really much more a physical thing. When I’m talking about other
perspectives on programs, I’m really thinking that there’s a thing there that
you can touch and feel it vibrate.
If you take the way he worked on TeX, it was very mathematical and
beautiful and elegant. Compare that with, say, the first Smalltalk engines:
they were very ad hoc. I just put together what was needed. It may be that
after a few go-rounds, I started to get some sort of mathematical picture of
it, or we did, but it’s different in that regard.
So the truth is, I read a fair amount of his basic data structures stuff, but I’m
not a big reader—I’m very much of a doer. If I have a flaw, it’s that I will
often do my own of X, Y, or Z rather than reading the literature and
knowing about it. It’s usually worked to my advantage I think, but who
knows?
Seibel: How much math do you think programmers need to know?
Dijkstra claimed that computer science is just a branch of mathematics. And
understanding The Art of Computer Programming requires a fair bit of math.
Download at Boykma.Com
Dan Ingalls
401
Ingalls: You’ve got to have a logical mind. But I spent a lot of time in the
country in Virginia while learning about computers. I always thought that if I
wanted to start a computer company in the mountains of Virginia, I’d go find
the mechanics. Except in certain fairly esoteric parts, the math isn’t near as
important as logic and intuition.
I think of a lot of it as being more architectural: the way the graphics work
together with the models; the way things need to be updated or cached.
That’s not esoteric mathematics. That said, I do view it as being very much a
part of the field of mathematics. To me, the really exciting thing about
computers is they allowed mathematics to become a synthetic, not just an
analytic art. What I have so much fun doing day to day is a mathematical
kind of thing, but it’s creative, generative, synthetic.
Seibel: You say you’re not a big reader. Are there any books you’d
recommend?
Ingalls: No. And I’m sure I’m in the minority this way. I was not a big
reader as a kid. I would occasionally read things thoroughly and not be
interrupted until I was done. There are some papers, absolutely, and I guess
with them come some books. Val Schorre’s paper on META II is one of
them. There is the LISP 1.5 book. There’s APL, but I didn’t find Iverson’s
book to be at all a good way to learn it. A mathematician probably would
have. I can’t even remember what I read to learn about it from. But I loved
it. So I think spending a bit of time in that language would be like a book to
read. Likewise, Smalltalk.
Seibel: So do you still enjoy programming as much as you used to when
you started?
Ingalls: Yeah, the programming itself. The last couple of years have been
interesting because I’ve been taken away from the environment that I got so
used to—Smalltalk and then Squeak—where the tools are just really great.
I’ve had to step back a little bit, working with JavaScript in browsers and
conventional development environments. It sometimes takes me longer to
debug stuff than it used to, but the basic process of getting an idea and
making it happen I still really enjoy.
Seibel: Do you feel at all that programming is a young person’s game?
Download at Boykma.Com
Dan Ingalls
402
Ingalls: Not really, no. There’s something about being able to survey all the
stuff that’s going on, and some of that endless energy that I don’t feel I have
the way I used to. But I still love to just take a problem and sit down and
pore over it until it’s right. There’s an analogy here: I tried to learn to play
piano fairly late in life. People said, “Oh, you should learn when you’re
young. You learn so much quicker.” Although I didn’t go very far, my
conclusion was that it isn’t that young people learn that much faster; it’s just
they have more time. When I would put time in, I made progress.
I feel a bit the same thing with programming. When I look back on earlier
times in my life, I had all the time I wanted. I would just work and work.
Now there are other things going on in my life and I’ve got responsibilities
that aren’t just programming. That undermines a bit of that intense focus.
Seibel: Leaving aside the amount of time you get to spend on it, doesn’t
programming require a certain intensity and focus? People talk about flow
and how if you get interrupted every 15 minutes, you never get anything
done because it takes that long to even start getting the stuff in your brain
sorted out.
Ingalls: That reminds me of something I said to somebody way back at
PARC. I was starting to have some other responsibilities besides just
hacking Smalltalk, but we were also making a lot of progress in making it a
really productive system. I kidded that I was racing to improve the Smalltalk
environment so that I could continue to get reasonable work done in
increasingly short amounts of time. So I’d gotten to where in 15 minutes I
could sit down and actually do something useful.
The other part of it is that you work with other people. I work with
younger people and it’s great—maybe I spend a bit more time thinking at
the level of goals and politics and arranging to make things happen, and
they’re making up for what I lack in available time for that deep, deep dive.
Seibel: You have a reputation of being a great guy to work for, when
you’re leading a project. How do you lead a team that makes people
productive and happy?
Ingalls: I love what I do. And it’s fun to share with other people. These
activities can go arbitrarily deep or wide, so it’s easy to find things for
Download at Boykma.Com
Dan Ingalls
403
various people to work on. I’ve just always enjoyed doing that with other
people. There have been times when it’s worked better or not—there are
very different phases to this kind of work. There are times when you see
everything that needs to be done and it’s just a matter of getting people to
do it. And there are other times when you don’t really know what needs to
be done and you’re trying to find out. Those are really different phases.
Seibel: Do you have any tips on how to be a good technical leader?
Ingalls: The first thing is being clear about what you’re trying to do. The
job is to get a clear vision. If you’ve been around, you can see actually how
you’re going to implement that, so you can start to see how that would
work out in terms of what various people can do and how it would all fit
together.
There have been times when I’ve been working on a project where I could
see everything. It felt really powerful because whenever anybody was stuck I
could tell them immediately what the next step was, or how to get around
it. And people feel that, too, if you know where you’re headed. They can
immediately sense it’s all there—he’s got it. And that’s empowering for the
team as well.
Seibel: Does having too clear a picture of what you want ever disempower
people—since you’ve got everything in your head, there’s nothing fun for
them to do?
Ingalls: Well, you can leave it open how they do their part of it and you
maybe only step in with micromanagement where it’s needed. And often
things turn out better. I was lucky to work with a great crew of people for a
long time whom I trusted. Trust is part of it, trust for the people that you’re
working with. The other thing is just confidence. When the picture’s clear,
it’s easy to be confident about it. I think the kind of thing that makes for bad
micromanagement is you’re worried and you’re insecure, and so you’re
feeling like you have to nail everything down.
Seibel: Have you ever had a really great team leader when you were a
worker bee?
Download at Boykma.Com
Dan Ingalls
404
Ingalls: The best boss in my life has been Alan Kay. I worked under him at
Xerox for some very formative years, and that was an interesting
combination, because he knew what he wanted but he hardly ever said how
I should do things. But he was technically savvy about all of it so he was a
good critic. I and the guys who came to work with me were really
productive so I think he felt enough progress that he didn’t have to interfere
much. He made an umbrella for me, for us, and he really had a picture of
what he wanted to do.
Seibel: When folks are working in a group, is it better for coders to each
own a piece of a system? Where “This is my code and no one touches it”
vs. the team owns the code and anybody can touch anything.
Ingalls: I don’t know. The way we work now on the Lively Kernel project
is different people have different areas they work on, but there aren’t any
fences. It’s more a matter of expertise, or focus, or goals. I’m trying to think
of these periods that seemed really successful, how that worked. I’ve never
worked in big teams, so it’s generally been that people pretty much worked
solely on a piece of code.
Seibel: Another topic, debugging: what’s the worst bug you’ve ever had to
track down?
Ingalls: It was a garbage-collection bug. Garbage collection is the worst
thing because the manifestation of the problem is long after the cause has
come and gone. Being able to track down an obscure bug like that makes
me think of breaking codes. My father was in the Office of Strategic
Services, and they worked in teams, and a lot of what they did was just
gathering information, just trying to be up on things. Then a code would
come in with a fragment of something they had seen in a newspaper and
they would put it together that way.
It was the same thing with being able to track this down. What I brought to
it was just having all this intuition of what can cause what to happen in these
situations. This particular one, I was in it deep down for at least a day.
When I finally got it done, I was elated, and my son, who was, I think, four
at the time, made me a “Determined Debugger Award.”
Download at Boykma.Com