[As described below, this is Part B of a 3-part course. Participants should complete Part A first -- Part B "dives right in" and refers often to material from Part A.] This course is an introduction to the basic concepts of programming languages, with a strong emphasis on functional programming. The course uses the languages ML, Racket, and Ruby as vehicles for teaching the concepts, but the real intent is to teach enough about how any language “fits together” to make you more effective programming in any language -- and in learning new ones. This course is neither particularly theoretical nor just about programming specifics -- it will give you a framework for understanding how to use language constructs effectively and how to design correct and elegant programs. By using different languages, you will learn to think more deeply than in terms of the particular syntax of one language. The emphasis on functional programming is essential for learning how to write robust, reusable, composable, and elegant programs. Indeed, many of the most important ideas in modern languages have their roots in functional programming. Get ready to learn a fresh and beautiful way to look at software and how to have fun building it. The course assumes some prior experience with programming, as described in more detail in the first module of Part A. Part B assumes successful completion of Part A. The course is divided into three Coursera courses: Part A, Part B, and Part C. As explained in more detail in the first module of Part A, the overall course is a substantial amount of challenging material, so the three-part format provides two intermediate milestones and opportunities for a pause before continuing. The three parts are designed to be completed in order and set up to motivate you to continue through to the end of Part C. Week 1 of Part A has a more detailed list of topics for all three parts of the course, but it is expected that most course participants will not (yet!) know what all these topics mean.
Weak Typing
Loading...
来自 University of Washington 的课程
Programming Languages, Part B
429 个评分
从本节课中
Section 7 Including a Quiz
In the last module of Part B we will use our experience programming in ML and Racket to
compare and contrast static typing and dynamic typing. This is not only the most important difference between these two languages, but it is a fundamental topic in the study of programming languages. Learning it can help you program more effectively in both kinds of languages. After completing this week's quiz, don't forget to watch the Part B Wrap-Up and Part C Preview video.
与讲师见面
Dan GrossmanProfessor
Computer Science & Engineering
[music] In this segment I want to briefly talk about weak typing, which is actually
a separate topic from static versus dynamic typing, but because it's often
confused with that issue, I thought it would be good to put it here in the course
and sort of set the record straight. So weak typing, which is probably best
known as the sort of type system that C and C++ have, is what I'll describe as
follows. There exist programs that according to the
language definition, must pass static checking.
But then when the program is run, they're allowed to do absoultely anything.
I like to say that they're allowed to set the computer on fire.
By which I mean they're allowed to crash, they're allowed to corrupt your data.
They're allowed to be a virus, they're allowed to delete all your files and so
on. So the idea is, there are programs that
perform operations that don't make any sense.
But the dynamic checking that would catch those errors before they do anything
disastrous is optional, and in practice and implementation of C and C++ is not
done. Now, if you've only ever programmed in
Racket, or Python, or ML, or Java, this would seem crazy.
Why would you allow a language implementation not to do these checks?
And the answers are numerous. So it turns out there are good reasons to
use C or C++ in some settings. One is that it make the language easier to
implement for all sorts of different computing platforms.
The checks are left to the programmer. Now I would prefer to have a harder
language implementation and make programming easier, but that's not always
the trade-off that people make. Probably the biggest one that people
emphasize is performance, that if you don't have to perform dynamic checks That
not only do you save the time of not doing them, but you don't have to use space in
your program to store the various tags and sizes and other things that you need in
order to do the dynamic checks. And that's actually related to this third
reason which is just lower level That if you want to language for a programmers can
control things like the representation of data, then you can type dynamic checks
that requires extra fields that the programmer doesn't have control over.
But that's really about C and C++ which is pa course is really not about.
I just want to emphasize the idea of weak typing and to tell you that its not a very
good name, right it doesn't really have to do.
With type systems. It has to be with there being things that
you're trying to prevent, some bad property X, and not checking for it.
You're not checking for it statically and you're not checking for it dynamically and
you're saying, if it happens, the computer is allowed to catch fire.
So you can see it's not really about typing.
When you realize that one of the biggest reasons that C and C++ programs do
unexplained things, up to and perhaps not including setting the computer on fire, is
when you have array bounds violations. Which are not checked, but most people
don't think of array bounds as related to the type system in these languages.
So, it turns out that decades ago people used to be fans of weak typing.
Some people were. Many people were not.
And, the people in favor of weak typing had this saying, strong types for weak
minds. So, the idea is that you would only want
to have strong typing, which is the opposite of weak typing, if you thought
the computers were smarter than humans. And we know they're not, we know that
static checking would never be perfect. So you always need the smart humans to
have some work around, who's turn off the checking that they may know is unnecessary
and the computer is not smart enough to realize is unnecessary.
That, in the end, languages should not get in the way of humans and humans should be
able to say, trust me. I say this is right.
And if I'm wrong, you can do anything. And I think the conventional wisdom has
really changed on this quite a bit. That, in reality, we've learned that
humans are really bad at avoiding bugs. And we need all the help we can get.
And if we can write a computer program that does some of the checking for us.
It's a nice division of responsibility. The person writing the static checker or
dynamic checker, can focus on getting just the checking right.
People writing applications can focus on the logic of their applications.
And rely on the automatic checking that we get in strongly type of programming
languages. It's also fair to point out that
conventional wisdom has changed because our type systems have just gotten a lot
better. They're more flexible we have things like
polymorphism. We have things like subtyping that made it
a lot easier to program in a language with a strong type system.
And not feel like it's getting in your way.
The weak typing argument in my mind, really breaks down when you look at the
modern size and complexity of software. There are operating systems today that
literally contain 30 or 40 or 50 million lines of C code.
If you're writing 200 or 1000 lines of C code, I'm sympathetic to the argument that
humans can look at that code and through hard work and staring carefully get most
of or even all of the bugs out of it. And maybe a type system shouldn't tell
them that it knows better. But when you get to 30 million lines, and
any one line of that program could cause the entire application to perform
arbitrary behavior. I just think it's ridiculous to expect
that we don't want some assurance that certain kinds of errors and preventable
properties are not possible in the software.
I like to joke when I'm teaching this stuff that there was an important bug
related to the weak typing of C and now it's just this week.
And I actually don't have to look that up because there are plenty of websites that
catalog these things, and I'm sure there is something posted on them every week.
So, you know, we just have too much software, too complex of software in the
world today, to sort of rely on humans to get their software right.
The thing I want to emphasize though, is that Racket is not weakly typed, this is
just a matter of definitions. Racket is dynamically typed.
It checks a lot of things like not using a number as a procedure At run time.
But that is very different than not checking them.
What we have is a language definition that says those errors will be detected at
particular points and an error will be raised.
Then in the language implementation, the checks can be removed.
If the implementation can perform some analysis to show that they would never
fail. So the language definition says these
checks have to be in there. If the check would ever fail, that has to
be indicated by an exception or an error in the program.
But you can remove checks in the implementation if you're sure they would
never fail. This is very different than the sort of
catch fire semantics we've been talking about with C and C++.
And just to finish up this segment, let me talk about another topic that is not weak
typing and is not dynamic typing, but is often confused with these things.
And that's this issue of, what primatives can, sorry, what operations can primitives
do? So in some languages, if you try to call
plus with strings, you get an error. That's a type error in ML, it's run time
error in Racket. But in other languages, plus just means
string concatenation. But maybe in those languages trying to add
a string and a number is an error. And then there are languages where it's
not an error. Where it will convert the number to a
string and you'll end up with a 4 character string like foo 3.
There are other things that I'm using to being errors.
And some languages say are not. If I access the 10th element of an array
and that array only has five elements is that an error?
Well, in some languages it says no you'll just get something like no back.
You'll get an empty list back of something like that.
Even if you assign to an array what if we just made an array bigger?
Suppose you had a function like in Racket you called it with the wrong number
arguments. That's an error in Racket.
But a different language can make a different choice.
It could say, if you pass too many arguments, we'll just ignore the extra
ones. Maybe if you pass too few arguments, we'll
pass in 79 for all the arguments you were missing.
You can define your language in these ways, to be more flexible, and beside that
fewer things are errors, but this is not actually static versus dynamic checking.
We sometimes think of languages that are more permissive on these things as being
more dynamic, but technically that's not what they are.
All that's going on is that we have changed the evaluation rules for a
primitive in our language, whether its accessing an array.
Calling a function or performing an addition.
And if you change the evaluation rules for your language, to allow more arguments,
more types of things, you are being more flexible.
You're increasing the set of legal programs.
At the expense, often times, of detecting bugs in a timely fashion.
Than, instead, you just silently continue, even if it's unlikely that that's what the
programmer meant. So, some of the trade offs between static
checking, which catches bugs earlier. And dynamic checking which catches them
later, apply here. Here we're being even more dynamic and
being even more delayed in announcing that something seems to have gone wrong.
But it's not really static checking at all, it's not really dynamic checking.
It's just saying that I've changed the evaluation rules of my language.
