本文是学习Coursera Programming Language课程的学习笔记，文章内容及代码均取自课程材料。
Interpreter or Compiler
实现编程语言的 workflow 如下图：
Parser 读取程序文本，检查 syntax，如果语法正确则输出 AST（abstract syntax tree）。如果该编程语言有 type checker，则将 AST 丢给 type checker 检查，通过 type check 后，就由 interpreter 或 compile 来运行程序并输出结果。
在实现编程语言 B 通常有下面两种办法：
- 使用另一种编程语言 A 来实现 interpreter（命名为 evaluator 或 executor 更恰当），输入 B 语言写的代码，输出结果
- 使用另一种编程语言 A 实现 compiler（命名为 translator 更恰当），将 B 翻译成第三种编程语言 C
Skipping Parsing
如果基于编程语言 A 来实现编程语言 B，就可以跳过 parsing 阶段：Have B programmers write ASTs directly in PL A
ML from a Racket perspective
ML is like a well-defined subset of Racket
Racket from an ML Perspective
One way to describe Racket is that it has “one big datatype”：all values have this type.
我的理解由于 Racket 是 dynamic typing，所以 ML 程序员看来 Racket 只有一个类型（这样 static type check 都能成功）
-
Constructors are applied implicitly (values are tagged)
42 is really like Int 42
-
Primitives implicitly check tags and extract data, raising errors for wrong constructors
1 2 3
fun car v = case v of Pair(a,b) => a | _ => raise … fun pair? v = case v of Pair _ => true | _ => false
Weak Typing
There exist programs that, by definition, must pass static checking but then when run can “set the computer on fire”?
- Ease of language implementation: Checks left to the programmer
- Performance: Dynamic checks take time
- Lower level: Compiler does not insert information like array sizes, so it cannot do the checks
Racket is not weakly typed
- It just checks most things dynamically*
- Dynamic checking is the definition – if the implementation can analyze the code to ensure some checks are not needed, then it can optimize them away
