PL Evaluation and Typing Systems

Note

These slides are also available in PDF format: 4:3 PDF, 16:9 PDF, 16:10 PDF.

Evaluating a Programming Language

Evaluation Metrics

Evaluating programming languages based on:

Writability:How easy is it to write good code?
Readability:How easy is it to read well written code? Is the language easy enough to learn?
Reliability:What features does the language provide to make sure our code works as it is supposed to?
Feasibility:Does an interpreter or compiler actually exist for the platform we need to use? Is it fast enough for our application?

A System of Trade-Offs

Often times, adding features which improve one metric can harm another metric. Examples to come…

Simplicity

The overall simplicity of a language plays a large role in both writability and readability.

For example, these features are non-simple:

  • Feature Multiplicity: 👍 Writability, 👎 Readability
  • Operator Overloading: 👍 Writability, 👎 Readability
  • Large Grammars: 👍 Writability, 👎 Readability

Simplicity can be carried too far

Assembly languages and esoteric languages generally aren’t considered very writable or readable.

Orthogonality

Orthogonality:how consistent is the language with itself?

Example of a lack of orthogonality (C++)

Parameters are passed by value, unless they were specified with an &.

Or unless they were an array.

Example of a lack of orthogonality (C/C++)

Arrays can contain data of any type, including pointers.

Unless it’s a function pointer.

But you can wrap that function pointer in a struct and you should be fine.

Impacts of poor orthogonality: poor readability, poor writability, and potentially reduced reliability.

Abstraction

Abstraction:The ability to define and use complicated structures and operations in a way that allows implementation to be ignored.

Examples:

  • Functions: Simplest form of abstraction. Often taken for granted, but gives us easy recursion.
  • Heap Memory: Imagine trying to create a large unbalanced binary tree in a single-dimensional array.
  • Generics: Allows us to define operations that apply to multiple data types without reimplementing for each type.

With your learning group…

What other kinds of PL-level abstractions can you name?

Good Abstractions: 👍 Writability, 👍 Readability, 👍 Reliability

Reliability Features

Some languages come with features designed for reliablitiy:

  • Type Checking: Making sure the type of data can be used with the function or operation you are calling. Independent of static/dynamic: more on this later.
  • Exception Handling: The ability of a running program to intercept run-time errors and take corrective measures.
  • Taint Protection: Protects the security of an application by not allowing privileged operations to be preformed on tainted data (e.g., user input from a web application).

Some features can harm a language’s reliability:

  • Goto: the ability to jump to different locations in the code without restriction.
  • Aliasing: allows two different symbolic names (variables, function names, etc.) to refer to the same data. Think pointers in C/C++.

Expressivitiy

Expressivity:How easy is it for the programmer to express their solution to a problem in the language?

With your learning group:

  1. Think of a scenario and two programming languages, where expressing a solution to the problem might be easier in one language than another.
  2. If one language is less expressive than another, how might it be less writable?
  3. If one language is less expressive than another, how might it be less readable?
  4. If one language is less expressive than another, how might it be less reliable?

Be prepared to share your answers with the class.

Type Systems

Type Systems: Overview

When we refer to “type systems”, we aren’t just talking about how you have to write the type of a variable in C, whereas you don’t in Python. There’s a lot that goes in:

  • Static/Dynamic Typing
  • Untyped Systems
  • Implicit/Explicit Type Specification
  • Strong/Weak Typing
  • Gradual Typing
  • Duck Typing – covered later in the course

Explicit/Implicit

In a language with explicit type specification, the type of a variable must be specified:

int x = 10;

In a language with implicit type specification, the type of a variable need not be specified:

var x = 10

Note

Explicit/implicit has nothing to do with static/dynamic. We will talk about that in a second…

Bindings

A binding refers to the association between:

  • a variable and its type,
  • a function and its definition,
  • a type and its representation (e.g., int is 32-bits),
  • or an operation and its symbol (e.g., multiplication is usually *)

Binding time refers to the time at which a binding takes place.

Common Binding Times

Design time, implementation time, compile time, link time, run time

Static Typing

In a statically typed language, the language design makes it possible to bind the type of any piece of data before run time.

Advantages:

  • No need to do type checking at run time, this can be done at compile time.
  • 👍 Reliability

Disadvantages:

  • Generics are needed to create operations and functions that apply to multiple types
  • 👎 Writability

Dynamic Typing

In a dynamically typed language, the language design makes it possible to bind the type of any piece of data during run time.

Commonly, the type of data is associated with the data itself.

Advantages:

  • Collections can be of mixed type without generics, functions can take multiple types without generics
  • Types can be dynamically created at run time
  • 👍 Writability

Disadvantages:

  • Type checking must be done at run time; makes things slow
  • 👎 Reliability

Gradual Typing

Gradual Typing can be used to refer to a language which allows optional explicit typing in a dynamically typed language.

Untyped Systems

In an untyped language, data cannot be bound to a type.

Commonly, the functions and operations called on the variables determine the type.

Advantages:

  • No need to do type checking, ever.
  • 👍 Feasibility

Disadvanges:

  • 👎 👎 👎 Reliability

Strongly and Weakly Typed

  • Type safety means a language will not allow bits to be intepreted as the incorrect data type. For example: treating the bits of an integer as a floating point number.
  • Implicit type conversions are when a language will automatically convert data types to allow an expression to be computed.
  • Strongly typed programming languages are both type safe and do not allow implicit type conversions.
  • Weakly typed programming languages are either not type safe or allow implicit type conversions.

Note

By definition, untyped languages are also weakly typed.

Exercise: Type Systems 1

Given the code snippet from a fake language below:

int a = 10
a += 5
print(a)
  • Explicit or implicit?
  • Is it possible that the language is statically typed?
  • Is it possible that the language is dynamically typed?
  • Weak or strong?

Exercise: Type Systems 2

Given the code snippet from a fake language below:

a = 10
a += 5
print(a)
  • Explicit or implicit?
  • Is it possible that the language is statically typed?
  • Is it possible that the language is dynamically typed?
  • Weak or strong?

Exercise: Type Systems 3

Given the code snippet from a fake language below:

a = eval(user_input())
a += 5
print(a)
  • Explicit or implicit?
  • Is it possible that the language is statically typed?
  • Is it possible that the language is dynamically typed?
  • Weak or strong?

Exercise: Type Systems 4

Given the code snippet from a fake language below:

++++++++[>++++[>++>+++>+++>+<<<<-]>+
>+>->>+[<]<-]>>.>---.+++++++..+++.>>
.<-.<.+++.------.-----+----.>>+.>++.
  • Explicit or implicit?
  • Is it possible that the language is statically typed?
  • Is it possible that the language is dynamically typed?
  • Weak or strong?

Type Systems: Language Examples

  Strong Weak
Static Haskell, Rust, Go C, C++
Dynamic Python, Ruby, Java PHP, JavaScript