Note: I corrected a few formatting errors on 29 January 2020.

Acknowledgements: These slides are adapted from the slide sets 0-5 for the course Programming in Lua created in July 2013 by Fabio Mascarenhas, Federal University of Rio de Janeiro, Brazil. It was presented at Nankai University, P. R. China, in July 2013. Although that course used Lua 5.2, I have attempted to update these slides to Lua 5.3. I produced these slided for CSci 450/503 (Organization of Programming Languages), Fall 2016.

Advisory: The HTML version of this document may require use of a browser that supports the display of MathML. A good choice as of January 2020 is a recent version of Firefox from Mozilla.

Lua Design Principles (1)

1. Portability

   • use common subset of standard C and C++
   • limit dependencies to standard C library
   • restrict memory used by core (small size)

2. Simplicity

   • provide small set of powerful primitives
   • add features conservatively
   • support mechanisms instead of policies

Lua Design Principles (2)

3. Embeddability

   • embed within applications
   • interoperate with C programs in both directions
   • enable extension

4. Efficiency

   • probably fastest, interpreted “scripting” language
   • LuaJIT (Just-In-Time compiler) supports Lua 5.1

Lua REPL

• REPL (Read-Evaluate-Print-Loop)

  • read a statement from command line
  • evaluate (execute) statement
  • print (display) result
  • loop to read next statement

• Supports interactive experimentation

• Execute REPL – run without script file argument

  lua

Statements & Chunks

• Statement is basic level – assignment, conditional, loop, function call

• Chunk is sequence of statements: unit of variable scoping

  • a loop body
  • a function body
  • all input to REPL execution

Separating Statements

• No statement separators required

  > a = 1 b = 2 print(a, b) 
  1       2 

• But using semicolon statement separator aids readability

  > a = 1; b = 2; print(a, b)
  1       2 

Running Scripts

• Better to put multiple-line input into file (a script)

• Example file defs.lua containing factorial function

  function fact(n)
      if n < 2 then
          return 1
      else
          return n * fact(n-1)
      end
  end

• Load file in REPL with dofile function

  > dofile("defs.lua")
  > print(fact(5))
  120

Basic Functions

• print and dofile built-in functions

• Function calls mostly familiar, with some unusual features

• Most built-in functions in different namespaces – math.sin, math.pi, math.sqrt

  > print(math.sin(math.pi/ 3))
  0.86602540378444
  > print(math.sqrt(3) / 2)
  0.86602540378444

Global Variables

• Lua variables are

  • global by default
  • mutable
  • not declared (except when local, addressed later)
  • initialized to special value nil

• Example

  > print(x)
  nil
  > x = 2
  > print(x)
  2

Variable Names

• Consist of sequences of letters, digits, & underscores not beginning with digit

• Include _ (one underscore) as dummy variable assigned but not accessed

• Reserved for Lua if begin with _ followed by all uppercase (e.g. _VERSION, _ENV, _G)

• Cannot include reserved words

  and      break    do       else     elseif   end
  false    for      function goto     if       in
  local    nil      not      or       repeat   return
  then     true     until    while

Comments

• End-of-line comments begin with -- (two hyphens)

• Block comments begin with --[[ and end with ]]

• ---[[ starts an end-of-line comment, not block (Useful for turning off block comments)

  --[[
  print(x) -- this line is commented out
  --]]
  ---[[
  print(x) -- this line is not
  --]]

Quiz

Which of the following names are valid variable names? Why/why not?

1. ---
2. _end
3. End
4. end
5. until?
6. Nil
7. NULL

Quiz Solution

Which of the following names are valid variable names? Why/why not?

1. --- – NO – a comment
2. _end – YES
3. End – Yes
4. end – No – reserved keyword
5. until? – No – ? not valid
6. Nil – Yes
7. NULL – Yes

Types

1. nil – only value is nil
2. boolean – values true and false
3. number
   • until 5.3, double precision floating point
   • beginning 5.3, both 64-bit integers and double precision floating point
4. string – immutable byte vectors, any byte values, in whatever coding
5. table – associative arrays
6. function – both named and anonymous
7. userdata – opaque blobs handled by external (e.g., C) libraries
8. thread – really coroutine

Dynamic Typing

• Variable does not have static type

• Variable’s value has type

• Assignment may change type of variable’s value

• Function type queries value, returns type as string

Nil Type

• Value nil denotes absence of useful value

• nil is value of

  • uninitialized variables
  • missing table fields
  • missing function parameters

• Most operations on nil are errors

Boolean Type

• Relational operators produce boolean value true or false

• But any value usable in condition or with logical operators

  • false and nil are false – called falsey
  • any other Lua value is true – called truthy
  • Careful! 0 and "" are truthy

• and operator yields first operand if falsey; otherwise second

• or operator yields first argument if truthy; otherwise second

• not operator always yields true or false

Logical Operator Idioms

• For optional parameter p of a function, use p or d where d is the default value

  function greeting(s)
      s = s or "Hello"
      print(s .. ", World!")
  end
  
  greeting()          -- "Hello, World!"
  greeting("Goodbye") -- "Goodbye, World!"

• For ternary operator like a ? b : c in C, use a and b or c

  Note: not binds more tightly than and, and binds more tightly than or

  function max(a, b)
      return (a > b) and a or b
  end

Number Type Before Lua 5.3

• Represented by IEEE double precision floating point

• Has no integer type, but double precision can precisely represent integers up to 53 bits

• Uses C and Java-like numeric literals, including scientific and hexadecimal (with “0x”) notation

• Supports usual arithmetic and relational operators: +, -, *, /, ^ (exponentiation), % (modulo), <, >, <=, >=, ==, and ~= (not equal)

• Careful! division by zero is not error

Number Type Changes in Lua 5.3

• Added alternative representation for type number: 64-bit integers

• Sought minimal changes to behavior

  • implicit conversions when needed
  • changes in 53-to-64-bit integer precision range
  • changes in output format for integers

• Added floor division operator // for integers

• Added math library functions maxinteger, mininteger

• Added math.type function to distinguish integer from float

String Type

• Lua string is immutable sequence of bytes

  • any byte value is okay, even 0
  • thus strings can store any binary data

• But what about text?

  • pick an encoding
  • Unicode UTF-8 has library support in Lua 5.3

• Built-in operations on strings

  • concatenation (..), length (#)
  • relational operators<, >, <=, >= use lexicographic ordering
  • automatic conversions (coercions) between numbers and strings

• More operations in string library, other libraries

String Literals

• Use either single or double quotes

  • double-quoted strings can include single quotes without escaping
  • vice versa

• Can also use escape sequences

  • \a (bell), \b (backspace), \f (form feed),\n (newline), \r (carriage return), \t (tab), \v (vertical tab), \\ (backslash), \" (double quote), \' (single quote)
  • \ddd (byte with decimal value ddd), \xhh (byte with hexadecimal value hh), \z (skips all spaces following it)
  • Latter 3 escapes for encoding binary data

Long String Literals

• Use delimiters [[ and ]]

• Do not interpret escape sequences

• Can extend for several lines; first character discarded if newline

• Can use other delimiters: [=[ and ]=], [==[ and ]==], [===[ and ]===], etc.

  escapes = [[
      \f (formfeed)
      \n (newline)
      \r (carriage return)
  ]]
  
  print("|" .. escapes .. "|")

Table Type

• Represent associative arrays – map key to value

  • key can be any value except nil
  • keys usually numbers or strings
  • efficiently implemented

• Can build flexible data structures

  • can represent arrays, structs, abstract data types, objects (in OO sense), modules, etc. (also in long comments)

• Used pervasively, everything is a table, more later

Function Type

• Are first-class values: can store functions in variables, pass as arguments, return as results

  • functions as table values basis for object-oriented programming, modules

• Can take variable number of parameters

• Can return multiple values (unusual feature!)

• Implemented with tail call optimization

• Used pervasively, more later

Quiz

How can you check whether a value is a boolean without using the type function?

Quiz Solution

How can you check whether a value is a boolean without using the type function?

    function isbool(x)
        return (not not x) == x
    end

if-then-else

• if statement executes then chunk if condition is truthy

• if statement executes optional else chunk if condition is falsey or does nothing if else omitted

  if a < 0 then
      print("a is negative")
      a = -a
  else
      print("a is positive")
  end

elseif

• Simplify structure using elseif clauses instead of nested if statements

  if op == "+" then
      r = a + b
  elseif op == "-" then
      r = a - b
  elseif op == "*" then
    r = a * b
  elseif op == "/" then
      r = a / b
  else
      error("invalid operation")
  end

• else clause remains optional

while and repeat

• while loop executes body chunk while condition truthy – tests at beginning

  i = 1; sum = 0
  while i <= 5 do
      sum = sum + (2 * i - 1)
      i = i + 1
  end
  print(sum)

• repeat loop executes body chunk until condition truthy – tests at end

   i = 1; sum = 0
   repeat
       sum = sum + (2 * i-1)
       i = i + 1
   until i > 5
   print(sum)

Numeric for (1)

• Numeric for loop steps control variable by 1 from starting to ending argument, executes body chunk for each value

  sum = 0
  for i = 1, 5 do
      sum = sum + (2 * i - 1)
  end
  print(sum)

• Control variable local to body, not known after loop

• Cannot assign to control variable

• Can use break statement to terminate loop early

Numeric for (2)

• Optional third argument to numeric for gives step value (1 if omitted)

  sum = 0
  for i = 5, 1, -1 do
      sum = sum + (2 * i - 1)
  end
  print(sum)

• Expressions for starting, ending, and step values evaluated only once before loop

• for loop executes zero times if starting value greater than ending (or lesser for negative step)

Local Variables

• local statement declares new variable visible from next statement to end of current chunk

  local sum = 0            -- local to the program 
  for i = 1, 5 do 
      local n = 2 * i - 1  -- local to the for body 
      sum = sum + n
  end 
  print(sum, n)            -- problem?!

• New local variables shadow global or local variables with same name

• Use local variables whenever possible

• Common idiom: cache value of global in local variable with same name

do-end

• Entering code from previous slide in REPL does not do what expected – each statement separate chunk

• do-end statement introduces new scope without changing control flow

  sum = 0
  do
      local i = 1
      while i <= 5 do
          sum = sum + (2 * i - 1)
          i = i + 1
      end
  end
  print(sum)

Multiple Assignment (1)

• Can assign to several different variables in single step

  > a, b = 10, 2 * sum
  > print(a, b)
  10      50

• First evaluate all expressions on right side, then do assignments

   > a, b = b, a
   > print(a, b)
   50      10

Multiple Assignment (2)

• If more variables than values to assign, assign nil to extra

  > a, b, sum = 10, 2 * sum 
  > print(a, b, sum) 
  10      50      nil 

• If more values than variables, ignore extra values

• Multiple assignments useful with functions that return multiple values

• local statement can declare and initialize several local variables with multiple assignment form

  local i, j, k = 1, 2, 3

Quiz

What is the result of running the following program? Why?

    local i = 5 
    while i do 
        print(i) 
        i = i - 1 
    end 

Quiz Solution

What is the result of running the following program? Why?

    local i = 5 
    while i do 
        print(i) 
        i = i - 1 
    end

Printing integers counting down from 5 – infinitely!

Functions

• Function abstracts and parameterizes chunk

• Function call either statement or expression

  a statement if executed for side effects only

  > print(8*9, 9/8) 
  72      1.125 

  an expression if returned value used

  > x = math.sin(3) + math.cos(10)
  > print(x, os.date())
  -0.69795152101659 06/23/13 14:49:03 

• Function has both side effects and return values

  Good style: Best to avoid (impure) functions that have side effects

Functions As Values

• Functions are first-class values

• Function names are just variables (in same namespace)

• Be careful with shadowing of function names

  > print(print) 
  function: 0000000068B94B40
  > do 
  >> local print = 10 
  >> print(print)
  >> end 
  stdin:3: attempt to call local 'print' (a number value) 
  stack traceback: 
          stdin:3: in main chunk 
          [C]: in ?

Defining Functions

• local functions local to chunk, just like local variables

  local function max(a, b) 
      return (a > b) and a or b 
  end

• Function body is chunk

• Function parameter local variable in body

• return statement returns control and values

Global Functions

• Functions can omit local

  function max(a, b)
      return (a > b) and a or b 
  end 

• If existing variable named max, then above function becomes new value

• If no variable named max, then above function becomes value of new global

• Following equivalent to local definition on previous slide:

  local max
  function max(a, b) 
      return (a > b) and a or b
  end 

Anonymous Functions

• Functions can be anonymous, defined in expression

  local max = function (a, b) 
      return (a > b) and a or b 
  end

• Actually all Lua functions are anonymous; other style just syntactic sugar

• Style note: Better to use function definition statements

Multiple Results

• Functions can return multiple values

• Useful with multiple assignments or multi-argument function calls

  > s, e = string.find("hello Lua users", "Lua") 
  > print(s, e)
  7       9 
  > print(string.find("hello Lua users", "Lua"))
  7       9

• List all return values after return

  function maxmin(a, b) 
      if a < b then 
          return b, a 
      else 
          return a, b 
      end 
  end

Using Multiple Results (1)

• If function call last in list of expressions, append all results

  > a, b = maxmin(2, 3) 
  > print(a, b) 
  3       2 
  > a, b, c = 1, maxmin(2, 3) 
  > print(a, b, c) 
  1       3       2 
  > print("maxmin(2, 3)", maxmin(2, 3)) 
  maxmin(2, 3)    3       2 
  > function f(x) return maxmin(x, 0) end 
  > print(f(-4)) 
  0       -4 

• Otherwise use just first result, ignore rest

• If no results, use nil

Using Multiple Results (2)

• Using just first result

  > a, b = maxmin(2, 3), 5 
  > print(a, b) 
  3       5 
  > print(maxmin(2, 3), 5) 
  3       5 
  > print(maxmin(2, 3) + 2) 
  5

• Enclosing function call in parenthesis forces just one result

  > a, b = (maxmin(2,3))
  > print(a, b) 
  3       nil 
  > print((maxmin(2,3))) 
  3

Variadic Functions

• Last parameter of function can be special token ...

• All arguments passed to call from position of ... forward referenced by ... inside body

  > function id(...) return ... end 
  > print(id(1, 2, 3)) 
  1       2       3 
  
  function printf(fmt, ...) 
    io.write(string.format(fmt, ...)) 
  end 
  
  > printf("%s(%d, %d)\n", "maxmin", 2, 3) 
  maxmin(2, 3) 

Quiz

What is the output of the following program?

    local function range(a, b, c) 
        if a > b then 
            return 
        else 
            return a, range(a + c, b, c) 
        end 
    end 
    print(range(1, 9, 2)) 

Quiz Solution

What is the output of the following program?

    local function range(a, b, c) 
        if a > b then 
            return 
        else 
            return a, range(a + c, b, c) 
        end 
    end 
    print(range(1, 9, 2)) 

    >1       3       5        7       9 

Tables for Everything

• Tables only way to structure data in pure Lua

• Can efficiently represent arrays, sets, records, objects, etc.

• Basic operations

  {} – construciton to make new table

  [] – indexing to read/write values

  > tab = {}        -- make a new table assign to tab
  > tab["x"] = 5    -- associate 5 with string key `"x"`
  > print(tab["x"]) -- read value of key `"x"` entry and print it

• Table is mutable reference type – careful of “pointer” aliasing (as with Java object references)

  > alias = tab
  > alias["x"] = "changed" 
  > print(tab["x"]) 
  changed

Arrays

• Table with values assigned to integer keys sequentially from 1

  local a = {} 
  for i= 1, 6 do 
      a[i] = math.random(10) 
  end 

• Initialize with list in table constructor

  local a = { math.random(10), math.random(10), math.random(10), 
              math.random(10), math.random(10), math.random(10) } 

• Do not allow value nil – array cannot have holes but can be filled in any order

Array Length

• Length operator # gives number of elements in array

• Use # to iterate over array with for loop

  local a = { math.random(10), math.random(10), math.random(10), 
              math.random(10), math.random(10), math.random(10) } 
  for i= 1, #a do 
      print(a[i]) 
  end 

• Common idiom: Use # to add/remove elements at end

  a[#a]     = nil             -- remove the last element 
  a[#a + 1] = math.random(10) -- add a new element to the end 

Array Insert & Remove

• Function table.insert inserts element at any position – shifts right

  > a = { 1, 2, 3, 4, 5 } 
  > table.insert(a, 3, 10) -- insert 10 in position 3 
  > print_array(a) 
  { 1, 2, 10, 3, 4, 5 } 

• Function table.remove deletes element from any position – shifts left

  > table.remove(a, 4)     -- remove fourth element 
  > print_array(a) 
  { 1, 2, 10, 4, 5 } 

Array Sort

• Function table.sort sorts elements efficiently

  > a = { "Python", "Lua", "C", "JavaScript", "Java", "Lisp" } 
  > table.sort(a) 
  > print_array(a) 
  { C, Java, JavaScript, Lisp, Lua, Python } 

Concatenation

• Function table.concat concatenates array of strings using optional separator

   function print_array(a) 
       print("{" .. table.concat(a, ", ") .. "}") 
   end 

• If no separator then use ""

• Common idiom: use array of strings as buffer, then concatenate

Iteration with ipairs

• Can iterate over array using generic for loop and ipairs function

  local a = { 1, 3, 5, 7, 9 } 
  local sum = 0 
  for i, x in ipairs(a) do 
      print("adding element " .. i) 
      sum = sum + x 
  end 
  print("the sum is " .. sum) 

• Has two control variables – first gets indices, second gets elements

• Often interested only in elements, so use _ as control variable for indices

Matrices

• Can represent multi-dimensional arrays with “jagged arrays” – arrays of arrays

  local mt = {}
  for i= 1, 3 do
      mt[i] = {}
      for j = 1, 5 do 
          mt[i][j] = 0 
      end 
  end 

• Alternatively can use multiplication to compute index directly

  local mt = {} 
  for i= 1, 3 do 
      for j = 1, 5 do 
          mt[(i-1)*5+j] = 0
      end 
  end 

Records

• Table with string keys that are valid Lua identifiers

  point1 = { x = 10, y = 20 } 
  point2 = { x = 50, y = 5  } 
  line = { from = point1, to = point2, color = "blue" } 

• Read and write fields with the . operator

  line.color = "red"               -- same as line["color"] = "red" 
  print(line.from.x, line["color"]) 

• No inherent order among keys

• Table can be both record and array – initialize both in single constructor

Sets and Bags

• Represent set with Table where keys are the elements, values are true (disallows nil as set member)

• Can test membership and add/remove elements by indexing

• Initialize using third table constructor syntax [expression]

  -- a set of four random integers from 1 to 10
  local set = { [math.random(10)] = true, [math.random(10)] = true, 
                [math.random(10)] = true, [math.random(10)] = true } 

• Replace true by number of occurrences to get bag (or multiset) data structure

Iteration with pairs

• Can iterate over table with generic for using pairs function to get all key-value pairs

  local tab = { x = 5, y = 3, 10, "foo" } 
  for k, v in pairs(tab) do 
      print(tostring(k) .. " = " .. tostring(v)) 
  end 

• First control variable gets keys, second gets values

• Function pairs does not guarantee order, even using numeric keys; ipairs does use order for arrays

• Function pairs works well for sets using _ to ignore control variable for values

Quiz

What will be the output of following program?

    sunday = "monday" ; monday = "sunday"
    t = { sunday = "monday", [sunday] = monday }
    print(t.sunday, t[sunday], t[t.sunday]) 

Quiz Solution

    sunday = "monday" ; monday = "sunday"
    t = { sunday = "monday", [sunday] = monday }
    print(t.sunday, t[sunday], t[t.sunday]) 

Second line same as:

    t = {} ; t["sunday"] = "monday" ; t["monday"] = "sunday"

Last? Variable sunday yields "monday". So t[sunday] same as t["monday"]. Variable monday yields "sunday". Thus assignment is t["monday"] = "sunday".

Thus t == { sunday = "monday", monday = "sunday" }. So print(t.sunday, t[sunday], t[t.sunday]):

    monday  sunday  sunday