blob: 5fb26b23c809187f6478215e1efdc76e528cbf25 [file] [log] [blame]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<TITLE>Lua 5.3 Reference Manual</TITLE>
<LINK REL="stylesheet" TYPE="text/css" HREF="lua.css">
<LINK REL="stylesheet" TYPE="text/css" HREF="manual.css">
<META HTTP-EQUIV="content-type" CONTENT="text/html; charset=iso-8859-1">
<A HREF=""><IMG SRC="logo.gif" ALT="Lua"></A>
Lua 5.3 Reference Manual
by Roberto Ierusalimschy, Luiz Henrique de Figueiredo, Waldemar Celes
Copyright &copy; 2015&ndash;2016, PUC-Rio.
Freely available under the terms of the
<a href="">Lua license</a>.
<DIV CLASS="menubar">
<A HREF="contents.html#contents">contents</A>
<A HREF="contents.html#index">index</A>
<A HREF="">other versions</A>
<!-- ====================================================================== -->
<!-- $Id: manual.of,v 1.162 2016/05/30 15:57:03 roberto Exp $ -->
<h1>1 &ndash; <a name="1">Introduction</a></h1>
Lua is a powerful, efficient, lightweight, embeddable scripting language.
It supports procedural programming,
object-oriented programming, functional programming,
data-driven programming, and data description.
Lua combines simple procedural syntax with powerful data description
constructs based on associative arrays and extensible semantics.
Lua is dynamically typed,
runs by interpreting bytecode with a register-based
virtual machine,
and has automatic memory management with
incremental garbage collection,
making it ideal for configuration, scripting,
and rapid prototyping.
Lua is implemented as a library, written in <em>clean C</em>,
the common subset of Standard&nbsp;C and C++.
The Lua distribution includes a host program called <code>lua</code>,
which uses the Lua library to offer a complete,
standalone Lua interpreter,
for interactive or batch use.
Lua is intended to be used both as a powerful, lightweight,
embeddable scripting language for any program that needs one,
and as a powerful but lightweight and efficient stand-alone language.
As an extension language, Lua has no notion of a "main" program:
it works <em>embedded</em> in a host client,
called the <em>embedding program</em> or simply the <em>host</em>.
(Frequently, this host is the stand-alone <code>lua</code> program.)
The host program can invoke functions to execute a piece of Lua code,
can write and read Lua variables,
and can register C&nbsp;functions to be called by Lua code.
Through the use of C&nbsp;functions, Lua can be augmented to cope with
a wide range of different domains,
thus creating customized programming languages sharing a syntactical framework.
Lua is free software,
and is provided as usual with no guarantees,
as stated in its license.
The implementation described in this manual is available
at Lua's official web site, <code></code>.
Like any other reference manual,
this document is dry in places.
For a discussion of the decisions behind the design of Lua,
see the technical papers available at Lua's web site.
For a detailed introduction to programming in Lua,
see Roberto's book, <em>Programming in Lua</em>.
<h1>2 &ndash; <a name="2">Basic Concepts</a></h1>
This section describes the basic concepts of the language.
<h2>2.1 &ndash; <a name="2.1">Values and Types</a></h2>
Lua is a <em>dynamically typed language</em>.
This means that
variables do not have types; only values do.
There are no type definitions in the language.
All values carry their own type.
All values in Lua are <em>first-class values</em>.
This means that all values can be stored in variables,
passed as arguments to other functions, and returned as results.
There are eight basic types in Lua:
<em>nil</em>, <em>boolean</em>, <em>number</em>,
<em>string</em>, <em>function</em>, <em>userdata</em>,
<em>thread</em>, and <em>table</em>.
The type <em>nil</em> has one single value, <b>nil</b>,
whose main property is to be different from any other value;
it usually represents the absence of a useful value.
The type <em>boolean</em> has two values, <b>false</b> and <b>true</b>.
Both <b>nil</b> and <b>false</b> make a condition false;
any other value makes it true.
The type <em>number</em> represents both
integer numbers and real (floating-point) numbers.
The type <em>string</em> represents immutable sequences of bytes.
Lua is 8-bit clean:
strings can contain any 8-bit value,
including embedded zeros ('<code>\0</code>').
Lua is also encoding-agnostic;
it makes no assumptions about the contents of a string.
The type <em>number</em> uses two internal representations,
or two subtypes,
one called <em>integer</em> and the other called <em>float</em>.
Lua has explicit rules about when each representation is used,
but it also converts between them automatically as needed (see <a href="#3.4.3">&sect;3.4.3</a>).
the programmer may choose to mostly ignore the difference
between integers and floats
or to assume complete control over the representation of each number.
Standard Lua uses 64-bit integers and double-precision (64-bit) floats,
but you can also compile Lua so that it
uses 32-bit integers and/or single-precision (32-bit) floats.
The option with 32 bits for both integers and floats
is particularly attractive
for small machines and embedded systems.
(See macro <code>LUA_32BITS</code> in file <code>luaconf.h</code>.)
Lua can call (and manipulate) functions written in Lua and
functions written in C (see <a href="#3.4.10">&sect;3.4.10</a>).
Both are represented by the type <em>function</em>.
The type <em>userdata</em> is provided to allow arbitrary C&nbsp;data to
be stored in Lua variables.
A userdata value represents a block of raw memory.
There are two kinds of userdata:
<em>full userdata</em>,
which is an object with a block of memory managed by Lua,
and <em>light userdata</em>,
which is simply a C&nbsp;pointer value.
Userdata has no predefined operations in Lua,
except assignment and identity test.
By using <em>metatables</em>,
the programmer can define operations for full userdata values
(see <a href="#2.4">&sect;2.4</a>).
Userdata values cannot be created or modified in Lua,
only through the C&nbsp;API.
This guarantees the integrity of data owned by the host program.
The type <em>thread</em> represents independent threads of execution
and it is used to implement coroutines (see <a href="#2.6">&sect;2.6</a>).
Lua threads are not related to operating-system threads.
Lua supports coroutines on all systems,
even those that do not support threads natively.
The type <em>table</em> implements associative arrays,
that is, arrays that can be indexed not only with numbers,
but with any Lua value except <b>nil</b> and NaN.
(<em>Not a Number</em> is a special value used to represent
undefined or unrepresentable numerical results, such as <code>0/0</code>.)
Tables can be <em>heterogeneous</em>;
that is, they can contain values of all types (except <b>nil</b>).
Any key with value <b>nil</b> is not considered part of the table.
Conversely, any key that is not part of a table has
an associated value <b>nil</b>.
Tables are the sole data-structuring mechanism in Lua;
they can be used to represent ordinary arrays, sequences,
symbol tables, sets, records, graphs, trees, etc.
To represent records, Lua uses the field name as an index.
The language supports this representation by
providing <code></code> as syntactic sugar for <code>a["name"]</code>.
There are several convenient ways to create tables in Lua
(see <a href="#3.4.9">&sect;3.4.9</a>).
We use the term <em>sequence</em> to denote a table where
the set of all positive numeric keys is equal to {1..<em>n</em>}
for some non-negative integer <em>n</em>,
which is called the length of the sequence (see <a href="#3.4.7">&sect;3.4.7</a>).
Like indices,
the values of table fields can be of any type.
In particular,
because functions are first-class values,
table fields can contain functions.
Thus tables can also carry <em>methods</em> (see <a href="#3.4.11">&sect;3.4.11</a>).
The indexing of tables follows
the definition of raw equality in the language.
The expressions <code>a[i]</code> and <code>a[j]</code>
denote the same table element
if and only if <code>i</code> and <code>j</code> are raw equal
(that is, equal without metamethods).
In particular, floats with integral values
are equal to their respective integers
(e.g., <code>1.0 == 1</code>).
To avoid ambiguities,
any float with integral value used as a key
is converted to its respective integer.
For instance, if you write <code>a[2.0] = true</code>,
the actual key inserted into the table will be the
integer <code>2</code>.
(On the other hand,
2 and "<code>2</code>" are different Lua values and therefore
denote different table entries.)
Tables, functions, threads, and (full) userdata values are <em>objects</em>:
variables do not actually <em>contain</em> these values,
only <em>references</em> to them.
Assignment, parameter passing, and function returns
always manipulate references to such values;
these operations do not imply any kind of copy.
The library function <a href="#pdf-type"><code>type</code></a> returns a string describing the type
of a given value (see <a href="#6.1">&sect;6.1</a>).
<h2>2.2 &ndash; <a name="2.2">Environments and the Global Environment</a></h2>
As will be discussed in <a href="#3.2">&sect;3.2</a> and <a href="#3.3.3">&sect;3.3.3</a>,
any reference to a free name
(that is, a name not bound to any declaration) <code>var</code>
is syntactically translated to <code>_ENV.var</code>.
Moreover, every chunk is compiled in the scope of
an external local variable named <code>_ENV</code> (see <a href="#3.3.2">&sect;3.3.2</a>),
so <code>_ENV</code> itself is never a free name in a chunk.
Despite the existence of this external <code>_ENV</code> variable and
the translation of free names,
<code>_ENV</code> is a completely regular name.
In particular,
you can define new variables and parameters with that name.
Each reference to a free name uses the <code>_ENV</code> that is
visible at that point in the program,
following the usual visibility rules of Lua (see <a href="#3.5">&sect;3.5</a>).
Any table used as the value of <code>_ENV</code> is called an <em>environment</em>.
Lua keeps a distinguished environment called the <em>global environment</em>.
This value is kept at a special index in the C registry (see <a href="#4.5">&sect;4.5</a>).
In Lua, the global variable <a href="#pdf-_G"><code>_G</code></a> is initialized with this same value.
(<a href="#pdf-_G"><code>_G</code></a> is never used internally.)
When Lua loads a chunk,
the default value for its <code>_ENV</code> upvalue
is the global environment (see <a href="#pdf-load"><code>load</code></a>).
Therefore, by default,
free names in Lua code refer to entries in the global environment
(and, therefore, they are also called <em>global variables</em>).
Moreover, all standard libraries are loaded in the global environment
and some functions there operate on that environment.
You can use <a href="#pdf-load"><code>load</code></a> (or <a href="#pdf-loadfile"><code>loadfile</code></a>)
to load a chunk with a different environment.
(In C, you have to load the chunk and then change the value
of its first upvalue.)
<h2>2.3 &ndash; <a name="2.3">Error Handling</a></h2>
Because Lua is an embedded extension language,
all Lua actions start from C&nbsp;code in the host program
calling a function from the Lua library.
(When you use Lua standalone,
the <code>lua</code> application is the host program.)
Whenever an error occurs during
the compilation or execution of a Lua chunk,
control returns to the host,
which can take appropriate measures
(such as printing an error message).
Lua code can explicitly generate an error by calling the
<a href="#pdf-error"><code>error</code></a> function.
If you need to catch errors in Lua,
you can use <a href="#pdf-pcall"><code>pcall</code></a> or <a href="#pdf-xpcall"><code>xpcall</code></a>
to call a given function in <em>protected mode</em>.
Whenever there is an error,
an <em>error object</em> (also called an <em>error message</em>)
is propagated with information about the error.
Lua itself only generates errors whose error object is a string,
but programs may generate errors with
any value as the error object.
It is up to the Lua program or its host to handle such error objects.
When you use <a href="#pdf-xpcall"><code>xpcall</code></a> or <a href="#lua_pcall"><code>lua_pcall</code></a>,
you may give a <em>message handler</em>
to be called in case of errors.
This function is called with the original error object
and returns a new error object.
It is called before the error unwinds the stack,
so that it can gather more information about the error,
for instance by inspecting the stack and creating a stack traceback.
This message handler is still protected by the protected call;
so, an error inside the message handler
will call the message handler again.
If this loop goes on for too long,
Lua breaks it and returns an appropriate message.
<h2>2.4 &ndash; <a name="2.4">Metatables and Metamethods</a></h2>
Every value in Lua can have a <em>metatable</em>.
This <em>metatable</em> is an ordinary Lua table
that defines the behavior of the original value
under certain special operations.
You can change several aspects of the behavior
of operations over a value by setting specific fields in its metatable.
For instance, when a non-numeric value is the operand of an addition,
Lua checks for a function in the field "<code>__add</code>" of the value's metatable.
If it finds one,
Lua calls this function to perform the addition.
The key for each event in a metatable is a string
with the event name prefixed by two underscores;
the corresponding values are called <em>metamethods</em>.
In the previous example, the key is "<code>__add</code>"
and the metamethod is the function that performs the addition.
You can query the metatable of any value
using the <a href="#pdf-getmetatable"><code>getmetatable</code></a> function.
Lua queries metamethods in metatables using a raw access (see <a href="#pdf-rawget"><code>rawget</code></a>).
So, to retrieve the metamethod for event <code>ev</code> in object <code>o</code>,
Lua does the equivalent to the following code:
rawget(getmetatable(<em>o</em>) or {}, "__<em>ev</em>")
You can replace the metatable of tables
using the <a href="#pdf-setmetatable"><code>setmetatable</code></a> function.
You cannot change the metatable of other types from Lua code
(except by using the debug library (<a href="#6.10">&sect;6.10</a>));
you should use the C&nbsp;API for that.
Tables and full userdata have individual metatables
(although multiple tables and userdata can share their metatables).
Values of all other types share one single metatable per type;
that is, there is one single metatable for all numbers,
one for all strings, etc.
By default, a value has no metatable,
but the string library sets a metatable for the string type (see <a href="#6.4">&sect;6.4</a>).
A metatable controls how an object behaves in
arithmetic operations, bitwise operations,
order comparisons, concatenation, length operation, calls, and indexing.
A metatable also can define a function to be called
when a userdata or a table is garbage collected (<a href="#2.5">&sect;2.5</a>).
For the unary operators (negation, length, and bitwise NOT),
the metamethod is computed and called with a dummy second operand,
equal to the first one.
This extra operand is only to simplify Lua's internals
(by making these operators behave like a binary operation)
and may be removed in future versions.
(For most uses this extra operand is irrelevant.)
A detailed list of events controlled by metatables is given next.
Each operation is identified by its corresponding key.
<li><b><code>__add</code>: </b>
the addition (<code>+</code>) operation.
If any operand for an addition is not a number
(nor a string coercible to a number),
Lua will try to call a metamethod.
First, Lua will check the first operand (even if it is valid).
If that operand does not define a metamethod for <code>__add</code>,
then Lua will check the second operand.
If Lua can find a metamethod,
it calls the metamethod with the two operands as arguments,
and the result of the call
(adjusted to one value)
is the result of the operation.
it raises an error.
<li><b><code>__sub</code>: </b>
the subtraction (<code>-</code>) operation.
Behavior similar to the addition operation.
<li><b><code>__mul</code>: </b>
the multiplication (<code>*</code>) operation.
Behavior similar to the addition operation.
<li><b><code>__div</code>: </b>
the division (<code>/</code>) operation.
Behavior similar to the addition operation.
<li><b><code>__mod</code>: </b>
the modulo (<code>%</code>) operation.
Behavior similar to the addition operation.
<li><b><code>__pow</code>: </b>
the exponentiation (<code>^</code>) operation.
Behavior similar to the addition operation.
<li><b><code>__unm</code>: </b>
the negation (unary <code>-</code>) operation.
Behavior similar to the addition operation.
<li><b><code>__idiv</code>: </b>
the floor division (<code>//</code>) operation.
Behavior similar to the addition operation.
<li><b><code>__band</code>: </b>
the bitwise AND (<code>&amp;</code>) operation.
Behavior similar to the addition operation,
except that Lua will try a metamethod
if any operand is neither an integer
nor a value coercible to an integer (see <a href="#3.4.3">&sect;3.4.3</a>).
<li><b><code>__bor</code>: </b>
the bitwise OR (<code>|</code>) operation.
Behavior similar to the bitwise AND operation.
<li><b><code>__bxor</code>: </b>
the bitwise exclusive OR (binary <code>~</code>) operation.
Behavior similar to the bitwise AND operation.
<li><b><code>__bnot</code>: </b>
the bitwise NOT (unary <code>~</code>) operation.
Behavior similar to the bitwise AND operation.
<li><b><code>__shl</code>: </b>
the bitwise left shift (<code>&lt;&lt;</code>) operation.
Behavior similar to the bitwise AND operation.
<li><b><code>__shr</code>: </b>
the bitwise right shift (<code>&gt;&gt;</code>) operation.
Behavior similar to the bitwise AND operation.
<li><b><code>__concat</code>: </b>
the concatenation (<code>..</code>) operation.
Behavior similar to the addition operation,
except that Lua will try a metamethod
if any operand is neither a string nor a number
(which is always coercible to a string).
<li><b><code>__len</code>: </b>
the length (<code>#</code>) operation.
If the object is not a string,
Lua will try its metamethod.
If there is a metamethod,
Lua calls it with the object as argument,
and the result of the call
(always adjusted to one value)
is the result of the operation.
If there is no metamethod but the object is a table,
then Lua uses the table length operation (see <a href="#3.4.7">&sect;3.4.7</a>).
Otherwise, Lua raises an error.
<li><b><code>__eq</code>: </b>
the equal (<code>==</code>) operation.
Behavior similar to the addition operation,
except that Lua will try a metamethod only when the values
being compared are either both tables or both full userdata
and they are not primitively equal.
The result of the call is always converted to a boolean.
<li><b><code>__lt</code>: </b>
the less than (<code>&lt;</code>) operation.
Behavior similar to the addition operation,
except that Lua will try a metamethod only when the values
being compared are neither both numbers nor both strings.
The result of the call is always converted to a boolean.
<li><b><code>__le</code>: </b>
the less equal (<code>&lt;=</code>) operation.
Unlike other operations,
the less-equal operation can use two different events.
First, Lua looks for the <code>__le</code> metamethod in both operands,
like in the less than operation.
If it cannot find such a metamethod,
then it will try the <code>__lt</code> metamethod,
assuming that <code>a &lt;= b</code> is equivalent to <code>not (b &lt; a)</code>.
As with the other comparison operators,
the result is always a boolean.
(This use of the <code>__lt</code> event can be removed in future versions;
it is also slower than a real <code>__le</code> metamethod.)
<li><b><code>__index</code>: </b>
The indexing access <code>table[key]</code>.
This event happens when <code>table</code> is not a table or
when <code>key</code> is not present in <code>table</code>.
The metamethod is looked up in <code>table</code>.
Despite the name,
the metamethod for this event can be either a function or a table.
If it is a function,
it is called with <code>table</code> and <code>key</code> as arguments,
and the result of the call
(adjusted to one value)
is the result of the operation.
If it is a table,
the final result is the result of indexing this table with <code>key</code>.
(This indexing is regular, not raw,
and therefore can trigger another metamethod.)
<li><b><code>__newindex</code>: </b>
The indexing assignment <code>table[key] = value</code>.
Like the index event,
this event happens when <code>table</code> is not a table or
when <code>key</code> is not present in <code>table</code>.
The metamethod is looked up in <code>table</code>.
Like with indexing,
the metamethod for this event can be either a function or a table.
If it is a function,
it is called with <code>table</code>, <code>key</code>, and <code>value</code> as arguments.
If it is a table,
Lua does an indexing assignment to this table with the same key and value.
(This assignment is regular, not raw,
and therefore can trigger another metamethod.)
Whenever there is a <code>__newindex</code> metamethod,
Lua does not perform the primitive assignment.
(If necessary,
the metamethod itself can call <a href="#pdf-rawset"><code>rawset</code></a>
to do the assignment.)
<li><b><code>__call</code>: </b>
The call operation <code>func(args)</code>.
This event happens when Lua tries to call a non-function value
(that is, <code>func</code> is not a function).
The metamethod is looked up in <code>func</code>.
If present,
the metamethod is called with <code>func</code> as its first argument,
followed by the arguments of the original call (<code>args</code>).
All results of the call
are the result of the operation.
(This is the only metamethod that allows multiple results.)
It is a good practice to add all needed metamethods to a table
before setting it as a metatable of some object.
In particular, the <code>__gc</code> metamethod works only when this order
is followed (see <a href="#2.5.1">&sect;2.5.1</a>).
Because metatables are regular tables,
they can contain arbitrary fields,
not only the event names defined above.
Some functions in the standard library
(e.g., <a href="#pdf-tostring"><code>tostring</code></a>)
use other fields in metatables for their own purposes.
<h2>2.5 &ndash; <a name="2.5">Garbage Collection</a></h2>
Lua performs automatic memory management.
This means that
you do not have to worry about allocating memory for new objects
or freeing it when the objects are no longer needed.
Lua manages memory automatically by running
a <em>garbage collector</em> to collect all <em>dead objects</em>
(that is, objects that are no longer accessible from Lua).
All memory used by Lua is subject to automatic management:
strings, tables, userdata, functions, threads, internal structures, etc.
Lua implements an incremental mark-and-sweep collector.
It uses two numbers to control its garbage-collection cycles:
the <em>garbage-collector pause</em> and
the <em>garbage-collector step multiplier</em>.
Both use percentage points as units
(e.g., a value of 100 means an internal value of 1).
The garbage-collector pause
controls how long the collector waits before starting a new cycle.
Larger values make the collector less aggressive.
Values smaller than 100 mean the collector will not wait to
start a new cycle.
A value of 200 means that the collector waits for the total memory in use
to double before starting a new cycle.
The garbage-collector step multiplier
controls the relative speed of the collector relative to
memory allocation.
Larger values make the collector more aggressive but also increase
the size of each incremental step.
You should not use values smaller than 100,
because they make the collector too slow and
can result in the collector never finishing a cycle.
The default is 200,
which means that the collector runs at "twice"
the speed of memory allocation.
If you set the step multiplier to a very large number
(larger than 10% of the maximum number of
bytes that the program may use),
the collector behaves like a stop-the-world collector.
If you then set the pause to 200,
the collector behaves as in old Lua versions,
doing a complete collection every time Lua doubles its
memory usage.
You can change these numbers by calling <a href="#lua_gc"><code>lua_gc</code></a> in C
or <a href="#pdf-collectgarbage"><code>collectgarbage</code></a> in Lua.
You can also use these functions to control
the collector directly (e.g., stop and restart it).
<h3>2.5.1 &ndash; <a name="2.5.1">Garbage-Collection Metamethods</a></h3>
You can set garbage-collector metamethods for tables
and, using the C&nbsp;API,
for full userdata (see <a href="#2.4">&sect;2.4</a>).
These metamethods are also called <em>finalizers</em>.
Finalizers allow you to coordinate Lua's garbage collection
with external resource management
(such as closing files, network or database connections,
or freeing your own memory).
For an object (table or userdata) to be finalized when collected,
you must <em>mark</em> it for finalization.
You mark an object for finalization when you set its metatable
and the metatable has a field indexed by the string "<code>__gc</code>".
Note that if you set a metatable without a <code>__gc</code> field
and later create that field in the metatable,
the object will not be marked for finalization.
When a marked object becomes garbage,
it is not collected immediately by the garbage collector.
Instead, Lua puts it in a list.
After the collection,
Lua goes through that list.
For each object in the list,
it checks the object's <code>__gc</code> metamethod:
If it is a function,
Lua calls it with the object as its single argument;
if the metamethod is not a function,
Lua simply ignores it.
At the end of each garbage-collection cycle,
the finalizers for objects are called in
the reverse order that the objects were marked for finalization,
among those collected in that cycle;
that is, the first finalizer to be called is the one associated
with the object marked last in the program.
The execution of each finalizer may occur at any point during
the execution of the regular code.
Because the object being collected must still be used by the finalizer,
that object (and other objects accessible only through it)
must be <em>resurrected</em> by Lua.
Usually, this resurrection is transient,
and the object memory is freed in the next garbage-collection cycle.
However, if the finalizer stores the object in some global place
(e.g., a global variable),
then the resurrection is permanent.
Moreover, if the finalizer marks a finalizing object for finalization again,
its finalizer will be called again in the next cycle where the
object is unreachable.
In any case,
the object memory is freed only in a GC cycle where
the object is unreachable and not marked for finalization.
When you close a state (see <a href="#lua_close"><code>lua_close</code></a>),
Lua calls the finalizers of all objects marked for finalization,
following the reverse order that they were marked.
If any finalizer marks objects for collection during that phase,
these marks have no effect.
<h3>2.5.2 &ndash; <a name="2.5.2">Weak Tables</a></h3>
A <em>weak table</em> is a table whose elements are
<em>weak references</em>.
A weak reference is ignored by the garbage collector.
In other words,
if the only references to an object are weak references,
then the garbage collector will collect that object.
A weak table can have weak keys, weak values, or both.
A table with weak values allows the collection of its values,
but prevents the collection of its keys.
A table with both weak keys and weak values allows the collection of
both keys and values.
In any case, if either the key or the value is collected,
the whole pair is removed from the table.
The weakness of a table is controlled by the
<code>__mode</code> field of its metatable.
If the <code>__mode</code> field is a string containing the character&nbsp;'<code>k</code>',
the keys in the table are weak.
If <code>__mode</code> contains '<code>v</code>',
the values in the table are weak.
A table with weak keys and strong values
is also called an <em>ephemeron table</em>.
In an ephemeron table,
a value is considered reachable only if its key is reachable.
In particular,
if the only reference to a key comes through its value,
the pair is removed.
Any change in the weakness of a table may take effect only
at the next collect cycle.
In particular, if you change the weakness to a stronger mode,
Lua may still collect some items from that table
before the change takes effect.
Only objects that have an explicit construction
are removed from weak tables.
Values, such as numbers and light C functions,
are not subject to garbage collection,
and therefore are not removed from weak tables
(unless their associated values are collected).
Although strings are subject to garbage collection,
they do not have an explicit construction,
and therefore are not removed from weak tables.
Resurrected objects
(that is, objects being finalized
and objects accessible only through objects being finalized)
have a special behavior in weak tables.
They are removed from weak values before running their finalizers,
but are removed from weak keys only in the next collection
after running their finalizers, when such objects are actually freed.
This behavior allows the finalizer to access properties
associated with the object through weak tables.
If a weak table is among the resurrected objects in a collection cycle,
it may not be properly cleared until the next cycle.
<h2>2.6 &ndash; <a name="2.6">Coroutines</a></h2>
Lua supports coroutines,
also called <em>collaborative multithreading</em>.
A coroutine in Lua represents an independent thread of execution.
Unlike threads in multithread systems, however,
a coroutine only suspends its execution by explicitly calling
a yield function.
You create a coroutine by calling <a href="#pdf-coroutine.create"><code>coroutine.create</code></a>.
Its sole argument is a function
that is the main function of the coroutine.
The <code>create</code> function only creates a new coroutine and
returns a handle to it (an object of type <em>thread</em>);
it does not start the coroutine.
You execute a coroutine by calling <a href="#pdf-coroutine.resume"><code>coroutine.resume</code></a>.
When you first call <a href="#pdf-coroutine.resume"><code>coroutine.resume</code></a>,
passing as its first argument
a thread returned by <a href="#pdf-coroutine.create"><code>coroutine.create</code></a>,
the coroutine starts its execution by
calling its main function.
Extra arguments passed to <a href="#pdf-coroutine.resume"><code>coroutine.resume</code></a> are passed
as arguments to that function.
After the coroutine starts running,
it runs until it terminates or <em>yields</em>.
A coroutine can terminate its execution in two ways:
normally, when its main function returns
(explicitly or implicitly, after the last instruction);
and abnormally, if there is an unprotected error.
In case of normal termination,
<a href="#pdf-coroutine.resume"><code>coroutine.resume</code></a> returns <b>true</b>,
plus any values returned by the coroutine main function.
In case of errors, <a href="#pdf-coroutine.resume"><code>coroutine.resume</code></a> returns <b>false</b>
plus an error object.
A coroutine yields by calling <a href="#pdf-coroutine.yield"><code>coroutine.yield</code></a>.
When a coroutine yields,
the corresponding <a href="#pdf-coroutine.resume"><code>coroutine.resume</code></a> returns immediately,
even if the yield happens inside nested function calls
(that is, not in the main function,
but in a function directly or indirectly called by the main function).
In the case of a yield, <a href="#pdf-coroutine.resume"><code>coroutine.resume</code></a> also returns <b>true</b>,
plus any values passed to <a href="#pdf-coroutine.yield"><code>coroutine.yield</code></a>.
The next time you resume the same coroutine,
it continues its execution from the point where it yielded,
with the call to <a href="#pdf-coroutine.yield"><code>coroutine.yield</code></a> returning any extra
arguments passed to <a href="#pdf-coroutine.resume"><code>coroutine.resume</code></a>.
Like <a href="#pdf-coroutine.create"><code>coroutine.create</code></a>,
the <a href="#pdf-coroutine.wrap"><code>coroutine.wrap</code></a> function also creates a coroutine,
but instead of returning the coroutine itself,
it returns a function that, when called, resumes the coroutine.
Any arguments passed to this function
go as extra arguments to <a href="#pdf-coroutine.resume"><code>coroutine.resume</code></a>.
<a href="#pdf-coroutine.wrap"><code>coroutine.wrap</code></a> returns all the values returned by <a href="#pdf-coroutine.resume"><code>coroutine.resume</code></a>,
except the first one (the boolean error code).
Unlike <a href="#pdf-coroutine.resume"><code>coroutine.resume</code></a>,
<a href="#pdf-coroutine.wrap"><code>coroutine.wrap</code></a> does not catch errors;
any error is propagated to the caller.
As an example of how coroutines work,
consider the following code:
function foo (a)
print("foo", a)
return coroutine.yield(2*a)
co = coroutine.create(function (a,b)
print("co-body", a, b)
local r = foo(a+1)
print("co-body", r)
local r, s = coroutine.yield(a+b, a-b)
print("co-body", r, s)
return b, "end"
print("main", coroutine.resume(co, 1, 10))
print("main", coroutine.resume(co, "r"))
print("main", coroutine.resume(co, "x", "y"))
print("main", coroutine.resume(co, "x", "y"))
When you run it, it produces the following output:
co-body 1 10
foo 2
main true 4
co-body r
main true 11 -9
co-body x y
main true 10 end
main false cannot resume dead coroutine
You can also create and manipulate coroutines through the C API:
see functions <a href="#lua_newthread"><code>lua_newthread</code></a>, <a href="#lua_resume"><code>lua_resume</code></a>,
and <a href="#lua_yield"><code>lua_yield</code></a>.
<h1>3 &ndash; <a name="3">The Language</a></h1>
This section describes the lexis, the syntax, and the semantics of Lua.
In other words,
this section describes
which tokens are valid,
how they can be combined,
and what their combinations mean.
Language constructs will be explained using the usual extended BNF notation,
in which
{<em>a</em>}&nbsp;means&nbsp;0 or more <em>a</em>'s, and
[<em>a</em>]&nbsp;means an optional <em>a</em>.
Non-terminals are shown like non-terminal,
keywords are shown like <b>kword</b>,
and other terminal symbols are shown like &lsquo;<b>=</b>&rsquo;.
The complete syntax of Lua can be found in <a href="#9">&sect;9</a>
at the end of this manual.
<h2>3.1 &ndash; <a name="3.1">Lexical Conventions</a></h2>
Lua is a free-form language.
It ignores spaces (including new lines) and comments
between lexical elements (tokens),
except as delimiters between names and keywords.
(also called <em>identifiers</em>)
in Lua can be any string of letters,
digits, and underscores,
not beginning with a digit and
not being a reserved word.
Identifiers are used to name variables, table fields, and labels.
The following <em>keywords</em> are reserved
and cannot be used as names:
and break do else elseif end
false for function goto if in
local nil not or repeat return
then true until while
Lua is a case-sensitive language:
<code>and</code> is a reserved word, but <code>And</code> and <code>AND</code>
are two different, valid names.
As a convention,
programs should avoid creating
names that start with an underscore followed by
one or more uppercase letters (such as <a href="#pdf-_VERSION"><code>_VERSION</code></a>).
The following strings denote other tokens:
+ - * / % ^ #
&amp; ~ | &lt;&lt; &gt;&gt; //
== ~= &lt;= &gt;= &lt; &gt; =
( ) { } [ ] ::
; : , . .. ...
<em>Literal strings</em>
can be delimited by matching single or double quotes,
and can contain the following C-like escape sequences:
'<code>\a</code>' (bell),
'<code>\b</code>' (backspace),
'<code>\f</code>' (form feed),
'<code>\n</code>' (newline),
'<code>\r</code>' (carriage return),
'<code>\t</code>' (horizontal tab),
'<code>\v</code>' (vertical tab),
'<code>\\</code>' (backslash),
'<code>\"</code>' (quotation mark [double quote]),
and '<code>\'</code>' (apostrophe [single quote]).
A backslash followed by a real newline
results in a newline in the string.
The escape sequence '<code>\z</code>' skips the following span
of white-space characters,
including line breaks;
it is particularly useful to break and indent a long literal string
into multiple lines without adding the newlines and spaces
into the string contents.
Strings in Lua can contain any 8-bit value, including embedded zeros,
which can be specified as '<code>\0</code>'.
More generally,
we can specify any byte in a literal string by its numeric value.
This can be done
with the escape sequence <code>\x<em>XX</em></code>,
where <em>XX</em> is a sequence of exactly two hexadecimal digits,
or with the escape sequence <code>\<em>ddd</em></code>,
where <em>ddd</em> is a sequence of up to three decimal digits.
(Note that if a decimal escape sequence is to be followed by a digit,
it must be expressed using exactly three digits.)
The UTF-8 encoding of a Unicode character
can be inserted in a literal string with
the escape sequence <code>\u{<em>XXX</em>}</code>
(note the mandatory enclosing brackets),
where <em>XXX</em> is a sequence of one or more hexadecimal digits
representing the character code point.
Literal strings can also be defined using a long format
enclosed by <em>long brackets</em>.
We define an <em>opening long bracket of level <em>n</em></em> as an opening
square bracket followed by <em>n</em> equal signs followed by another
opening square bracket.
So, an opening long bracket of level&nbsp;0 is written as <code>[[</code>,
an opening long bracket of level&nbsp;1 is written as <code>[=[</code>,
and so on.
A <em>closing long bracket</em> is defined similarly;
for instance,
a closing long bracket of level&nbsp;4 is written as <code>]====]</code>.
A <em>long literal</em> starts with an opening long bracket of any level and
ends at the first closing long bracket of the same level.
It can contain any text except a closing bracket of the same level.
Literals in this bracketed form can run for several lines,
do not interpret any escape sequences,
and ignore long brackets of any other level.
Any kind of end-of-line sequence
(carriage return, newline, carriage return followed by newline,
or newline followed by carriage return)
is converted to a simple newline.
Any byte in a literal string not
explicitly affected by the previous rules represents itself.
However, Lua opens files for parsing in text mode,
and the system file functions may have problems with
some control characters.
So, it is safer to represent
non-text data as a quoted literal with
explicit escape sequences for the non-text characters.
For convenience,
when the opening long bracket is immediately followed by a newline,
the newline is not included in the string.
As an example, in a system using ASCII
(in which '<code>a</code>' is coded as&nbsp;97,
newline is coded as&nbsp;10, and '<code>1</code>' is coded as&nbsp;49),
the five literal strings below denote the same string:
a = 'alo\n123"'
a = "alo\n123\""
a = '\97lo\10\04923"'
a = [[alo
a = [==[
A <em>numeric constant</em> (or <em>numeral</em>)
can be written with an optional fractional part
and an optional decimal exponent,
marked by a letter '<code>e</code>' or '<code>E</code>'.
Lua also accepts hexadecimal constants,
which start with <code>0x</code> or <code>0X</code>.
Hexadecimal constants also accept an optional fractional part
plus an optional binary exponent,
marked by a letter '<code>p</code>' or '<code>P</code>'.
A numeric constant with a radix point or an exponent
denotes a float;
if its value fits in an integer,
it denotes an integer.
Examples of valid integer constants are
3 345 0xff 0xBEBADA
Examples of valid float constants are
3.0 3.1416 314.16e-2 0.31416E1 34e1
0x0.1E 0xA23p-4 0X1.921FB54442D18P+1
A <em>comment</em> starts with a double hyphen (<code>--</code>)
anywhere outside a string.
If the text immediately after <code>--</code> is not an opening long bracket,
the comment is a <em>short comment</em>,
which runs until the end of the line.
Otherwise, it is a <em>long comment</em>,
which runs until the corresponding closing long bracket.
Long comments are frequently used to disable code temporarily.
<h2>3.2 &ndash; <a name="3.2">Variables</a></h2>
Variables are places that store values.
There are three kinds of variables in Lua:
global variables, local variables, and table fields.
A single name can denote a global variable or a local variable
(or a function's formal parameter,
which is a particular kind of local variable):
var ::= Name
Name denotes identifiers, as defined in <a href="#3.1">&sect;3.1</a>.
Any variable name is assumed to be global unless explicitly declared
as a local (see <a href="#3.3.7">&sect;3.3.7</a>).
Local variables are <em>lexically scoped</em>:
local variables can be freely accessed by functions
defined inside their scope (see <a href="#3.5">&sect;3.5</a>).
Before the first assignment to a variable, its value is <b>nil</b>.
Square brackets are used to index a table:
var ::= prefixexp &lsquo;<b>[</b>&rsquo; exp &lsquo;<b>]</b>&rsquo;
The meaning of accesses to table fields can be changed via metatables.
An access to an indexed variable <code>t[i]</code> is equivalent to
a call <code>gettable_event(t,i)</code>.
(See <a href="#2.4">&sect;2.4</a> for a complete description of the
<code>gettable_event</code> function.
This function is not defined or callable in Lua.
We use it here only for explanatory purposes.)
The syntax <code>var.Name</code> is just syntactic sugar for
var ::= prefixexp &lsquo;<b>.</b>&rsquo; Name
An access to a global variable <code>x</code>
is equivalent to <code>_ENV.x</code>.
Due to the way that chunks are compiled,
<code>_ENV</code> is never a global name (see <a href="#2.2">&sect;2.2</a>).
<h2>3.3 &ndash; <a name="3.3">Statements</a></h2>
Lua supports an almost conventional set of statements,
similar to those in Pascal or C.
This set includes
assignments, control structures, function calls,
and variable declarations.
<h3>3.3.1 &ndash; <a name="3.3.1">Blocks</a></h3>
A block is a list of statements,
which are executed sequentially:
block ::= {stat}
Lua has <em>empty statements</em>
that allow you to separate statements with semicolons,
start a block with a semicolon
or write two semicolons in sequence:
stat ::= &lsquo;<b>;</b>&rsquo;
Function calls and assignments
can start with an open parenthesis.
This possibility leads to an ambiguity in Lua's grammar.
Consider the following fragment:
a = b + c
(print or io.write)('done')
The grammar could see it in two ways:
a = b + c(print or io.write)('done')
a = b + c; (print or io.write)('done')
The current parser always sees such constructions
in the first way,
interpreting the open parenthesis
as the start of the arguments to a call.
To avoid this ambiguity,
it is a good practice to always precede with a semicolon
statements that start with a parenthesis:
;(print or io.write)('done')
A block can be explicitly delimited to produce a single statement:
stat ::= <b>do</b> block <b>end</b>
Explicit blocks are useful
to control the scope of variable declarations.
Explicit blocks are also sometimes used to
add a <b>return</b> statement in the middle
of another block (see <a href="#3.3.4">&sect;3.3.4</a>).
<h3>3.3.2 &ndash; <a name="3.3.2">Chunks</a></h3>
The unit of compilation of Lua is called a <em>chunk</em>.
a chunk is simply a block:
chunk ::= block
Lua handles a chunk as the body of an anonymous function
with a variable number of arguments
(see <a href="#3.4.11">&sect;3.4.11</a>).
As such, chunks can define local variables,
receive arguments, and return values.
Moreover, such anonymous function is compiled as in the
scope of an external local variable called <code>_ENV</code> (see <a href="#2.2">&sect;2.2</a>).
The resulting function always has <code>_ENV</code> as its only upvalue,
even if it does not use that variable.
A chunk can be stored in a file or in a string inside the host program.
To execute a chunk,
Lua first <em>loads</em> it,
precompiling the chunk's code into instructions for a virtual machine,
and then Lua executes the compiled code
with an interpreter for the virtual machine.
Chunks can also be precompiled into binary form;
see program <code>luac</code> and function <a href="#pdf-string.dump"><code>string.dump</code></a> for details.
Programs in source and compiled forms are interchangeable;
Lua automatically detects the file type and acts accordingly (see <a href="#pdf-load"><code>load</code></a>).
<h3>3.3.3 &ndash; <a name="3.3.3">Assignment</a></h3>
Lua allows multiple assignments.
Therefore, the syntax for assignment
defines a list of variables on the left side
and a list of expressions on the right side.
The elements in both lists are separated by commas:
stat ::= varlist &lsquo;<b>=</b>&rsquo; explist
varlist ::= var {&lsquo;<b>,</b>&rsquo; var}
explist ::= exp {&lsquo;<b>,</b>&rsquo; exp}
Expressions are discussed in <a href="#3.4">&sect;3.4</a>.
Before the assignment,
the list of values is <em>adjusted</em> to the length of
the list of variables.
If there are more values than needed,
the excess values are thrown away.
If there are fewer values than needed,
the list is extended with as many <b>nil</b>'s as needed.
If the list of expressions ends with a function call,
then all values returned by that call enter the list of values,
before the adjustment
(except when the call is enclosed in parentheses; see <a href="#3.4">&sect;3.4</a>).
The assignment statement first evaluates all its expressions
and only then the assignments are performed.
Thus the code
i = 3
i, a[i] = i+1, 20
sets <code>a[3]</code> to 20, without affecting <code>a[4]</code>
because the <code>i</code> in <code>a[i]</code> is evaluated (to 3)
before it is assigned&nbsp;4.
Similarly, the line
x, y = y, x
exchanges the values of <code>x</code> and <code>y</code>,
x, y, z = y, z, x
cyclically permutes the values of <code>x</code>, <code>y</code>, and <code>z</code>.
The meaning of assignments to global variables
and table fields can be changed via metatables.
An assignment to an indexed variable <code>t[i] = val</code> is equivalent to
(See <a href="#2.4">&sect;2.4</a> for a complete description of the
<code>settable_event</code> function.
This function is not defined or callable in Lua.
We use it here only for explanatory purposes.)
An assignment to a global name <code>x = val</code>
is equivalent to the assignment
<code>_ENV.x = val</code> (see <a href="#2.2">&sect;2.2</a>).
<h3>3.3.4 &ndash; <a name="3.3.4">Control Structures</a></h3><p>
The control structures
<b>if</b>, <b>while</b>, and <b>repeat</b> have the usual meaning and
familiar syntax:
stat ::= <b>while</b> exp <b>do</b> block <b>end</b>
stat ::= <b>repeat</b> block <b>until</b> exp
stat ::= <b>if</b> exp <b>then</b> block {<b>elseif</b> exp <b>then</b> block} [<b>else</b> block] <b>end</b>
Lua also has a <b>for</b> statement, in two flavors (see <a href="#3.3.5">&sect;3.3.5</a>).
The condition expression of a
control structure can return any value.
Both <b>false</b> and <b>nil</b> are considered false.
All values different from <b>nil</b> and <b>false</b> are considered true
(in particular, the number 0 and the empty string are also true).
In the <b>repeat</b>&ndash;<b>until</b> loop,
the inner block does not end at the <b>until</b> keyword,
but only after the condition.
So, the condition can refer to local variables
declared inside the loop block.
The <b>goto</b> statement transfers the program control to a label.
For syntactical reasons,
labels in Lua are considered statements too:
stat ::= <b>goto</b> Name
stat ::= label
label ::= &lsquo;<b>::</b>&rsquo; Name &lsquo;<b>::</b>&rsquo;
A label is visible in the entire block where it is defined,
inside nested blocks where a label with the same name is defined and
inside nested functions.
A goto may jump to any visible label as long as it does not
enter into the scope of a local variable.
Labels and empty statements are called <em>void statements</em>,
as they perform no actions.
The <b>break</b> statement terminates the execution of a
<b>while</b>, <b>repeat</b>, or <b>for</b> loop,
skipping to the next statement after the loop:
stat ::= <b>break</b>
A <b>break</b> ends the innermost enclosing loop.
The <b>return</b> statement is used to return values
from a function or a chunk
(which is an anonymous function).
Functions can return more than one value,
so the syntax for the <b>return</b> statement is
stat ::= <b>return</b> [explist] [&lsquo;<b>;</b>&rsquo;]
The <b>return</b> statement can only be written
as the last statement of a block.
If it is really necessary to <b>return</b> in the middle of a block,
then an explicit inner block can be used,
as in the idiom <code>do return end</code>,
because now <b>return</b> is the last statement in its (inner) block.
<h3>3.3.5 &ndash; <a name="3.3.5">For Statement</a></h3>
The <b>for</b> statement has two forms:
one numerical and one generic.
The numerical <b>for</b> loop repeats a block of code while a
control variable runs through an arithmetic progression.
It has the following syntax:
stat ::= <b>for</b> Name &lsquo;<b>=</b>&rsquo; exp &lsquo;<b>,</b>&rsquo; exp [&lsquo;<b>,</b>&rsquo; exp] <b>do</b> block <b>end</b>
The <em>block</em> is repeated for <em>name</em> starting at the value of
the first <em>exp</em>, until it passes the second <em>exp</em> by steps of the
third <em>exp</em>.
More precisely, a <b>for</b> statement like
for v = <em>e1</em>, <em>e2</em>, <em>e3</em> do <em>block</em> end
is equivalent to the code:
local <em>var</em>, <em>limit</em>, <em>step</em> = tonumber(<em>e1</em>), tonumber(<em>e2</em>), tonumber(<em>e3</em>)
if not (<em>var</em> and <em>limit</em> and <em>step</em>) then error() end
<em>var</em> = <em>var</em> - <em>step</em>
while true do
<em>var</em> = <em>var</em> + <em>step</em>
if (<em>step</em> &gt;= 0 and <em>var</em> &gt; <em>limit</em>) or (<em>step</em> &lt; 0 and <em>var</em> &lt; <em>limit</em>) then
local v = <em>var</em>
Note the following:
All three control expressions are evaluated only once,
before the loop starts.
They must all result in numbers.
<code><em>var</em></code>, <code><em>limit</em></code>, and <code><em>step</em></code> are invisible variables.
The names shown here are for explanatory purposes only.
If the third expression (the step) is absent,
then a step of&nbsp;1 is used.
You can use <b>break</b> and <b>goto</b> to exit a <b>for</b> loop.
The loop variable <code>v</code> is local to the loop body.
If you need its value after the loop,
assign it to another variable before exiting the loop.
The generic <b>for</b> statement works over functions,
called <em>iterators</em>.
On each iteration, the iterator function is called to produce a new value,
stopping when this new value is <b>nil</b>.
The generic <b>for</b> loop has the following syntax:
stat ::= <b>for</b> namelist <b>in</b> explist <b>do</b> block <b>end</b>
namelist ::= Name {&lsquo;<b>,</b>&rsquo; Name}
A <b>for</b> statement like
for <em>var_1</em>, &middot;&middot;&middot;, <em>var_n</em> in <em>explist</em> do <em>block</em> end
is equivalent to the code:
local <em>f</em>, <em>s</em>, <em>var</em> = <em>explist</em>
while true do
local <em>var_1</em>, &middot;&middot;&middot;, <em>var_n</em> = <em>f</em>(<em>s</em>, <em>var</em>)
if <em>var_1</em> == nil then break end
<em>var</em> = <em>var_1</em>
Note the following:
<code><em>explist</em></code> is evaluated only once.
Its results are an <em>iterator</em> function,
a <em>state</em>,
and an initial value for the first <em>iterator variable</em>.
<code><em>f</em></code>, <code><em>s</em></code>, and <code><em>var</em></code> are invisible variables.
The names are here for explanatory purposes only.
You can use <b>break</b> to exit a <b>for</b> loop.
The loop variables <code><em>var_i</em></code> are local to the loop;
you cannot use their values after the <b>for</b> ends.
If you need these values,
then assign them to other variables before breaking or exiting the loop.
<h3>3.3.6 &ndash; <a name="3.3.6">Function Calls as Statements</a></h3><p>
To allow possible side-effects,
function calls can be executed as statements:
stat ::= functioncall
In this case, all returned values are thrown away.
Function calls are explained in <a href="#3.4.10">&sect;3.4.10</a>.
<h3>3.3.7 &ndash; <a name="3.3.7">Local Declarations</a></h3><p>
Local variables can be declared anywhere inside a block.
The declaration can include an initial assignment:
stat ::= <b>local</b> namelist [&lsquo;<b>=</b>&rsquo; explist]
If present, an initial assignment has the same semantics
of a multiple assignment (see <a href="#3.3.3">&sect;3.3.3</a>).
Otherwise, all variables are initialized with <b>nil</b>.
A chunk is also a block (see <a href="#3.3.2">&sect;3.3.2</a>),
and so local variables can be declared in a chunk outside any explicit block.
The visibility rules for local variables are explained in <a href="#3.5">&sect;3.5</a>.
<h2>3.4 &ndash; <a name="3.4">Expressions</a></h2>
The basic expressions in Lua are the following:
exp ::= prefixexp
exp ::= <b>nil</b> | <b>false</b> | <b>true</b>
exp ::= Numeral
exp ::= LiteralString
exp ::= functiondef
exp ::= tableconstructor
exp ::= &lsquo;<b>...</b>&rsquo;
exp ::= exp binop exp
exp ::= unop exp
prefixexp ::= var | functioncall | &lsquo;<b>(</b>&rsquo; exp &lsquo;<b>)</b>&rsquo;
Numerals and literal strings are explained in <a href="#3.1">&sect;3.1</a>;
variables are explained in <a href="#3.2">&sect;3.2</a>;
function definitions are explained in <a href="#3.4.11">&sect;3.4.11</a>;
function calls are explained in <a href="#3.4.10">&sect;3.4.10</a>;
table constructors are explained in <a href="#3.4.9">&sect;3.4.9</a>.
Vararg expressions,
denoted by three dots ('<code>...</code>'), can only be used when
directly inside a vararg function;
they are explained in <a href="#3.4.11">&sect;3.4.11</a>.
Binary operators comprise arithmetic operators (see <a href="#3.4.1">&sect;3.4.1</a>),
bitwise operators (see <a href="#3.4.2">&sect;3.4.2</a>),
relational operators (see <a href="#3.4.4">&sect;3.4.4</a>), logical operators (see <a href="#3.4.5">&sect;3.4.5</a>),
and the concatenation operator (see <a href="#3.4.6">&sect;3.4.6</a>).
Unary operators comprise the unary minus (see <a href="#3.4.1">&sect;3.4.1</a>),
the unary bitwise NOT (see <a href="#3.4.2">&sect;3.4.2</a>),
the unary logical <b>not</b> (see <a href="#3.4.5">&sect;3.4.5</a>),
and the unary <em>length operator</em> (see <a href="#3.4.7">&sect;3.4.7</a>).
Both function calls and vararg expressions can result in multiple values.
If a function call is used as a statement (see <a href="#3.3.6">&sect;3.3.6</a>),
then its return list is adjusted to zero elements,
thus discarding all returned values.
If an expression is used as the last (or the only) element
of a list of expressions,
then no adjustment is made
(unless the expression is enclosed in parentheses).
In all other contexts,
Lua adjusts the result list to one element,
either discarding all values except the first one
or adding a single <b>nil</b> if there are no values.
Here are some examples:
f() -- adjusted to 0 results
g(f(), x) -- f() is adjusted to 1 result
g(x, f()) -- g gets x plus all results from f()
a,b,c = f(), x -- f() is adjusted to 1 result (c gets nil)
a,b = ... -- a gets the first vararg parameter, b gets
-- the second (both a and b can get nil if there
-- is no corresponding vararg parameter)
a,b,c = x, f() -- f() is adjusted to 2 results
a,b,c = f() -- f() is adjusted to 3 results
return f() -- returns all results from f()
return ... -- returns all received vararg parameters
return x,y,f() -- returns x, y, and all results from f()
{f()} -- creates a list with all results from f()
{...} -- creates a list with all vararg parameters
{f(), nil} -- f() is adjusted to 1 result
Any expression enclosed in parentheses always results in only one value.
<code>(f(x,y,z))</code> is always a single value,
even if <code>f</code> returns several values.
(The value of <code>(f(x,y,z))</code> is the first value returned by <code>f</code>
or <b>nil</b> if <code>f</code> does not return any values.)
<h3>3.4.1 &ndash; <a name="3.4.1">Arithmetic Operators</a></h3><p>
Lua supports the following arithmetic operators:
<li><b><code>+</code>: </b>addition</li>
<li><b><code>-</code>: </b>subtraction</li>
<li><b><code>*</code>: </b>multiplication</li>
<li><b><code>/</code>: </b>float division</li>
<li><b><code>//</code>: </b>floor division</li>
<li><b><code>%</code>: </b>modulo</li>
<li><b><code>^</code>: </b>exponentiation</li>
<li><b><code>-</code>: </b>unary minus</li>
With the exception of exponentiation and float division,
the arithmetic operators work as follows:
If both operands are integers,
the operation is performed over integers and the result is an integer.
Otherwise, if both operands are numbers
or strings that can be converted to
numbers (see <a href="#3.4.3">&sect;3.4.3</a>),
then they are converted to floats,
the operation is performed following the usual rules
for floating-point arithmetic
(usually the IEEE 754 standard),
and the result is a float.
Exponentiation and float division (<code>/</code>)
always convert their operands to floats
and the result is always a float.
Exponentiation uses the ISO&nbsp;C function <code>pow</code>,
so that it works for non-integer exponents too.
Floor division (<code>//</code>) is a division
that rounds the quotient towards minus infinity,
that is, the floor of the division of its operands.
Modulo is defined as the remainder of a division
that rounds the quotient towards minus infinity (floor division).
In case of overflows in integer arithmetic,
all operations <em>wrap around</em>,
according to the usual rules of two-complement arithmetic.
(In other words,
they return the unique representable integer
that is equal modulo <em>2<sup>64</sup></em> to the mathematical result.)
<h3>3.4.2 &ndash; <a name="3.4.2">Bitwise Operators</a></h3><p>
Lua supports the following bitwise operators:
<li><b><code>&amp;</code>: </b>bitwise AND</li>
<li><b><code>&#124;</code>: </b>bitwise OR</li>
<li><b><code>~</code>: </b>bitwise exclusive OR</li>
<li><b><code>&gt;&gt;</code>: </b>right shift</li>
<li><b><code>&lt;&lt;</code>: </b>left shift</li>
<li><b><code>~</code>: </b>unary bitwise NOT</li>
All bitwise operations convert its operands to integers
(see <a href="#3.4.3">&sect;3.4.3</a>),
operate on all bits of those integers,
and result in an integer.
Both right and left shifts fill the vacant bits with zeros.
Negative displacements shift to the other direction;
displacements with absolute values equal to or higher than
the number of bits in an integer
result in zero (as all bits are shifted out).
<h3>3.4.3 &ndash; <a name="3.4.3">Coercions and Conversions</a></h3><p>
Lua provides some automatic conversions between some
types and representations at run time.
Bitwise operators always convert float operands to integers.
Exponentiation and float division
always convert integer operands to floats.
All other arithmetic operations applied to mixed numbers
(integers and floats) convert the integer operand to a float;
this is called the <em>usual rule</em>.
The C API also converts both integers to floats and
floats to integers, as needed.
Moreover, string concatenation accepts numbers as arguments,
besides strings.
Lua also converts strings to numbers,
whenever a number is expected.
In a conversion from integer to float,
if the integer value has an exact representation as a float,
that is the result.
the conversion gets the nearest higher or
the nearest lower representable value.
This kind of conversion never fails.
The conversion from float to integer
checks whether the float has an exact representation as an integer
(that is, the float has an integral value and
it is in the range of integer representation).
If it does, that representation is the result.
Otherwise, the conversion fails.
The conversion from strings to numbers goes as follows:
First, the string is converted to an integer or a float,
following its syntax and the rules of the Lua lexer.
(The string may have also leading and trailing spaces and a sign.)
Then, the resulting number (float or integer)
is converted to the type (float or integer) required by the context
(e.g., the operation that forced the conversion).
All conversions from strings to numbers
accept both a dot and the current locale mark
as the radix character.
(The Lua lexer, however, accepts only a dot.)
The conversion from numbers to strings uses a
non-specified human-readable format.
For complete control over how numbers are converted to strings,
use the <code>format</code> function from the string library
(see <a href="#pdf-string.format"><code>string.format</code></a>).
<h3>3.4.4 &ndash; <a name="3.4.4">Relational Operators</a></h3><p>
Lua supports the following relational operators:
<li><b><code>==</code>: </b>equality</li>
<li><b><code>~=</code>: </b>inequality</li>
<li><b><code>&lt;</code>: </b>less than</li>
<li><b><code>&gt;</code>: </b>greater than</li>
<li><b><code>&lt;=</code>: </b>less or equal</li>
<li><b><code>&gt;=</code>: </b>greater or equal</li>
These operators always result in <b>false</b> or <b>true</b>.
Equality (<code>==</code>) first compares the type of its operands.
If the types are different, then the result is <b>false</b>.
Otherwise, the values of the operands are compared.
Strings are compared in the obvious way.
Numbers are equal if they denote the same mathematical value.
Tables, userdata, and threads
are compared by reference:
two objects are considered equal only if they are the same object.
Every time you create a new object
(a table, userdata, or thread),
this new object is different from any previously existing object.
Closures with the same reference are always equal.
Closures with any detectable difference
(different behavior, different definition) are always different.
You can change the way that Lua compares tables and userdata
by using the "eq" metamethod (see <a href="#2.4">&sect;2.4</a>).
Equality comparisons do not convert strings to numbers
or vice versa.
Thus, <code>"0"==0</code> evaluates to <b>false</b>,
and <code>t[0]</code> and <code>t["0"]</code> denote different
entries in a table.
The operator <code>~=</code> is exactly the negation of equality (<code>==</code>).
The order operators work as follows.
If both arguments are numbers,
then they are compared according to their mathematical values
(regardless of their subtypes).
Otherwise, if both arguments are strings,
then their values are compared according to the current locale.
Otherwise, Lua tries to call the "lt" or the "le"
metamethod (see <a href="#2.4">&sect;2.4</a>).
A comparison <code>a &gt; b</code> is translated to <code>b &lt; a</code>
and <code>a &gt;= b</code> is translated to <code>b &lt;= a</code>.
Following the IEEE 754 standard,
NaN is considered neither smaller than,
nor equal to, nor greater than any value (including itself).
<h3>3.4.5 &ndash; <a name="3.4.5">Logical Operators</a></h3><p>
The logical operators in Lua are
<b>and</b>, <b>or</b>, and <b>not</b>.
Like the control structures (see <a href="#3.3.4">&sect;3.3.4</a>),
all logical operators consider both <b>false</b> and <b>nil</b> as false
and anything else as true.
The negation operator <b>not</b> always returns <b>false</b> or <b>true</b>.
The conjunction operator <b>and</b> returns its first argument
if this value is <b>false</b> or <b>nil</b>;
otherwise, <b>and</b> returns its second argument.
The disjunction operator <b>or</b> returns its first argument
if this value is different from <b>nil</b> and <b>false</b>;
otherwise, <b>or</b> returns its second argument.
Both <b>and</b> and <b>or</b> use short-circuit evaluation;
that is,
the second operand is evaluated only if necessary.
Here are some examples:
10 or 20 --&gt; 10
10 or error() --&gt; 10
nil or "a" --&gt; "a"
nil and 10 --&gt; nil
false and error() --&gt; false
false and nil --&gt; false
false or nil --&gt; nil
10 and 20 --&gt; 20
(In this manual,
<code>--&gt;</code> indicates the result of the preceding expression.)
<h3>3.4.6 &ndash; <a name="3.4.6">Concatenation</a></h3><p>
The string concatenation operator in Lua is
denoted by two dots ('<code>..</code>').
If both operands are strings or numbers, then they are converted to
strings according to the rules described in <a href="#3.4.3">&sect;3.4.3</a>.
Otherwise, the <code>__concat</code> metamethod is called (see <a href="#2.4">&sect;2.4</a>).
<h3>3.4.7 &ndash; <a name="3.4.7">The Length Operator</a></h3>
The length operator is denoted by the unary prefix operator <code>#</code>.
The length of a string is its number of bytes
(that is, the usual meaning of string length when each
character is one byte).
A program can modify the behavior of the length operator for
any value but strings through the <code>__len</code> metamethod (see <a href="#2.4">&sect;2.4</a>).
Unless a <code>__len</code> metamethod is given,
the length of a table <code>t</code> is only defined if the
table is a <em>sequence</em>,
that is,
the set of its positive numeric keys is equal to <em>{1..n}</em>
for some non-negative integer <em>n</em>.
In that case, <em>n</em> is its length.
Note that a table like
{10, 20, nil, 40}
is not a sequence, because it has the key <code>4</code>
but does not have the key <code>3</code>.
(So, there is no <em>n</em> such that the set <em>{1..n}</em> is equal
to the set of positive numeric keys of that table.)
Note, however, that non-numeric keys do not interfere
with whether a table is a sequence.
<h3>3.4.8 &ndash; <a name="3.4.8">Precedence</a></h3><p>
Operator precedence in Lua follows the table below,
from lower to higher priority:
&lt; &gt; &lt;= &gt;= ~= ==
&lt;&lt; &gt;&gt;
+ -
* / // %
unary operators (not # - ~)
As usual,
you can use parentheses to change the precedences of an expression.
The concatenation ('<code>..</code>') and exponentiation ('<code>^</code>')
operators are right associative.
All other binary operators are left associative.
<h3>3.4.9 &ndash; <a name="3.4.9">Table Constructors</a></h3><p>
Table constructors are expressions that create tables.
Every time a constructor is evaluated, a new table is created.
A constructor can be used to create an empty table
or to create a table and initialize some of its fields.
The general syntax for constructors is
tableconstructor ::= &lsquo;<b>{</b>&rsquo; [fieldlist] &lsquo;<b>}</b>&rsquo;
fieldlist ::= field {fieldsep field} [fieldsep]
field ::= &lsquo;<b>[</b>&rsquo; exp &lsquo;<b>]</b>&rsquo; &lsquo;<b>=</b>&rsquo; exp | Name &lsquo;<b>=</b>&rsquo; exp | exp
fieldsep ::= &lsquo;<b>,</b>&rsquo; | &lsquo;<b>;</b>&rsquo;
Each field of the form <code>[exp1] = exp2</code> adds to the new table an entry
with key <code>exp1</code> and value <code>exp2</code>.
A field of the form <code>name = exp</code> is equivalent to
<code>["name"] = exp</code>.
Finally, fields of the form <code>exp</code> are equivalent to
<code>[i] = exp</code>, where <code>i</code> are consecutive integers
starting with 1.
Fields in the other formats do not affect this counting.
For example,
a = { [f(1)] = g; "x", "y"; x = 1, f(x), [30] = 23; 45 }
is equivalent to
local t = {}
t[f(1)] = g
t[1] = "x" -- 1st exp
t[2] = "y" -- 2nd exp
t.x = 1 -- t["x"] = 1
t[3] = f(x) -- 3rd exp
t[30] = 23
t[4] = 45 -- 4th exp
a = t
The order of the assignments in a constructor is undefined.
(This order would be relevant only when there are repeated keys.)
If the last field in the list has the form <code>exp</code>
and the expression is a function call or a vararg expression,
then all values returned by this expression enter the list consecutively
(see <a href="#3.4.10">&sect;3.4.10</a>).
The field list can have an optional trailing separator,
as a convenience for machine-generated code.
<h3>3.4.10 &ndash; <a name="3.4.10">Function Calls</a></h3><p>
A function call in Lua has the following syntax:
functioncall ::= prefixexp args
In a function call,
first prefixexp and args are evaluated.
If the value of prefixexp has type <em>function</em>,
then this function is called
with the given arguments.
Otherwise, the prefixexp "call" metamethod is called,
having as first parameter the value of prefixexp,
followed by the original call arguments
(see <a href="#2.4">&sect;2.4</a>).
The form
functioncall ::= prefixexp &lsquo;<b>:</b>&rsquo; Name args
can be used to call "methods".
A call <code>v:name(<em>args</em>)</code>
is syntactic sugar for <code>,<em>args</em>)</code>,
except that <code>v</code> is evaluated only once.
Arguments have the following syntax:
args ::= &lsquo;<b>(</b>&rsquo; [explist] &lsquo;<b>)</b>&rsquo;
args ::= tableconstructor
args ::= LiteralString
All argument expressions are evaluated before the call.
A call of the form <code>f{<em>fields</em>}</code> is
syntactic sugar for <code>f({<em>fields</em>})</code>;
that is, the argument list is a single new table.
A call of the form <code>f'<em>string</em>'</code>
(or <code>f"<em>string</em>"</code> or <code>f[[<em>string</em>]]</code>)
is syntactic sugar for <code>f('<em>string</em>')</code>;
that is, the argument list is a single literal string.
A call of the form <code>return <em>functioncall</em></code> is called
a <em>tail call</em>.
Lua implements <em>proper tail calls</em>
(or <em>proper tail recursion</em>):
in a tail call,
the called function reuses the stack entry of the calling function.
Therefore, there is no limit on the number of nested tail calls that
a program can execute.
However, a tail call erases any debug information about the
calling function.
Note that a tail call only happens with a particular syntax,
where the <b>return</b> has one single function call as argument;
this syntax makes the calling function return exactly
the returns of the called function.
So, none of the following examples are tail calls:
return (f(x)) -- results adjusted to 1
return 2 * f(x)
return x, f(x) -- additional results
f(x); return -- results discarded
return x or f(x) -- results adjusted to 1
<h3>3.4.11 &ndash; <a name="3.4.11">Function Definitions</a></h3>
The syntax for function definition is
functiondef ::= <b>function</b> funcbody
funcbody ::= &lsquo;<b>(</b>&rsquo; [parlist] &lsquo;<b>)</b>&rsquo; block <b>end</b>
The following syntactic sugar simplifies function definitions:
stat ::= <b>function</b> funcname funcbody
stat ::= <b>local</b> <b>function</b> Name funcbody
funcname ::= Name {&lsquo;<b>.</b>&rsquo; Name} [&lsquo;<b>:</b>&rsquo; Name]
The statement
function f () <em>body</em> end
translates to
f = function () <em>body</em> end
The statement
function t.a.b.c.f () <em>body</em> end
translates to
t.a.b.c.f = function () <em>body</em> end
The statement
local function f () <em>body</em> end
translates to
local f; f = function () <em>body</em> end
not to
local f = function () <em>body</em> end
(This only makes a difference when the body of the function
contains references to <code>f</code>.)
A function definition is an executable expression,
whose value has type <em>function</em>.
When Lua precompiles a chunk,
all its function bodies are precompiled too.
Then, whenever Lua executes the function definition,
the function is <em>instantiated</em> (or <em>closed</em>).
This function instance (or <em>closure</em>)
is the final value of the expression.
Parameters act as local variables that are
initialized with the argument values:
parlist ::= namelist [&lsquo;<b>,</b>&rsquo; &lsquo;<b>...</b>&rsquo;] | &lsquo;<b>...</b>&rsquo;
When a function is called,
the list of arguments is adjusted to
the length of the list of parameters,
unless the function is a <em>vararg function</em>,
which is indicated by three dots ('<code>...</code>')
at the end of its parameter list.
A vararg function does not adjust its argument list;
instead, it collects all extra arguments and supplies them
to the function through a <em>vararg expression</em>,
which is also written as three dots.
The value of this expression is a list of all actual extra arguments,
similar to a function with multiple results.
If a vararg expression is used inside another expression
or in the middle of a list of expressions,
then its return list is adjusted to one element.
If the expression is used as the last element of a list of expressions,
then no adjustment is made
(unless that last expression is enclosed in parentheses).
As an example, consider the following definitions:
function f(a, b) end
function g(a, b, ...) end
function r() return 1,2,3 end
Then, we have the following mapping from arguments to parameters and
to the vararg expression:
f(3) a=3, b=nil
f(3, 4) a=3, b=4
f(3, 4, 5) a=3, b=4
f(r(), 10) a=1, b=10
f(r()) a=1, b=2
g(3) a=3, b=nil, ... --&gt; (nothing)
g(3, 4) a=3, b=4, ... --&gt; (nothing)
g(3, 4, 5, 8) a=3, b=4, ... --&gt; 5 8
g(5, r()) a=5, b=1, ... --&gt; 2 3
Results are returned using the <b>return</b> statement (see <a href="#3.3.4">&sect;3.3.4</a>).
If control reaches the end of a function
without encountering a <b>return</b> statement,
then the function returns with no results.
There is a system-dependent limit on the number of values
that a function may return.
This limit is guaranteed to be larger than 1000.
The <em>colon</em> syntax
is used for defining <em>methods</em>,
that is, functions that have an implicit extra parameter <code>self</code>.
Thus, the statement
function t.a.b.c:f (<em>params</em>) <em>body</em> end
is syntactic sugar for
t.a.b.c.f = function (self, <em>params</em>) <em>body</em> end
<h2>3.5 &ndash; <a name="3.5">Visibility Rules</a></h2>
Lua is a lexically scoped language.
The scope of a local variable begins at the first statement after
its declaration and lasts until the last non-void statement
of the innermost block that includes the declaration.
Consider the following example:
x = 10 -- global variable
do -- new block
local x = x -- new 'x', with value 10
print(x) --&gt; 10
x = x+1
do -- another block
local x = x+1 -- another 'x'
print(x) --&gt; 12
print(x) --&gt; 11
print(x) --&gt; 10 (the global one)
Notice that, in a declaration like <code>local x = x</code>,
the new <code>x</code> being declared is not in scope yet,
and so the second <code>x</code> refers to the outside variable.
Because of the lexical scoping rules,
local variables can be freely accessed by functions
defined inside their scope.
A local variable used by an inner function is called
an <em>upvalue</em>, or <em>external local variable</em>,
inside the inner function.
Notice that each execution of a <b>local</b> statement
defines new local variables.
Consider the following example:
a = {}
local x = 20
for i=1,10 do
local y = 0
a[i] = function () y=y+1; return x+y end
The loop creates ten closures
(that is, ten instances of the anonymous function).
Each of these closures uses a different <code>y</code> variable,
while all of them share the same <code>x</code>.
<h1>4 &ndash; <a name="4">The Application Program Interface</a></h1>
This section describes the C&nbsp;API for Lua, that is,
the set of C&nbsp;functions available to the host program to communicate
with Lua.
All API functions and related types and constants
are declared in the header file <a name="pdf-lua.h"><code>lua.h</code></a>.
Even when we use the term "function",
any facility in the API may be provided as a macro instead.
Except where stated otherwise,
all such macros use each of their arguments exactly once
(except for the first argument, which is always a Lua state),
and so do not generate any hidden side-effects.
As in most C&nbsp;libraries,
the Lua API functions do not check their arguments for validity or consistency.
However, you can change this behavior by compiling Lua
with the macro <a name="pdf-LUA_USE_APICHECK"><code>LUA_USE_APICHECK</code></a> defined.
<h2>4.1 &ndash; <a name="4.1">The Stack</a></h2>
Lua uses a <em>virtual stack</em> to pass values to and from C.
Each element in this stack represents a Lua value
(<b>nil</b>, number, string, etc.).
Whenever Lua calls C, the called function gets a new stack,
which is independent of previous stacks and of stacks of
C&nbsp;functions that are still active.
This stack initially contains any arguments to the C&nbsp;function
and it is where the C&nbsp;function pushes its results
to be returned to the caller (see <a href="#lua_CFunction"><code>lua_CFunction</code></a>).
For convenience,
most query operations in the API do not follow a strict stack discipline.
Instead, they can refer to any element in the stack
by using an <em>index</em>:
A positive index represents an absolute stack position
(starting at&nbsp;1);
a negative index represents an offset relative to the top of the stack.
More specifically, if the stack has <em>n</em> elements,
then index&nbsp;1 represents the first element
(that is, the element that was pushed onto the stack first)
index&nbsp;<em>n</em> represents the last element;
index&nbsp;-1 also represents the last element
(that is, the element at the&nbsp;top)
and index <em>-n</em> represents the first element.
<h2>4.2 &ndash; <a name="4.2">Stack Size</a></h2>
When you interact with the Lua API,
you are responsible for ensuring consistency.
In particular,
<em>you are responsible for controlling stack overflow</em>.
You can use the function <a href="#lua_checkstack"><code>lua_checkstack</code></a>
to ensure that the stack has enough space for pushing new elements.
Whenever Lua calls C,
it ensures that the stack has space for
at least <a name="pdf-LUA_MINSTACK"><code>LUA_MINSTACK</code></a> extra slots.
<code>LUA_MINSTACK</code> is defined as 20,
so that usually you do not have to worry about stack space
unless your code has loops pushing elements onto the stack.
When you call a Lua function
without a fixed number of results (see <a href="#lua_call"><code>lua_call</code></a>),
Lua ensures that the stack has enough space for all results,
but it does not ensure any extra space.
So, before pushing anything in the stack after such a call
you should use <a href="#lua_checkstack"><code>lua_checkstack</code></a>.
<h2>4.3 &ndash; <a name="4.3">Valid and Acceptable Indices</a></h2>
Any function in the API that receives stack indices
works only with <em>valid indices</em> or <em>acceptable indices</em>.
A <em>valid index</em> is an index that refers to a
position that stores a modifiable Lua value.
It comprises stack indices between&nbsp;1 and the stack top
(<code>1 &le; abs(index) &le; top</code>)
plus <em>pseudo-indices</em>,
which represent some positions that are accessible to C&nbsp;code
but that are not in the stack.
Pseudo-indices are used to access the registry (see <a href="#4.5">&sect;4.5</a>)
and the upvalues of a C&nbsp;function (see <a href="#4.4">&sect;4.4</a>).
Functions that do not need a specific mutable position,
but only a value (e.g., query functions),
can be called with acceptable indices.
An <em>acceptable index</em> can be any valid index,
but it also can be any positive index after the stack top
within the space allocated for the stack,
that is, indices up to the stack size.
(Note that 0 is never an acceptable index.)
Except when noted otherwise,
functions in the API work with acceptable indices.
Acceptable indices serve to avoid extra tests
against the stack top when querying the stack.
For instance, a C&nbsp;function can query its third argument
without the need to first check whether there is a third argument,
that is, without the need to check whether 3 is a valid index.
For functions that can be called with acceptable indices,
any non-valid index is treated as if it
contains a value of a virtual type <a name="pdf-LUA_TNONE"><code>LUA_TNONE</code></a>,
which behaves like a nil value.
<h2>4.4 &ndash; <a name="4.4">C Closures</a></h2>
When a C&nbsp;function is created,
it is possible to associate some values with it,
thus creating a <em>C&nbsp;closure</em>
(see <a href="#lua_pushcclosure"><code>lua_pushcclosure</code></a>);
these values are called <em>upvalues</em> and are
accessible to the function whenever it is called.
Whenever a C&nbsp;function is called,
its upvalues are located at specific pseudo-indices.
These pseudo-indices are produced by the macro
<a href="#lua_upvalueindex"><code>lua_upvalueindex</code></a>.
The first upvalue associated with a function is at index
<code>lua_upvalueindex(1)</code>, and so on.
Any access to <code>lua_upvalueindex(<em>n</em>)</code>,
where <em>n</em> is greater than the number of upvalues of the
current function
(but not greater than 256,
which is one plus the maximum number of upvalues in a closure),
produces an acceptable but invalid index.
<h2>4.5 &ndash; <a name="4.5">Registry</a></h2>
Lua provides a <em>registry</em>,
a predefined table that can be used by any C&nbsp;code to
store whatever Lua values it needs to store.
The registry table is always located at pseudo-index
<a name="pdf-LUA_REGISTRYINDEX"><code>LUA_REGISTRYINDEX</code></a>.
Any C&nbsp;library can store data into this table,
but it must take care to choose keys
that are different from those used
by other libraries, to avoid collisions.
Typically, you should use as key a string containing your library name,
or a light userdata with the address of a C&nbsp;object in your code,
or any Lua object created by your code.
As with variable names,
string keys starting with an underscore followed by
uppercase letters are reserved for Lua.
The integer keys in the registry are used
by the reference mechanism (see <a href="#luaL_ref"><code>luaL_ref</code></a>)
and by some predefined values.
Therefore, integer keys must not be used for other purposes.
When you create a new Lua state,
its registry comes with some predefined values.
These predefined values are indexed with integer keys
defined as constants in <code>lua.h</code>.
The following constants are defined:
<li><b><a name="pdf-LUA_RIDX_MAINTHREAD"><code>LUA_RIDX_MAINTHREAD</code></a>: </b> At this index the registry has
the main thread of the state.
(The main thread is the one created together with the state.)
<li><b><a name="pdf-LUA_RIDX_GLOBALS"><code>LUA_RIDX_GLOBALS</code></a>: </b> At this index the registry has
the global environment.
<h2>4.6 &ndash; <a name="4.6">Error Handling in C</a></h2>
Internally, Lua uses the C <code>longjmp</code> facility to handle errors.
(Lua will use exceptions if you compile it as C++;
search for <code>LUAI_THROW</code> in the source code for details.)
When Lua faces any error
(such as a memory allocation error, type errors, syntax errors,
and runtime errors)
it <em>raises</em> an error;
that is, it does a long jump.
A <em>protected environment</em> uses <code>setjmp</code>
to set a recovery point;
any error jumps to the most recent active recovery point.
If an error happens outside any protected environment,
Lua calls a <em>panic function</em> (see <a href="#lua_atpanic"><code>lua_atpanic</code></a>)
and then calls <code>abort</code>,
thus exiting the host application.
Your panic function can avoid this exit by
never returning
(e.g., doing a long jump to your own recovery point outside Lua).
The panic function runs as if it were a message handler (see <a href="#2.3">&sect;2.3</a>);
in particular, the error object is at the top of the stack.
However, there is no guarantee about stack space.
To push anything on the stack,
the panic function must first check the available space (see <a href="#4.2">&sect;4.2</a>).
Most functions in the API can raise an error,
for instance due to a memory allocation error.
The documentation for each function indicates whether
it can raise errors.
Inside a C&nbsp;function you can raise an error by calling <a href="#lua_error"><code>lua_error</code></a>.
<h2>4.7 &ndash; <a name="4.7">Handling Yields in C</a></h2>
Internally, Lua uses the C <code>longjmp</code> facility to yield a coroutine.
Therefore, if a C function <code>foo</code> calls an API function
and this API function yields
(directly or indirectly by calling another function that yields),
Lua cannot return to <code>foo</code> any more,
because the <code>longjmp</code> removes its frame from the C stack.
To avoid this kind of problem,
Lua raises an error whenever it tries to yield across an API call,
except for three functions:
<a href="#lua_yieldk"><code>lua_yieldk</code></a>, <a href="#lua_callk"><code>lua_callk</code></a>, and <a href="#lua_pcallk"><code>lua_pcallk</code></a>.
All those functions receive a <em>continuation function</em>
(as a parameter named <code>k</code>) to continue execution after a yield.
We need to set some terminology to explain continuations.
We have a C function called from Lua which we will call
the <em>original function</em>.
This original function then calls one of those three functions in the C API,
which we will call the <em>callee function</em>,
that then yields the current thread.
(This can happen when the callee function is <a href="#lua_yieldk"><code>lua_yieldk</code></a>,
or when the callee function is either <a href="#lua_callk"><code>lua_callk</code></a> or <a href="#lua_pcallk"><code>lua_pcallk</code></a>
and the function called by them yields.)
Suppose the running thread yields while executing the callee function.
After the thread resumes,
it eventually will finish running the callee function.
the callee function cannot return to the original function,
because its frame in the C stack was destroyed by the yield.
Instead, Lua calls a <em>continuation function</em>,
which was given as an argument to the callee function.
As the name implies,
the continuation function should continue the task
of the original function.
As an illustration, consider the following function:
int original_function (lua_State *L) {
... /* code 1 */
status = lua_pcall(L, n, m, h); /* calls Lua */
... /* code 2 */
Now we want to allow
the Lua code being run by <a href="#lua_pcall"><code>lua_pcall</code></a> to yield.
First, we can rewrite our function like here:
int k (lua_State *L, int status, lua_KContext ctx) {
... /* code 2 */
int original_function (lua_State *L) {
... /* code 1 */
return k(L, lua_pcall(L, n, m, h), ctx);
In the above code,
the new function <code>k</code> is a
<em>continuation function</em> (with type <a href="#lua_KFunction"><code>lua_KFunction</code></a>),
which should do all the work that the original function
was doing after calling <a href="#lua_pcall"><code>lua_pcall</code></a>.
Now, we must inform Lua that it must call <code>k</code> if the Lua code
being executed by <a href="#lua_pcall"><code>lua_pcall</code></a> gets interrupted in some way
(errors or yielding),
so we rewrite the code as here,
replacing <a href="#lua_pcall"><code>lua_pcall</code></a> by <a href="#lua_pcallk"><code>lua_pcallk</code></a>:
int original_function (lua_State *L) {
... /* code 1 */
return k(L, lua_pcallk(L, n, m, h, ctx2, k), ctx1);
Note the external, explicit call to the continuation:
Lua will call the continuation only if needed, that is,
in case of errors or resuming after a yield.
If the called function returns normally without ever yielding,
<a href="#lua_pcallk"><code>lua_pcallk</code></a> (and <a href="#lua_callk"><code>lua_callk</code></a>) will also return normally.
(Of course, instead of calling the continuation in that case,
you can do the equivalent work directly inside the original function.)
Besides the Lua state,
the continuation function has two other parameters:
the final status of the call plus the context value (<code>ctx</code>) that
was passed originally to <a href="#lua_pcallk"><code>lua_pcallk</code></a>.
(Lua does not use this context value;
it only passes this value from the original function to the
continuation function.)
For <a href="#lua_pcallk"><code>lua_pcallk</code></a>,
the status is the same value that would be returned by <a href="#lua_pcallk"><code>lua_pcallk</code></a>,
except that it is <a href="#pdf-LUA_YIELD"><code>LUA_YIELD</code></a> when being executed after a yield
(instead of <a href="#pdf-LUA_OK"><code>LUA_OK</code></a>).
For <a href="#lua_yieldk"><code>lua_yieldk</code></a> and <a href="#lua_callk"><code>lua_callk</code></a>,
the status is always <a href="#pdf-LUA_YIELD"><code>LUA_YIELD</code></a> when Lua calls the continuation.
(For these two functions,
Lua will not call the continuation in case of errors,
because they do not handle errors.)
Similarly, when using <a href="#lua_callk"><code>lua_callk</code></a>,
you should call the continuation function
with <a href="#pdf-LUA_OK"><code>LUA_OK</code></a> as the status.
(For <a href="#lua_yieldk"><code>lua_yieldk</code></a>, there is not much point in calling
directly the continuation function,
because <a href="#lua_yieldk"><code>lua_yieldk</code></a> usually does not return.)
Lua treats the continuation function as if it were the original function.
The continuation function receives the same Lua stack
from the original function,
in the same state it would be if the callee function had returned.
(For instance,
after a <a href="#lua_callk"><code>lua_callk</code></a> the function and its arguments are
removed from the stack and replaced by the results from the call.)
It also has the same upvalues.
Whatever it returns is handled by Lua as if it were the return
of the original function.
<h2>4.8 &ndash; <a name="4.8">Functions and Types</a></h2>
Here we list all functions and types from the C&nbsp;API in
alphabetical order.
Each function has an indicator like this:
<span class="apii">[-o, +p, <em>x</em>]</span>
The first field, <code>o</code>,
is how many elements the function pops from the stack.
The second field, <code>p</code>,
is how many elements the function pushes onto the stack.
(Any function always pushes its results after popping its arguments.)
A field in the form <code>x|y</code> means the function can push (or pop)
<code>x</code> or <code>y</code> elements,
depending on the situation;
an interrogation mark '<code>?</code>' means that
we cannot know how many elements the function pops/pushes
by looking only at its arguments
(e.g., they may depend on what is on the stack).
The third field, <code>x</code>,
tells whether the function may raise errors:
'<code>-</code>' means the function never raises any error;
'<code>m</code>' means the function may raise out-of-memory errors
and errors running a <code>__gc</code> metamethod;
'<code>e</code>' means the function may raise any errors
(it can run arbitrary Lua code,
either directly or through metamethods);
'<code>v</code>' means the function may raise an error on purpose.
<hr><h3><a name="lua_absindex"><code>lua_absindex</code></a></h3><p>
<span class="apii">[-0, +0, &ndash;]</span>
<pre>int lua_absindex (lua_State *L, int idx);</pre>
Converts the acceptable index <code>idx</code>
into an equivalent absolute index
(that is, one that does not depend on the stack top).
<hr><h3><a name="lua_Alloc"><code>lua_Alloc</code></a></h3>
<pre>typedef void * (*lua_Alloc) (void *ud,
void *ptr,
size_t osize,
size_t nsize);</pre>
The type of the memory-allocation function used by Lua states.
The allocator function must provide a
functionality similar to <code>realloc</code>,
but not exactly the same.
Its arguments are
<code>ud</code>, an opaque pointer passed to <a href="#lua_newstate"><code>lua_newstate</code></a>;
<code>ptr</code>, a pointer to the block being allocated/reallocated/freed;
<code>osize</code>, the original size of the block or some code about what
is being allocated;
and <code>nsize</code>, the new size of the block.
When <code>ptr</code> is not <code>NULL</code>,
<code>osize</code> is the size of the block pointed by <code>ptr</code>,
that is, the size given when it was allocated or reallocated.
When <code>ptr</code> is <code>NULL</code>,
<code>osize</code> encodes the kind of object that Lua is allocating.
<code>osize</code> is any of
<a href="#pdf-LUA_TSTRING"><code>LUA_TSTRING</code></a>, <a href="#pdf-LUA_TTABLE"><code>LUA_TTABLE</code></a>, <a href="#pdf-LUA_TFUNCTION"><code>LUA_TFUNCTION</code></a>,
<a href="#pdf-LUA_TUSERDATA"><code>LUA_TUSERDATA</code></a>, or <a href="#pdf-LUA_TTHREAD"><code>LUA_TTHREAD</code></a> when (and only when)
Lua is creating a new object of that type.
When <code>osize</code> is some other value,
Lua is allocating memory for something else.
Lua assumes the following behavior from the allocator function:
When <code>nsize</code> is zero,
the allocator must behave like <code>free</code>
and return <code>NULL</code>.
When <code>nsize</code> is not zero,
the allocator must behave like <code>realloc</code>.
The allocator returns <code>NULL</code>
if and only if it cannot fulfill the request.
Lua assumes that the allocator never fails when
<code>osize &gt;= nsize</code>.
Here is a simple implementation for the allocator function.
It is used in the auxiliary library by <a href="#luaL_newstate"><code>luaL_newstate</code></a>.
static void *l_alloc (void *ud, void *ptr, size_t osize,
size_t nsize) {
(void)ud; (void)osize; /* not used */
if (nsize == 0) {
return NULL;
return realloc(ptr, nsize);
Note that Standard&nbsp;C ensures
that <code>free(NULL)</code> has no effect and that
<code>realloc(NULL,size)</code> is equivalent to <code>malloc(size)</code>.
This code assumes that <code>realloc</code> does not fail when shrinking a block.
(Although Standard&nbsp;C does not ensure this behavior,
it seems to be a safe assumption.)
<hr><h3><a name="lua_arith"><code>lua_arith</code></a></h3><p>
<span class="apii">[-(2|1), +1, <em>e</em>]</span>
<pre>void lua_arith (lua_State *L, int op);</pre>
Performs an arithmetic or bitwise operation over the two values
(or one, in the case of negations)
at the top of the stack,
with the value at the top being the second operand,
pops these values, and pushes the result of the operation.
The function follows the semantics of the corresponding Lua operator
(that is, it may call metamethods).
The value of <code>op</code> must be one of the following constants:
<li><b><a name="pdf-LUA_OPADD"><code>LUA_OPADD</code></a>: </b> performs addition (<code>+</code>)</li>
<li><b><a name="pdf-LUA_OPSUB"><code>LUA_OPSUB</code></a>: </b> performs subtraction (<code>-</code>)</li>
<li><b><a name="pdf-LUA_OPMUL"><code>LUA_OPMUL</code></a>: </b> performs multiplication (<code>*</code>)</li>
<li><b><a name="pdf-LUA_OPDIV"><code>LUA_OPDIV</code></a>: </b> performs float division (<code>/</code>)</li>
<li><b><a name="pdf-LUA_OPIDIV"><code>LUA_OPIDIV</code></a>: </b> performs floor division (<code>//</code>)</li>
<li><b><a name="pdf-LUA_OPMOD"><code>LUA_OPMOD</code></a>: </b> performs modulo (<code>%</code>)</li>
<li><b><a name="pdf-LUA_OPPOW"><code>LUA_OPPOW</code></a>: </b> performs exponentiation (<code>^</code>)</li>
<li><b><a name="pdf-LUA_OPUNM"><code>LUA_OPUNM</code></a>: </b> performs mathematical negation (unary <code>-</code>)</li>
<li><b><a name="pdf-LUA_OPBNOT"><code>LUA_OPBNOT</code></a>: </b> performs bitwise NOT (<code>~</code>)</li>
<li><b><a name="pdf-LUA_OPBAND"><code>LUA_OPBAND</code></a>: </b> performs bitwise AND (<code>&amp;</code>)</li>
<li><b><a name="pdf-LUA_OPBOR"><code>LUA_OPBOR</code></a>: </b> performs bitwise OR (<code>|</code>)</li>
<li><b><a name="pdf-LUA_OPBXOR"><code>LUA_OPBXOR</code></a>: </b> performs bitwise exclusive OR (<code>~</code>)</li>
<li><b><a name="pdf-LUA_OPSHL"><code>LUA_OPSHL</code></a>: </b> performs left shift (<code>&lt;&lt;</code>)</li>
<li><b><a name="pdf-LUA_OPSHR"><code>LUA_OPSHR</code></a>: </b> performs right shift (<code>&gt;&gt;</code>)</li>
<hr><h3><a name="lua_atpanic"><code>lua_atpanic</code></a></h3><p>
<span class="apii">[-0, +0, &ndash;]</span>
<pre>lua_CFunction lua_atpanic (lua_State *L, lua_CFunction panicf);</pre>
Sets a new panic function and returns the old one (see <a href="#4.6">&sect;4.6</a>).
<hr><h3><a name="lua_call"><code>lua_call</code></a></h3><p>
<span class="apii">[-(nargs+1), +nresults, <em>e</em>]</span>
<pre>void lua_call (lua_State *L, int nargs, int nresults);</pre>
Calls a function.
To call a function you must use the following protocol:
first, the function to be called is pushed onto the stack;
then, the arguments to the function are pushed
in direct order;
that is, the first argument is pushed first.
Finally you call <a href="#lua_call"><code>lua_call</code></a>;
<code>nargs</code> is the number of arguments that you pushed onto the stack.
All arguments and the function value are popped from the stack
when the function is called.
The function results are pushed onto the stack when the function returns.
The number of results is adjusted to <code>nresults</code>,
unless <code>nresults</code> is <a name="pdf-LUA_MULTRET"><code>LUA_MULTRET</code></a>.
In this case, all results from the function are pushed.
Lua takes care that the returned values fit into the stack space,
but it does not ensure any extra space in the stack.
The function results are pushed onto the stack in direct order
(the first result is pushed first),
so that after the call the last result is on the top of the stack.
Any error inside the called function is propagated upwards
(with a <code>longjmp</code>).
The following example shows how the host program can do the
equivalent to this Lua code:
a = f("how", t.x, 14)
Here it is in&nbsp;C:
lua_getglobal(L, "f"); /* function to be called */
lua_pushliteral(L, "how"); /* 1st argument */
lua_getglobal(L, "t"); /* table to be indexed */
lua_getfield(L, -1, "x"); /* push result of t.x (2nd arg) */
lua_remove(L, -2); /* remove 't' from the stack */
lua_pushinteger(L, 14); /* 3rd argument */
lua_call(L, 3, 1); /* call 'f' with 3 arguments and 1 result */
lua_setglobal(L, "a"); /* set global 'a' */
Note that the code above is <em>balanced</em>:
at its end, the stack is back to its original configuration.
This is considered good programming practice.
<hr><h3><a name="lua_callk"><code>lua_callk</code></a></h3><p>
<span class="apii">[-(nargs + 1), +nresults, <em>e</em>]</span>
<pre>void lua_callk (lua_State *L,
int nargs,
int nresults,
lua_KContext ctx,
lua_KFunction k);</pre>
This function behaves exactly like <a href="#lua_call"><code>lua_call</code></a>,
but allows the called function to yield (see <a href="#4.7">&sect;4.7</a>).
<hr><h3><a name="lua_CFunction"><code>lua_CFunction</code></a></h3>
<pre>typedef int (*lua_CFunction) (lua_State *L);</pre>
Type for C&nbsp;functions.
In order to communicate properly with Lua,
a C&nbsp;function must use the following protocol,
which defines the way parameters and results are passed:
a C&nbsp;function receives its arguments from Lua in its stack