<?xml version="1.0" encoding="latin1" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd">
<chapter>
<header>
<copyright>
<year>2003</year><year>2009</year>
<holder>Ericsson AB. All Rights Reserved.</holder>
</copyright>
<legalnotice>
The contents of this file are subject to the Erlang Public License,
Version 1.1, (the "License"); you may not use this file except in
compliance with the License. You should have received a copy of the
Erlang Public License along with this software. If not, it can be
retrieved online at http://www.erlang.org/.
Software distributed under the License is distributed on an "AS IS"
basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
the License for the specific language governing rights and limitations
under the License.
</legalnotice>
<title>Expressions</title>
<prepared></prepared>
<docno></docno>
<date></date>
<rev></rev>
<file>expressions.xml</file>
</header>
<p>In this chapter, all valid Erlang expressions are listed.
When writing Erlang programs, it is also allowed to use macro-
and record expressions. However, these expressions are expanded
during compilation and are in that sense not true Erlang
expressions. Macro- and record expressions are covered in
separate chapters: <seealso marker="macros">Macros</seealso> and
<seealso marker="records">Records</seealso>.</p>
<section>
<title>Expression Evaluation</title>
<p>All subexpressions are evaluated before an expression itself is
evaluated, unless explicitly stated otherwise. For example,
consider the expression:</p>
<code type="none">
Expr1 + Expr2</code>
<p><c>Expr1</c> and <c>Expr2</c>, which are also expressions, are
evaluated first - in any order - before the addition is
performed.</p>
<p>Many of the operators can only be applied to arguments of a
certain type. For example, arithmetic operators can only be
applied to numbers. An argument of the wrong type will cause
a <c>badarg</c> run-time error.</p>
</section>
<section>
<marker id="term"></marker>
<title>Terms</title>
<p>The simplest form of expression is a term, that is an integer,
float, atom, string, list or tuple.
The return value is the term itself.</p>
</section>
<section>
<title>Variables</title>
<p>A variable is an expression. If a variable is bound to a value,
the return value is this value. Unbound variables are only
allowed in patterns.</p>
<p>Variables start with an uppercase letter or underscore (_)
and may contain alphanumeric characters, underscore and @.
Examples:</p>
<pre>
X
Name1
PhoneNumber
Phone_number
_
_Height</pre>
<p>Variables are bound to values using
<seealso marker="patterns">pattern matching</seealso>. Erlang
uses <em>single assignment</em>, a variable can only be bound
once.</p>
<p>The <em>anonymous variable</em> is denoted by underscore (_) and
can be used when a variable is required but its value can be
ignored. Example:</p>
<pre>
[H|_] = [1,2,3]</pre>
<p>Variables starting with underscore (_), for example
<c>_Height</c>, are normal variables, not anonymous. They are
however ignored by the compiler in the sense that they will not
generate any warnings for unused variables. Example: The following
code</p>
<pre>
member(_, []) ->
[].</pre>
<p>can be rewritten to be more readable:</p>
<pre>
member(Elem, []) ->
[].</pre>
<p>This will however cause a warning for an unused variable
<c>Elem</c>, if the code is compiled with the flag
<c>warn_unused_vars</c> set. Instead, the code can be rewritten
to:</p>
<pre>
member(_Elem, []) ->
[].</pre>
<p>Note that since variables starting with an underscore are
not anonymous, this will match:</p>
<pre>
{_,_} = {1,2}</pre>
<p>But this will fail:</p>
<pre>
{_N,_N} = {1,2}</pre>
<p>The scope for a variable is its function clause.
Variables bound in a branch of an <c>if</c>, <c>case</c>,
or <c>receive</c> expression must be bound in all branches
to have a value outside the expression, otherwise they
will be regarded as 'unsafe' outside the expression.</p>
<p>For the <c>try</c> expression introduced in
Erlang 5.4/OTP-R10B, variable scoping is limited so that
variables bound in the expression are always 'unsafe' outside
the expression. This will be improved.</p>
</section>
<section>
<marker id="pattern"></marker>
<title>Patterns</title>
<p>A pattern has the same structure as a term but may contain
unbound variables. Example:</p>
<pre>
Name1
[H|T]
{error,Reason}</pre>
<p>Patterns are allowed in clause heads, <c>case</c> and
<c>receive</c> expressions, and match expressions.</p>
<section>
<title>Match Operator = in Patterns</title>
<p>If <c>Pattern1</c> and <c>Pattern2</c> are valid patterns,
then the following is also a valid pattern:</p>
<pre>
Pattern1 = Pattern2</pre>
<p>When matched against a term, both <c>Pattern1</c> and
<c>Pattern2</c> will be matched against the term. The idea
behind this feature is to avoid reconstruction of terms.
Example:</p>
<pre>
f({connect,From,To,Number,Options}, To) ->
Signal = {connect,From,To,Number,Options},
...;
f(Signal, To) ->
ignore.</pre>
<p>can instead be written as</p>
<pre>
f({connect,_,To,_,_} = Signal, To) ->
...;
f(Signal, To) ->
ignore.</pre>
</section>
<section>
<title>String Prefix in Patterns</title>
<p>When matching strings, the following is a valid pattern:</p>
<pre>
f("prefix" ++ Str) -> ...</pre>
<p>This is syntactic sugar for the equivalent, but harder to
read</p>
<pre>
f([$p,$r,$e,$f,$i,$x | Str]) -> ...</pre>
</section>
<section>
<title>Expressions in Patterns</title>
<p>An arithmetic expression can be used within a pattern, if
it uses only numeric or bitwise operators, and if its value
can be evaluated to a constant at compile-time. Example:</p>
<pre>
case {Value, Result} of
{?THRESHOLD+1, ok} -> ...</pre>
<p>This feature was added in Erlang 5.0/OTP R7.</p>
</section>
</section>
<section>
<title>Match</title>
<pre>
Expr1 = Expr2</pre>
<p>Matches <c>Expr1</c>, a pattern, against <c>Expr2</c>.
If the matching succeeds, any unbound variable in the pattern
becomes bound and the value of <c>Expr2</c> is returned.</p>
<p>If the matching fails, a <c>badmatch</c> run-time error will
occur.</p>
<p>Examples:</p>
<pre>
1> <input>{A, B} = {answer, 42}.</input>
{answer,42}
2> <input>A.</input>
answer
3> <input>{C, D} = [1, 2].</input>
** exception error: no match of right hand side value [1,2]</pre>
</section>
<section>
<marker id="calls"></marker>
<title>Function Calls</title>
<pre>
ExprF(Expr1,...,ExprN)
ExprM:ExprF(Expr1,...,ExprN)</pre>
<p>In the first form of function calls,
<c>ExprM:ExprF(Expr1,...,ExprN)</c>, each of <c>ExprM</c> and
<c>ExprF</c> must be an atom or an expression that evaluates to
an atom. The function is said to be called by using the
<em>fully qualified function name</em>. This is often referred
to as a <em>remote</em> or <em>external function call</em>.
Example:</p>
<code type="none">
lists:keysearch(Name, 1, List)</code>
<p>In the second form of function calls,
<c>ExprF(Expr1,...,ExprN)</c>, <c>ExprF</c> must be an atom or
evaluate to a fun.</p>
<p>If <c>ExprF</c> is an atom the function is said to be called by
using the <em>implicitly qualified function name</em>. If
<c>ExprF/N</c> is the name of a function explicitly or
automatically imported from module <c>M</c>, then the call is
short for <c>M:ExprF(Expr1,...,ExprN)</c>. Otherwise,
<c>ExprF/N</c> must be a locally defined function. Examples:</p>
<code type="none">
handle(Msg, State)
spawn(m, init, [])</code>
<p>Examples where ExprF is a fun:</p>
<code type="none">
Fun1 = fun(X) -> X+1 end
Fun1(3)
=> 4
Fun2 = {lists,append}
Fun2([1,2], [3,4])
=> [1,2,3,4]
fun lists:append/2([1,2], [3,4])
=> [1,2,3,4]</code>
<p>To avoid possible ambiguities, the fully qualified function
name must be used when calling a function with the same name as
a BIF, and the compiler does not allow defining a function with
the same name as an explicitly imported function.</p>
<p>Note that when calling a local function, there is a difference
between using the implicitly or fully qualified function name, as
the latter always refers to the latest version of the module. See
<seealso marker="code_loading">Compilation and Code Loading</seealso>.</p>
<p>See also the chapter about
<seealso marker="functions#eval">Function Evaluation</seealso>.</p>
</section>
<section>
<title>If</title>
<pre>
if
GuardSeq1 ->
Body1;
...;
GuardSeqN ->
BodyN
end</pre>
<p>The branches of an <c>if</c>-expression are scanned sequentially
until a guard sequence <c>GuardSeq</c> which evaluates to true is
found. Then the corresponding <c>Body</c> (sequence of expressions
separated by ',') is evaluated.</p>
<p>The return value of <c>Body</c> is the return value of
the <c>if</c> expression.</p>
<p>If no guard sequence is true, an <c>if_clause</c> run-time error
will occur. If necessary, the guard expression <c>true</c> can be
used in the last branch, as that guard sequence is always true.</p>
<p>Example:</p>
<pre>
is_greater_than(X, Y) ->
if
X>Y ->
true;
true -> % works as an 'else' branch
false
end</pre>
</section>
<section>
<marker id="case"></marker>
<title>Case</title>
<pre>
case Expr of
Pattern1 [when GuardSeq1] ->
Body1;
...;
PatternN [when GuardSeqN] ->
BodyN
end</pre>
<p>The expression <c>Expr</c> is evaluated and the patterns
<c>Pattern</c> are sequentially matched against the result. If a
match succeeds and the optional guard sequence <c>GuardSeq</c> is
true, the corresponding <c>Body</c> is evaluated.</p>
<p>The return value of <c>Body</c> is the return value of
the <c>case</c> expression.</p>
<p>If there is no matching pattern with a true guard sequence,
a <c>case_clause</c> run-time error will occur.</p>
<p>Example:</p>
<pre>
is_valid_signal(Signal) ->
case Signal of
{signal, _What, _From, _To} ->
true;
{signal, _What, _To} ->
true;
_Else ->
false
end.</pre>
</section>
<section>
<marker id="send"></marker>
<title>Send</title>
<pre>
Expr1 ! Expr2</pre>
<p>Sends the value of <c>Expr2</c> as a message to the process
specified by <c>Expr1</c>. The value of <c>Expr2</c> is also
the return value of the expression.</p>
<p><c>Expr1</c> must evaluate to a pid, a registered name (atom) or
a tuple <c>{Name,Node}</c>, where <c>Name</c> is an atom and
<c>Node</c> a node name, also an atom.</p>
<list type="bulleted">
<item>If <c>Expr1</c> evaluates to a name, but this name is not
registered, a <c>badarg</c> run-time error will occur.</item>
<item>Sending a message to a pid never fails, even if the pid
identifies a non-existing process.</item>
<item>Distributed message sending, that is if <c>Expr1</c>
evaluates to a tuple <c>{Name,Node}</c> (or a pid located at
another node), also never fails.</item>
</list>
</section>
<section>
<marker id="receive"></marker>
<title>Receive</title>
<pre>
receive
Pattern1 [when GuardSeq1] ->
Body1;
...;
PatternN [when GuardSeqN] ->
BodyN
end</pre>
<p>Receives messages sent to the process using the send operator
(!). The patterns <c>Pattern</c> are sequentially matched
against the first message in time order in the mailbox, then
the second, and so on. If a match succeeds and the optional
guard sequence <c>GuardSeq</c> is true, the corresponding
<c>Body</c> is evaluated. The matching message is consumed, that
is removed from the mailbox, while any other messages in
the mailbox remain unchanged.</p>
<p>The return value of <c>Body</c> is the return value of
the <c>receive</c> expression.</p>
<p><c>receive</c> never fails. Execution is suspended, possibly
indefinitely, until a message arrives that does match one of
the patterns and with a true guard sequence. </p>
<p>Example:</p>
<pre>
wait_for_onhook() ->
receive
onhook ->
disconnect(),
idle();
{connect, B} ->
B ! {busy, self()},
wait_for_onhook()
end.</pre>
<p>It is possible to augment the <c>receive</c> expression with a
timeout:</p>
<pre>
receive
Pattern1 [when GuardSeq1] ->
Body1;
...;
PatternN [when GuardSeqN] ->
BodyN
after
ExprT ->
BodyT
end</pre>
<p><c>ExprT</c> should evaluate to an integer. The highest allowed
value is 16#ffffffff, that is, the value must fit in 32 bits.
<c>receive..after</c> works exactly as <c>receive</c>, except
that if no matching message has arrived within <c>ExprT</c>
milliseconds, then <c>BodyT</c> is evaluated instead and its
return value becomes the return value of the <c>receive..after</c>
expression.</p>
<p>Example:</p>
<pre>
wait_for_onhook() ->
receive
onhook ->
disconnect(),
idle();
{connect, B} ->
B ! {busy, self()},
wait_for_onhook()
after
60000 ->
disconnect(),
error()
end.</pre>
<p>It is legal to use a <c>receive..after</c> expression with no
branches:</p>
<pre>
receive
after
ExprT ->
BodyT
end</pre>
<p>This construction will not consume any messages, only suspend
execution in the process for <c>ExprT</c> milliseconds and can be
used to implement simple timers.</p>
<p>Example:</p>
<pre>
timer() ->
spawn(m, timer, [self()]).
timer(Pid) ->
receive
after
5000 ->
Pid ! timeout
end.</pre>
<p>There are two special cases for the timeout value <c>ExprT</c>:</p>
<taglist>
<tag><c>infinity</c></tag>
<item>The process should wait indefinitely for a matching message
-- this is the same as not using a timeout. Can be
useful for timeout values that are calculated at run-time.</item>
<tag>0</tag>
<item>If there is no matching message in the mailbox, the timeout
will occur immediately.</item>
</taglist>
</section>
<section>
<title>Term Comparisons</title>
<pre>
Expr1 <input>op</input> Expr2</pre>
<table>
<row>
<cell align="left" valign="middle"><em>op</em></cell>
<cell align="left" valign="middle"><em>Description</em></cell>
</row>
<row>
<cell align="left" valign="middle">==</cell>
<cell align="left" valign="middle">equal to</cell>
</row>
<row>
<cell align="left" valign="middle">/=</cell>
<cell align="left" valign="middle">not equal to</cell>
</row>
<row>
<cell align="left" valign="middle">=<</cell>
<cell align="left" valign="middle">less than or equal to</cell>
</row>
<row>
<cell align="left" valign="middle"><</cell>
<cell align="left" valign="middle">less than</cell>
</row>
<row>
<cell align="left" valign="middle">>=</cell>
<cell align="left" valign="middle">greater than or equal to</cell>
</row>
<row>
<cell align="left" valign="middle">></cell>
<cell align="left" valign="middle">greater than</cell>
</row>
<row>
<cell align="left" valign="middle">=:=</cell>
<cell align="left" valign="middle">exactly equal to</cell>
</row>
<row>
<cell align="left" valign="middle">=/=</cell>
<cell align="left" valign="middle">exactly not equal to</cell>
</row>
<tcaption>Term Comparison Operators.</tcaption>
</table>
<p>The arguments may be of different data types. The following
order is defined:</p>
<pre>
number < atom < reference < fun < port < pid < tuple < list < bit string</pre>
<p>Lists are compared element by element. Tuples are ordered by
size, two tuples with the same size are compared element by
element.</p>
<p>If one of the compared terms is an integer and the other a
float, the integer is first converted into a float, unless the
operator is one of =:= and =/=. If the integer is too big to fit
in a float no conversion is done, but the order is determined by
inspecting the sign of the numbers.</p>
<p>Returns the Boolean value of the expression, <c>true</c> or
<c>false</c>.</p>
<p>Examples:</p>
<pre>
1> <input>1==1.0.</input>
true
2> <input>1=:=1.0.</input>
false
3> <input>1 > a.</input>
false</pre>
</section>
<section>
<title>Arithmetic Expressions</title>
<pre>
<input>op</input> Expr
Expr1 <input>op</input> Expr2</pre>
<table>
<row>
<cell align="left" valign="middle"><em>op</em></cell>
<cell align="left" valign="middle"><em>Description</em></cell>
<cell align="left" valign="middle"><em>Argument type</em></cell>
</row>
<row>
<cell align="left" valign="middle">+</cell>
<cell align="left" valign="middle">unary +</cell>
<cell align="left" valign="middle">number</cell>
</row>
<row>
<cell align="left" valign="middle">-</cell>
<cell align="left" valign="middle">unary -</cell>
<cell align="left" valign="middle">number</cell>
</row>
<row>
<cell align="left" valign="middle">+</cell>
<cell align="left" valign="middle"> </cell>
<cell align="left" valign="middle">number</cell>
</row>
<row>
<cell align="left" valign="middle">-</cell>
<cell align="left" valign="middle"> </cell>
<cell align="left" valign="middle">number</cell>
</row>
<row>
<cell align="left" valign="middle">*</cell>
<cell align="left" valign="middle"> </cell>
<cell align="left" valign="middle">number</cell>
</row>
<row>
<cell align="left" valign="middle">/</cell>
<cell align="left" valign="middle">floating point division</cell>
<cell align="left" valign="middle">number</cell>
</row>
<row>
<cell align="left" valign="middle">bnot</cell>
<cell align="left" valign="middle">unary bitwise not</cell>
<cell align="left" valign="middle">integer</cell>
</row>
<row>
<cell align="left" valign="middle">div</cell>
<cell align="left" valign="middle">integer division</cell>
<cell align="left" valign="middle">integer</cell>
</row>
<row>
<cell align="left" valign="middle">rem</cell>
<cell align="left" valign="middle">integer remainder of X/Y</cell>
<cell align="left" valign="middle">integer</cell>
</row>
<row>
<cell align="left" valign="middle">band</cell>
<cell align="left" valign="middle">bitwise and</cell>
<cell align="left" valign="middle">integer</cell>
</row>
<row>
<cell align="left" valign="middle">bor</cell>
<cell align="left" valign="middle">bitwise or</cell>
<cell align="left" valign="middle">integer</cell>
</row>
<row>
<cell align="left" valign="middle">bxor</cell>
<cell align="left" valign="middle">arithmetic bitwise xor</cell>
<cell align="left" valign="middle">integer</cell>
</row>
<row>
<cell align="left" valign="middle">bsl</cell>
<cell align="left" valign="middle">arithmetic bitshift left</cell>
<cell align="left" valign="middle">integer</cell>
</row>
<row>
<cell align="left" valign="middle">bsr</cell>
<cell align="left" valign="middle">bitshift right</cell>
<cell align="left" valign="middle">integer</cell>
</row>
<tcaption>Arithmetic Operators.</tcaption>
</table>
<p>Examples:</p>
<pre>
1> <input>+1.</input>
1
2> <input>-1.</input>
-1
3> <input>1+1.</input>
2
4> <input>4/2.</input>
2.0
5> <input>5 div 2.</input>
2
6> <input>5 rem 2.</input>
1
7> <input>2#10 band 2#01.</input>
0
8> <input>2#10 bor 2#01.</input>
3
9> <input>a + 10.</input>
** exception error: bad argument in an arithmetic expression
in operator +/2
called as a + 10
10> <input>1 bsl (1 bsl 64).</input>
** exception error: a system limit has been reached
in operator bsl/2
called as 1 bsl 18446744073709551616</pre>
</section>
<section>
<title>Boolean Expressions</title>
<pre>
<input>op</input> Expr
Expr1 <input>op</input> Expr2</pre>
<table>
<row>
<cell align="left" valign="middle"><em>op</em></cell>
<cell align="left" valign="middle"><em>Description</em></cell>
</row>
<row>
<cell align="left" valign="middle">not</cell>
<cell align="left" valign="middle">unary logical not</cell>
</row>
<row>
<cell align="left" valign="middle">and</cell>
<cell align="left" valign="middle">logical and</cell>
</row>
<row>
<cell align="left" valign="middle">or</cell>
<cell align="left" valign="middle">logical or</cell>
</row>
<row>
<cell align="left" valign="middle">xor</cell>
<cell align="left" valign="middle">logical xor</cell>
</row>
<tcaption>Logical Operators.</tcaption>
</table>
<p>Examples:</p>
<pre>
1> <input>not true.</input>
false
2> <input>true and false.</input>
false
3> <input>true xor false.</input>
true
4> <input>true or garbage.</input>
** exception error: bad argument
in operator or/2
called as true or garbage</pre>
</section>
<section>
<title>Short-Circuit Expressions</title>
<pre>
Expr1 orelse Expr2
Expr1 andalso Expr2</pre>
<p>Expressions where <c>Expr2</c> is evaluated only if
necessary. That is, <c>Expr2</c> is evaluated only if <c>Expr1</c>
evaluates to <c>false</c> in an <c>orelse</c> expression, or only
if <c>Expr1</c> evaluates to <c>true</c> in an <c>andalso</c>
expression. Returns either the value of <c>Expr1</c> (that is,
<c>true</c> or <c>false</c>) or the value of <c>Expr2</c>
(if <c>Expr2</c> was evaluated).</p>
<p>Example 1:</p>
<pre>
case A >= -1.0 andalso math:sqrt(A+1) > B of</pre>
<p>This will work even if <c>A</c> is less than <c>-1.0</c>,
since in that case, <c>math:sqrt/1</c> is never evaluated.</p>
<p>Example 2:</p>
<pre>
OnlyOne = is_atom(L) orelse
(is_list(L) andalso length(L) == 1),</pre>
<p>From R13A, <c>Expr2</c> is no longer required to evaluate to a
boolean value. As a consequence, <c>andalso</c> and <c>orelse</c>
are now tail-recursive. For instance, the following function is
tail-recursive in R13A and later:</p>
<pre>
all(Pred, [Hd|Tail]) ->
Pred(Hd) andalso all(Pred, Tail);
all(_, []) ->
true.</pre>
</section>
<section>
<title>List Operations</title>
<pre>
Expr1 ++ Expr2
Expr1 -- Expr2</pre>
<p>The list concatenation operator <c>++</c> appends its second
argument to its first and returns the resulting list.</p>
<p>The list subtraction operator <c>--</c> produces a list which
is a copy of the first argument, subjected to the following
procedure: for each element in the second argument, the first
occurrence of this element (if any) is removed.</p>
<p>Example:</p>
<pre>
1> <input>[1,2,3]++[4,5].</input>
[1,2,3,4,5]
2> <input>[1,2,3,2,1,2]--[2,1,2].</input>
[3,1,2]</pre>
<warning><p>The complexity of <c>A -- B</c> is
proportional to <c>length(A)*length(B)</c>, meaning that it
will be very slow if both <c>A</c> and <c>B</c> are
long lists.</p></warning>
</section>
<section>
<marker id="bit_syntax"></marker>
<title>Bit Syntax Expressions</title>
<code type="none"><![CDATA[<<>>
<<E1,...,En>>]]></code>
<p>Each element <c>Ei</c> specifies a <em>segment</em> of
the bit string. Each element <c>Ei</c> is a value, followed by an
optional <em>size expression</em> and an optional <em>type specifier list</em>.</p>
<pre>
Ei = Value |
Value:Size |
Value/TypeSpecifierList |
Value:Size/TypeSpecifierList</pre>
<p>Used in a bit string construction, <c>Value</c> is an expression
which should evaluate to an integer, float or bit string. If the
expression is something else than a single literal or variable, it
should be enclosed in parenthesis.</p>
<p>Used in a bit string matching, <c>Value</c> must be a variable,
or an integer, float or string.</p>
<p>Note that, for example, using a string literal as in
<c><![CDATA[<<"abc">>]]></c> is syntactic sugar for
<c><![CDATA[<<$a,$b,$c>>]]></c>.</p>
<p>Used in a bit string construction, <c>Size</c> is an expression
which should evaluate to an integer.</p>
<p>Used in a bit string matching, <c>Size</c> must be an integer or a
variable bound to an integer.</p>
<p>The value of <c>Size</c> specifies the size of the segment in
units (see below). The default value depends on the type (see
below). For <c>integer</c> it is 8, for
<c>float</c> it is 64, for <c>binary</c> and <c>bitstring</c> it is
the whole binary or bit string. In matching, this default value is only
valid for the very last element. All other bit string or binary
elements in the matching must have a size specification.</p>
<p>For the <c>utf8</c>, <c>utf16</c>, and <c>utf32</c> types,
<c>Size</c> must not be given. The size of the segment is implicitly
determined by the type and value itself.</p>
<p><c>TypeSpecifierList</c> is a list of type specifiers, in any
order, separated by hyphens (-). Default values are used for any
omitted type specifiers.</p>
<taglist>
<tag><c>Type</c>= <c>integer</c> | <c>float</c> | <c>binary</c> |
<c>bytes</c> | <c>bitstring</c> | <c>bits</c> |
<c>utf8</c> | <c>utf16</c> | <c>utf32</c> </tag>
<item>The default is <c>integer</c>. <c>bytes</c> is a shorthand for
<c>binary</c> and <c>bits</c> is a shorthand for <c>bitstring</c>.
See below for more information about the <c>utf</c> types.
</item>
<tag><c>Signedness</c>= <c>signed</c> | <c>unsigned</c></tag>
<item>Only matters for matching and when the type is <c>integer</c>.
The default is <c>unsigned</c>.</item>
<tag><c>Endianness</c>= <c>big</c> | <c>little</c> | <c>native</c></tag>
<item>Native-endian means that the endianness will be resolved at load
time to be either big-endian or little-endian, depending on
what is native for the CPU that the Erlang machine is run on.
Endianness only matters when the Type is either <c>integer</c>,
<c>utf16</c>, <c>utf32</c>, or <c>float</c>. The default is <c>big</c>.
</item>
<tag><c>Unit</c>= <c>unit:IntegerLiteral</c></tag>
<item>The allowed range is 1..256. Defaults to 1 for <c>integer</c>,
<c>float</c> and <c>bitstring</c>, and to 8 for <c>binary</c>.
No unit specifier must be given for the types
<c>utf8</c>, <c>utf16</c>, and <c>utf32</c>.
</item>
</taglist>
<p>The value of <c>Size</c> multiplied with the unit gives
the number of bits. A segment of type <c>binary</c> must have
a size that is evenly divisible by 8.</p>
<note><p>When constructing binaries, if the size <c>N</c> of an integer
segment is too small to contain the given integer, the most significant
bits of the integer will be silently discarded and only the <c>N</c> least
significant bits will be put into the binary.</p></note>
<p>The types <c>utf8</c>, <c>utf16</c>, and <c>utf32</c> specifies
encoding/decoding of the <em>Unicode Transformation Format</em>s UTF-8, UTF-16,
and UTF-32, respectively.</p>
<p>When constructing a segment of a <c>utf</c> type, <c>Value</c>
must be an integer in one of the ranges 0..16#D7FF,
16#E000..16#FFFD, or 16#10000..16#10FFFF
(i.e. a valid Unicode code point). Construction
will fail with a <c>badarg</c> exception if <c>Value</c> is
outside the allowed ranges. The size of the resulting binary
segment depends on the type and/or <c>Value</c>. For <c>utf8</c>,
<c>Value</c> will be encoded in 1 through 4 bytes. For
<c>utf16</c>, <c>Value</c> will be encoded in 2 or 4
bytes. Finally, for <c>utf32</c>, <c>Value</c> will always be
encoded in 4 bytes.</p>
<p>When constructing, a literal string may be given followed
by one of the UTF types, for example: <c><![CDATA[<<"abc"/utf8>>]]></c>
which is syntatic sugar for
<c><![CDATA[<<$a/utf8,$b/utf8,$c/utf8>>]]></c>.</p>
<p>A successful match of a segment of a <c>utf</c> type results
in an integer in one of the ranges 0..16#D7FF, 16#E000..16#FFFD,
or 16#10000..16#10FFFF
(i.e. a valid Unicode code point). The match will fail if returned value
would fall outside those ranges.</p>
<p>A segment of type <c>utf8</c> will match 1 to 4 bytes in the binary,
if the binary at the match position contains a valid UTF-8 sequence.
(See RFC-2279 or the Unicode standard.)</p>
<p>A segment of type <c>utf16</c> may match 2 or 4 bytes in the binary.
The match will fail if the binary at the match position does not contain
a legal UTF-16 encoding of a Unicode code point. (See RFC-2781 or
the Unicode standard.)</p>
<p>A segment of type <c>utf32</c> may match 4 bytes in the binary in the
same way as an <c>integer</c> segment matching 32 bits.
The match will fail if the resulting integer is outside the legal ranges
mentioned above.</p>
<p>Examples:</p>
<pre>
1> <input>Bin1 = <<1,17,42>>.</input>
<<1,17,42>>
2> <input>Bin2 = <<"abc">>.</input>
<<97,98,99>>
3> <input>Bin3 = <<1,17,42:16>>.</input>
<<1,17,0,42>>
4> <input><<A,B,C:16>> = <<1,17,42:16>>.</input>
<<1,17,0,42>>
5> <input>C.</input>
42
6> <input><<D:16,E,F>> = <<1,17,42:16>>.</input>
<<1,17,0,42>>
7> <input>D.</input>
273
8> <input>F.</input>
42
9> <input><<G,H/binary>> = <<1,17,42:16>>.</input>
<<1,17,0,42>>
10> <input>H.</input>
<<17,0,42>>
11> <input><<G,H/bitstring>> = <<1,17,42:12>>.</input>
<<1,17,1,10:4>>
12> <input>H.</input>
<<17,1,10:4>>
13> <input><<1024/utf8>>.</input>
<<208,128>>
</pre>
<p>Note that bit string patterns cannot be nested.</p>
<p>Note also that "<c><![CDATA[B=<<1>>]]></c>" is interpreted as
"<c><![CDATA[B =<<1>>]]></c>" which is a syntax error. The correct way is
to write a space after '=': "<c><![CDATA[B= <<1>>]]></c>.</p>
<p>More examples can be found in <em>Programming Examples</em>.</p>
</section>
<section>
<marker id="funs"></marker>
<title>Fun Expressions</title>
<pre>
fun
(Pattern11,...,Pattern1N) [when GuardSeq1] ->
Body1;
...;
(PatternK1,...,PatternKN) [when GuardSeqK] ->
BodyK
end</pre>
<p>A fun expression begins with the keyword <c>fun</c> and ends
with the keyword <c>end</c>. Between them should be a function
declaration, similar to a
<seealso marker="functions#syntax">regular function declaration</seealso>, except that no function name is
specified.</p>
<p>Variables in a fun head shadow variables in the
function clause surrounding the fun expression, and
variables bound in a fun body are local to the fun body.</p>
<p>The return value of the expression is the resulting fun.</p>
<p>Examples:</p>
<pre>
1> <input>Fun1 = fun (X) -> X+1 end.</input>
#Fun<erl_eval.6.39074546>
2> <input>Fun1(2).</input>
3
3> <input>Fun2 = fun (X) when X>=5 -> gt; (X) -> lt end.</input>
#Fun<erl_eval.6.39074546>
4> <input>Fun2(7).</input>
gt</pre>
<p>The following fun expressions are also allowed:</p>
<pre>
fun Name/Arity
fun Module:Name/Arity</pre>
<p>In <c>Name/Arity</c>, <c>Name</c> is an atom and <c>Arity</c> is an integer.
<c>Name/Arity</c> must specify an existing local function. The expression is
syntactic sugar for:</p>
<pre>
fun (Arg1,...,ArgN) -> Name(Arg1,...,ArgN) end</pre>
<p>In <c>Module:Name/Arity</c>, <c>Module</c> and <c>Name</c> are atoms
and <c>Arity</c> is an integer.
A fun defined in this way will refer to the function <c>Name</c>
with arity <c>Arity</c> in the <em>latest</em> version of module <c>Module</c>.
</p>
<p>When applied to a number N of arguments, a tuple
<c>{Module,FunctionName}</c> is interpreted as a fun, referring
to the function <c>FunctionName</c> with arity N in the module
<c>Module</c>. The function must be exported.
<em>This usage is deprecated.</em>
See <seealso marker="#calls">Function Calls</seealso> for an example.</p>
<p>More examples can be found in <em>Programming Examples</em>.</p>
</section>
<section>
<marker id="catch"></marker>
<title>Catch and Throw</title>
<code type="none">
catch Expr</code>
<p>Returns the value of <c>Expr</c> unless an exception
occurs during the evaluation. In that case, the exception is
caught. For exceptions of class <c>error</c>,
that is run-time errors: <c>{'EXIT',{Reason,Stack}}</c>
is returned. For exceptions of class <c>exit</c>, that is
the code called <c>exit(Term)</c>: <c>{'EXIT',Term}</c> is returned.
For exceptions of class <c>throw</c>, that is
the code called <c>throw(Term)</c>: <c>Term</c> is returned.</p>
<p><c>Reason</c> depends on the type of error that occurred, and
<c>Stack</c> is the stack of recent function calls, see
<seealso marker="errors#exit_reasons">Errors and Error Handling</seealso>.</p>
<p>Examples:</p>
<p></p>
<pre>
1> <input>catch 1+2.</input>
3
2> <input>catch 1+a.</input>
{'EXIT',{badarith,[...]}}</pre>
<p>Note that <c>catch</c> has low precedence and catch
subexpressions often needs to be enclosed in a block
expression or in parenthesis:</p>
<pre>
3> <input>A = catch 1+2.</input>
** 1: syntax error before: 'catch' **
4> <input>A = (catch 1+2).</input>
3</pre>
<p>The BIF <c>throw(Any)</c> can be used for non-local return from
a function. It must be evaluated within a <c>catch</c>, which will
return the value <c>Any</c>. Example:</p>
<pre>
5> <input>catch throw(hello).</input>
hello</pre>
<p>If <c>throw/1</c> is not evaluated within a catch, a
<c>nocatch</c> run-time error will occur.</p>
</section>
<section>
<marker id="try"></marker>
<title>Try</title>
<code type="none">
try Exprs
catch
[Class1:]ExceptionPattern1 [when ExceptionGuardSeq1] ->
ExceptionBody1;
[ClassN:]ExceptionPatternN [when ExceptionGuardSeqN] ->
ExceptionBodyN
end</code>
<p>This is an enhancement of
<seealso marker="#catch">catch</seealso> that appeared in
Erlang 5.4/OTP-R10B. It gives the possibility do distinguish
between different exception classes, and to choose to handle only
the desired ones, passing the others on to an enclosing
<c>try</c> or <c>catch</c> or to default error handling.</p>
<p>Note that although the keyword <c>catch</c> is used in
the <c>try</c> expression, there is not a <c>catch</c> expression
within the <c>try</c> expression.</p>
<p>Returns the value of <c>Exprs</c> (a sequence of expressions
<c>Expr1, ..., ExprN</c>) unless an exception occurs during
the evaluation. In that case the exception is caught and
the patterns <c>ExceptionPattern</c> with the right exception
class <c>Class</c> are sequentially matched against the caught
exception. An omitted <c>Class</c> is shorthand for <c>throw</c>.
If a match succeeds and the optional guard sequence
<c>ExceptionGuardSeq</c> is true, the corresponding
<c>ExceptionBody</c> is evaluated to become the return value.</p>
<p>If an exception occurs during evaluation of <c>Exprs</c> but
there is no matching <c>ExceptionPattern</c> of the right
<c>Class</c> with a true guard sequence, the exception is passed
on as if <c>Exprs</c> had not been enclosed in a <c>try</c>
expression.</p>
<p>If an exception occurs during evaluation of <c>ExceptionBody</c>
it is not caught.</p>
<p>The <c>try</c> expression can have an <c>of</c>
section:
</p>
<code type="none">
try Exprs of
Pattern1 [when GuardSeq1] ->
Body1;
...;
PatternN [when GuardSeqN] ->
BodyN
catch
[Class1:]ExceptionPattern1 [when ExceptionGuardSeq1] ->
ExceptionBody1;
...;
[ClassN:]ExceptionPatternN [when ExceptionGuardSeqN] ->
ExceptionBodyN
end</code>
<p>If the evaluation of <c>Exprs</c> succeeds without an exception,
the patterns <c>Pattern</c> are sequentially matched against
the result in the same way as for a
<seealso marker="#case">case</seealso> expression, except that if
the matching fails, a <c>try_clause</c> run-time error will occur.</p>
<p>An exception occurring during the evaluation of <c>Body</c> is
not caught.</p>
<p>The <c>try</c> expression can also be augmented with an
<c>after</c> section, intended to be used for cleanup with side
effects:</p>
<code type="none">
try Exprs of
Pattern1 [when GuardSeq1] ->
Body1;
...;
PatternN [when GuardSeqN] ->
BodyN
catch
[Class1:]ExceptionPattern1 [when ExceptionGuardSeq1] ->
ExceptionBody1;
...;
[ClassN:]ExceptionPatternN [when ExceptionGuardSeqN] ->
ExceptionBodyN
after
AfterBody
end</code>
<p><c>AfterBody</c> is evaluated after either <c>Body</c> or
<c>ExceptionBody</c> no matter which one. The evaluated value of
<c>AfterBody</c> is lost; the return value of the <c>try</c>
expression is the same with an <c>after</c> section as without.</p>
<p>Even if an exception occurs during evaluation of <c>Body</c> or
<c>ExceptionBody</c>, <c>AfterBody</c> is evaluated. In this case
the exception is passed on after <c>AfterBody</c> has been
evaluated, so the exception from the <c>try</c> expression is
the same with an <c>after</c> section as without.</p>
<p>If an exception occurs during evaluation of <c>AfterBody</c>
itself it is not caught, so if <c>AfterBody</c> is evaluated after
an exception in <c>Exprs</c>, <c>Body</c> or <c>ExceptionBody</c>,
that exception is lost and masked by the exception in
<c>AfterBody</c>.</p>
<p>The <c>of</c>, <c>catch</c> and <c>after</c> sections are all
optional, as long as there is at least a <c>catch</c> or an
<c>after</c> section, so the following are valid <c>try</c>
expressions:</p>
<code type="none">
try Exprs of
Pattern when GuardSeq ->
Body
after
AfterBody
end
try Exprs
catch
ExpressionPattern ->
ExpressionBody
after
AfterBody
end
try Exprs after AfterBody end</code>
<p>Example of using <c>after</c>, this code will close the file
even in the event of exceptions in <c>file:read/2</c> or in
<c>binary_to_term/1</c>, and exceptions will be the same as
without the <c>try</c>...<c>after</c>...<c>end</c> expression:</p>
<code type="none">
termize_file(Name) ->
{ok,F} = file:open(Name, [read,binary]),
try
{ok,Bin} = file:read(F, 1024*1024),
binary_to_term(Bin)
after
file:close(F)
end.</code>
<p>Example: Using <c>try</c> to emulate <c>catch Expr</c>.</p>
<code type="none">
try Expr
catch
throw:Term -> Term;
exit:Reason -> {'EXIT',Reason}
error:Reason -> {'EXIT',{Reason,erlang:get_stacktrace()}}
end</code>
</section>
<section>
<title>Parenthesized Expressions</title>
<pre>
(Expr)</pre>
<p>Parenthesized expressions are useful to override
<seealso marker="#prec">operator precedences</seealso>,
for example in arithmetic expressions:</p>
<pre>
1> <input>1 + 2 * 3.</input>
7
2> <input>(1 + 2) * 3.</input>
9</pre>
</section>
<section>
<title>Block Expressions</title>
<pre>
begin
Expr1,
...,
ExprN
end</pre>
<p>Block expressions provide a way to group a sequence of
expressions, similar to a clause body. The return value is
the value of the last expression <c>ExprN</c>.</p>
</section>
<section>
<marker id="lcs"></marker>
<title>List Comprehensions</title>
<p>List comprehensions are a feature of many modern functional
programming languages. Subject to certain rules, they provide a
succinct notation for generating elements in a list.</p>
<p>List comprehensions are analogous to set comprehensions in
Zermelo-Frankel set theory and are called ZF expressions in
Miranda. They are analogous to the <c>setof</c> and
<c>findall</c> predicates in Prolog.</p>
<p>List comprehensions are written with the following syntax:</p>
<pre>
[Expr || Qualifier1,...,QualifierN]</pre>
<p><c>Expr</c> is an arbitrary expression, and each
<c>Qualifier</c> is either a generator or a filter.</p>
<list type="bulleted">
<item>A <em>generator</em> is written as: <br></br>
<c><![CDATA[Pattern <- ListExpr]]></c>. <br></br>
<c>ListExpr</c> must be an expression which evaluates to a
list of terms.</item>
<item>A <em>bit string generator</em> is written as: <br></br>
<c><![CDATA[BitstringPattern <= BitStringExpr]]></c>. <br></br>
<c>BitStringExpr</c> must be an expression which evaluates to a
bitstring.</item>
<item>A <em>filter</em> is an expression which evaluates to
<c>true</c> or <c>false</c>.</item>
</list>
<p>The variables in the generator patterns shadow variables in the function
clause surrounding the list comprehensions.</p> <p>A list comprehension
returns a list, where the elements are the result of evaluating <c>Expr</c>
for each combination of generator list elements and bit string generator
elements for which all filters are true.</p> <p></p> <p>Example:</p>
<pre>
1> <input>[X*2 || X <- [1,2,3]].</input>
[2,4,6]</pre>
<p>More examples can be found in <em>Programming Examples</em>.</p>
</section>
<section>
<title>Bit String Comprehensions</title>
<p>Bit string comprehensions are
analogous to List Comprehensions. They are used to generate bit strings
efficiently and succinctly.</p>
<p>Bit string comprehensions are written with
the following syntax:</p>
<pre>
<< BitString || Qualifier1,...,QualifierN >></pre>
<p><c>BitString</c> is a bit string expression, and each
<c>Qualifier</c> is either a generator, a bit string generator or a filter.</p>
<list type="bulleted">
<item>A <em>generator</em> is written as: <br></br>
<c><![CDATA[Pattern <- ListExpr]]></c>. <br></br>
<c>ListExpr</c> must be an expression which evaluates to a
list of terms.</item>
<item>A <em>bit string generator</em> is written as: <br></br>
<c><![CDATA[BitstringPattern <= BitStringExpr]]></c>. <br></br>
<c>BitStringExpr</c> must be an expression which evaluates to a
bitstring.</item>
<item>A <em>filter</em> is an expression which evaluates to
<c>true</c> or <c>false</c>.</item>
</list>
<p>The variables in the generator patterns shadow variables in
the function clause surrounding the bit string comprehensions.</p>
<p>A bit string comprehension returns a bit string, which is
created by concatenating the results of evaluating <c>BitString</c>
for each combination of bit string generator elements for which all
filters are true.</p>
<p></p>
<p>Example:</p>
<pre>
1> <input><< << (X*2) >> ||
<<X>> <= << 1,2,3 >> >>.</input>
<<2,4,6>></pre>
<p>More examples can be found in <em>Programming Examples</em>.</p>
</section>
<section>
<marker id="guards"></marker>
<title>Guard Sequences</title>
<p>A <em>guard sequence</em> is a sequence of guards, separated
by semicolon (;). The guard sequence is true if at least one of
the guards is true. (The remaining guards, if any, will not be
evaluated.)<br></br>
<c>Guard1;...;GuardK</c></p>
<p>A <em>guard</em> is a sequence of guard expressions, separated
by comma (,). The guard is true if all guard expressions
evaluate to <c>true</c>.<br></br>
<c>GuardExpr1,...,GuardExprN</c></p>
<p>The set of valid <em>guard expressions</em> (sometimes called
guard tests) is a subset of the set of valid Erlang expressions.
The reason for restricting the set of valid expressions is that
evaluation of a guard expression must be guaranteed to be free
of side effects. Valid guard expressions are:</p>
<list type="bulleted">
<item>the atom <c>true</c>,</item>
<item>other constants (terms and bound variables), all regarded
as false,</item>
<item>calls to the BIFs specified below,</item>
<item>term comparisons,</item>
<item>arithmetic expressions,</item>
<item>boolean expressions, and</item>
<item>short-circuit expressions (<c>andalso</c>/<c>orelse</c>).</item>
</list>
<table>
<row>
<cell align="left" valign="middle"><c>is_atom/1</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_binary/1</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_bitstring/1</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_float/1</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_function/1</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_function/2</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_integer/1</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_list/1</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_number/1</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_pid/1</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_port/1</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_record/2</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_record/3</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_reference/1</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>is_tuple/1</c></cell>
</row>
<tcaption>Type Test BIFs.</tcaption>
</table>
<p>Note that most type test BIFs have older equivalents, without
the <c>is_</c> prefix. These old BIFs are retained for backwards
compatibility only and should not be used in new code. They are
also only allowed at top level. For example, they are not allowed
in boolean expressions in guards.</p>
<table>
<row>
<cell align="left" valign="middle"><c>abs(Number)</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>bit_size(Bitstring)</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>byte_size(Bitstring)</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>element(N, Tuple)</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>float(Term)</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>hd(List)</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>length(List)</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>node()</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>node(Pid|Ref|Port)</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>round(Number)</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>self()</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>size(Tuple|Bitstring)</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>tl(List)</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>trunc(Number)</c></cell>
</row>
<row>
<cell align="left" valign="middle"><c>tuple_size(Tuple)</c></cell>
</row>
<tcaption>Other BIFs Allowed in Guard Expressions.</tcaption>
</table>
<p>If an arithmetic expression, a boolean expression, a
short-circuit expression, or a call to a guard BIF fails (because
of invalid arguments), the entire guard fails. If the guard was
part of a guard sequence, the next guard in the sequence (that is,
the guard following the next semicolon) will be evaluated.</p>
</section>
<section>
<marker id="prec"></marker>
<title>Operator Precedence</title>
<p>Operator precedence in falling priority:</p>
<table>
<row>
<cell align="left" valign="middle">:</cell>
<cell align="left" valign="middle"> </cell>
</row>
<row>
<cell align="left" valign="middle">#</cell>
<cell align="left" valign="middle"> </cell>
</row>
<row>
<cell align="left" valign="middle">Unary + - bnot not</cell>
<cell align="left" valign="middle"> </cell>
</row>
<row>
<cell align="left" valign="middle">/ * div rem band and</cell>
<cell align="left" valign="middle">Left associative</cell>
</row>
<row>
<cell align="left" valign="middle">+ - bor bxor bsl bsr or xor</cell>
<cell align="left" valign="middle">Left associative</cell>
</row>
<row>
<cell align="left" valign="middle">++ --</cell>
<cell align="left" valign="middle">Right associative</cell>
</row>
<row>
<cell align="left" valign="middle">== /= =< < >= > =:= =/=</cell>
<cell align="left" valign="middle"> </cell>
</row>
<row>
<cell align="left" valign="middle">andalso</cell>
<cell align="left" valign="middle"> </cell>
</row>
<row>
<cell align="left" valign="middle">orelse</cell>
<cell align="left" valign="middle"> </cell>
</row>
<row>
<cell align="left" valign="middle">= !</cell>
<cell align="left" valign="middle">Right associative</cell>
</row>
<row>
<cell align="left" valign="middle">catch</cell>
<cell align="left" valign="middle"> </cell>
</row>
<tcaption>Operator Precedence.</tcaption>
</table>
<p>When evaluating an expression, the operator with the highest
priority is evaluated first. Operators with the same priority
are evaluated according to their associativity. Example:
The left associative arithmetic operators are evaluated left to
right:</p>
<pre>
<input>6 + 5 * 4 - 3 / 2</input> evaluates to
<input>6 + 20 - 1.5</input> evaluates to
<input>26 - 1.5</input> evaluates to
<input>24.5</input></pre>
</section>
</chapter>
