<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd">
<chapter>
<header>
<copyright>
<year>2003</year><year>2015</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>Compilation and Code Loading</title>
<prepared></prepared>
<docno></docno>
<date></date>
<rev></rev>
<file>code_loading.xml</file>
</header>
<p>How code is compiled and loaded is not a language issue, but
is system-dependent. This section describes compilation and
code loading in Erlang/OTP with references to relevant parts of
the documentation.</p>
<section>
<title>Compilation</title>
<p>Erlang programs must be <em>compiled</em> to object code.
The compiler can generate a new file that contains the object
code. The current abstract machine, which runs the object code, is
called BEAM, therefore the object files get the suffix
<c>.beam</c>. The compiler can also generate a binary which can
be loaded directly.</p>
<p>The compiler is located in the module <c>compile</c> (see the
<seealso marker="compiler:compile">compile(3)</seealso> manual page in
Compiler).</p>
<pre>
compile:file(Module)
compile:file(Module, Options)</pre>
<p>The Erlang shell understands the command <c>c(Module)</c> which
both compiles and loads <c>Module</c>.</p>
<p>There is also a module <c>make</c>, which provides a set of
functions similar to the UNIX type Make functions, see the
<seealso marker="tools:make">make(3)</seealso>
manual page in Tools.</p>
<p>The compiler can also be accessed from the OS prompt, see the
<seealso marker="erts:erl">erl(1)</seealso> manual page in ERTS.</p>
<pre>
% erl -compile <input>Module1</input>...<input>ModuleN</input>
% erl -make</pre>
<p>The <c>erlc</c> program provides an even better way to compile
modules from the shell, see the
<seealso marker="erts:erlc">erlc(1)</seealso> manual page in ERTS.
It understands a
number of flags that can be used to define macros, add search
paths for include files, and more.</p>
<pre>
% erlc <input><flags></input> <input>File1.erl</input>...<input>FileN.erl</input></pre>
</section>
<section>
<marker id="loading"></marker>
<title>Code Loading</title>
<p>The object code must be <em>loaded</em> into the Erlang runtime
system. This is handled by the <em>code server</em>, see the
<seealso marker="kernel:code">code(3)</seealso>
manual page in Kernel.</p>
<p>The code server loads code according to a code loading strategy,
which is either <em>interactive</em> (default) or
<em>embedded</em>. In interactive mode, code is searched for in
a <em>code path</em> and loaded when first referenced. In
embedded mode, code is loaded at start-up according to a
<em>boot script</em>. This is described in
<seealso marker="doc/system_principles:system_principles#code_loading">
System Principles </seealso>.</p>
</section>
<section>
<title>Code Replacement</title>
<p>Erlang supports change of code in a running system. Code
replacement is done on module level.</p>
<p>The code of a module can exist in two variants in a system:
<em>current</em> and <em>old</em>. When a module is loaded into
the system for the first time, the code becomes 'current'. If then
a new instance of the module is loaded, the code of the previous
instance becomes 'old' and the new instance becomes 'current'.</p>
<p>Both old and current code is valid, and can be evaluated
concurrently. Fully qualified function calls always refer to
current code. Old code can still be evaluated because of processes
lingering in the old code.</p>
<p>If a third instance of the module is loaded, the code server
removes (purges) the old code and any processes lingering in it is
terminated. Then the third instance becomes 'current' and
the previously current code becomes 'old'.</p>
<p>To change from old code to current code, a process must make a
fully qualified function call.</p>
<p><em>Example:</em></p>
<pre>
-module(m).
-export([loop/0]).
loop() ->
receive
code_switch ->
m:loop();
Msg ->
...
loop()
end.</pre>
<p>To make the process change code, send the message
<c>code_switch</c> to it. The process then makes a fully
qualified call to <c>m:loop()</c> and changes to current code.
Notice that <c>m:loop/0</c> must be exported.</p>
<p>For code replacement of funs to work, use the syntax
<c>fun Module:FunctionName/Arity</c>.</p>
</section>
<section>
<marker id="on_load"></marker>
<title>Running a Function When a Module is Loaded</title>
<warning>
<p>The <c>on_load</c> feature is to be considered experimental
as there are a number of known weak points in current semantics,
which therefore might change in future Erlang/OTP releases:</p>
<list>
<item><p>Doing external call in <c>on_load</c> to the module itself
leads to deadlock.</p></item>
<item><p>At module upgrade, other processes calling the module
get suspended waiting for <c>on_load</c> to finish. This can be very bad
for applications with demands on realtime characteristics.</p></item>
<item><p>At module upgrade, no rollback is done if the
<c>on_load</c> function fails.
The system is left in a bad limbo state without any working
and reachable instance of the module.</p></item>
</list>
<p>The problems with module upgrade described above can be fixed in future
Erlang/OTP releases by changing the behaviour to not make the module reachable until
after the <c>on_load</c> function has successfully returned.</p>
</warning>
<p>The <c>-on_load()</c> directive names a function that is to
be run automatically when a module is loaded.</p>
<p>Its syntax is as follows:</p>
<pre>
-on_load(Name/0).</pre>
<p>It is not necessary to export the function. It is called in a
freshly spawned process (which terminates as soon as the function
returns). The function must return <c>ok</c> if the module is to
remain loaded and become callable, or any other value if the module
is to be unloaded. Generating an exception also causes the
module to be unloaded. If the return value is not an atom,
a warning error report is sent to the error logger.</p>
<p>A process that calls any function in a module whose <c>on_load</c>
function has not yet returned, is suspended until the <c>on_load</c>
function has returned.</p>
<p>In embedded mode, first all modules are loaded.
Then all <c>on_load</c> functions are called. The system is
terminated unless all of the <c>on_load</c> functions return
<c>ok</c></p>.
<p><em>Example:</em></p>
<pre>
-module(m).
-on_load(load_my_nifs/0).
load_my_nifs() ->
NifPath = ..., %Set up the path to the NIF library.
Info = ..., %Initialize the Info term
erlang:load_nif(NifPath, Info).</pre>
<p>If the call to <c>erlang:load_nif/2</c> fails, the module
is unloaded and a warning report is sent to
the error loader.</p>
</section>
</chapter>
