<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd">
<chapter>
<header>
<copyright>
<year>1997</year><year>2013</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>Overview</title>
<prepared></prepared>
<docno></docno>
<date></date>
<rev></rev>
<file>des_princ.xml</file>
</header>
<marker id="otp design principles"></marker>
<p>The <em>OTP design principles</em> define how to
structure Erlang code in terms of processes, modules,
and directories.</p>
<section>
<title>Supervision Trees</title>
<p>A basic concept in Erlang/OTP is the <em>supervision tree</em>.
This is a process structuring model based on the idea of
<em>workers</em> and <em>supervisors</em>:</p>
<list type="bulleted">
<item>Workers are processes that perform computations, that is,
they do the actual work.</item>
<item>Supervisors are processes that monitor the behaviour of
workers. A supervisor can restart a worker if something goes
wrong.</item>
<item>The supervision tree is a hierarchical arrangement of
code into supervisors and workers, which makes it possible to
design and program fault-tolerant software.</item>
</list>
<p>In the following figure, square boxes represents supervisors and
circles represent workers:</p>
<marker id="sup6"></marker>
<image file="../design_principles/sup6.gif">
<icaption>Supervision Tree</icaption>
</image>
</section>
<section>
<title>Behaviours</title>
<p>In a supervision tree, many of the processes have similar
structures, they follow similar patterns. For example,
the supervisors are similar in structure. The only difference
between them is which child processes they supervise. Many
of the workers are servers in a server-client relation,
finite-state machines, or event handlers such as error loggers.</p>
<p><em>Behaviours</em> are formalizations of these common patterns.
The idea is to divide the code for a process in a generic part
(a behaviour module) and a specific part (a
<em>callback module</em>).</p>
<p>The behaviour module is part of Erlang/OTP. To implement a
process such as a supervisor, the user only has to implement
the callback module which is to export a pre-defined set of
functions, the <em>callback functions</em>.</p>
<p>The following example illustrate how code can be divided into a
generic and a specific part. Consider the following code (written in
plain Erlang) for a simple server, which keeps track of a number
of "channels". Other processes can allocate and free the channels
by calling the functions <c>alloc/0</c> and <c>free/1</c>,
respectively.</p>
<marker id="ch1"></marker>
<code type="none">
-module(ch1).
-export([start/0]).
-export([alloc/0, free/1]).
-export([init/0]).
start() ->
spawn(ch1, init, []).
alloc() ->
ch1 ! {self(), alloc},
receive
{ch1, Res} ->
Res
end.
free(Ch) ->
ch1 ! {free, Ch},
ok.
init() ->
register(ch1, self()),
Chs = channels(),
loop(Chs).
loop(Chs) ->
receive
{From, alloc} ->
{Ch, Chs2} = alloc(Chs),
From ! {ch1, Ch},
loop(Chs2);
{free, Ch} ->
Chs2 = free(Ch, Chs),
loop(Chs2)
end.</code>
<p>The code for the server can be rewritten into a generic part
<c>server.erl</c>:</p>
<code type="none">
-module(server).
-export([start/1]).
-export([call/2, cast/2]).
-export([init/1]).
start(Mod) ->
spawn(server, init, [Mod]).
call(Name, Req) ->
Name ! {call, self(), Req},
receive
{Name, Res} ->
Res
end.
cast(Name, Req) ->
Name ! {cast, Req},
ok.
init(Mod) ->
register(Mod, self()),
State = Mod:init(),
loop(Mod, State).
loop(Mod, State) ->
receive
{call, From, Req} ->
{Res, State2} = Mod:handle_call(Req, State),
From ! {Mod, Res},
loop(Mod, State2);
{cast, Req} ->
State2 = Mod:handle_cast(Req, State),
loop(Mod, State2)
end.</code>
<p>And a callback module <c>ch2.erl</c>:</p>
<code type="none">
-module(ch2).
-export([start/0]).
-export([alloc/0, free/1]).
-export([init/0, handle_call/2, handle_cast/2]).
start() ->
server:start(ch2).
alloc() ->
server:call(ch2, alloc).
free(Ch) ->
server:cast(ch2, {free, Ch}).
init() ->
channels().
handle_call(alloc, Chs) ->
alloc(Chs). % => {Ch,Chs2}
handle_cast({free, Ch}, Chs) ->
free(Ch, Chs). % => Chs2</code>
<p>Notice the following:</p>
<list type="bulleted">
<item>The code in <c>server</c> can be reused to build many
different servers.</item>
<item>The server name, in this example the atom
<c>ch2</c>, is hidden from the users of the client functions. This
means that the name can be changed without affecting them.</item>
<item>The protcol (messages sent to and received from the server)
is also hidden. This is good programming practice and allows
one to change the protocol without changing the code using
the interface functions.</item>
<item>The functionality of <c>server</c> can be extended without
having to change <c>ch2</c> or any other callback module.</item>
</list>
<p>In <c>ch1.erl</c> and <c>ch2.erl</c> above, the implementation
of <c>channels/0</c>, <c>alloc/1</c>, and <c>free/2</c> has been
intentionally left out, as it is not relevant to the example.
For completeness, one way to write these functions are given
below. This is an example only, a realistic
implementation must be able to handle situations like running out
of channels to allocate, and so on.</p>
<code type="none">
channels() ->
{_Allocated = [], _Free = lists:seq(1,100)}.
alloc({Allocated, [H|T] = _Free}) ->
{H, {[H|Allocated], T}}.
free(Ch, {Alloc, Free} = Channels) ->
case lists:member(Ch, Alloc) of
true ->
{lists:delete(Ch, Alloc), [Ch|Free]};
false ->
Channels
end. </code>
<p>Code written without using behaviours can be more
efficient, but the increased efficiency is at the expense of
generality. The ability to manage all applications in the system
in a consistent manner is important.</p>
<p>Using behaviours also makes it easier to read and understand
code written by other programmers. Improvised programming structures,
while possibly more efficient, are always more difficult to
understand.</p>
<p>The <c>server</c> module corresponds, greatly simplified,
to the Erlang/OTP behaviour <c>gen_server</c>.</p>
<p>The standard Erlang/OTP behaviours are:</p>
<list type="bulleted">
<item><p><seealso marker="gen_server_concepts">gen_server</seealso></p>
<p>For implementing the server of a client-server relation</p></item>
<item><p><seealso marker="fsm">gen_fsm</seealso></p>
<p>For implementing finite-state machines</p></item>
<item><p><seealso marker="events">gen_event</seealso></p>
<p>For implementing event handling functionality</p></item>
<item><p><seealso marker="sup_princ">supervisor</seealso></p>
<p>For implementing a supervisor in a supervision tree</p></item>
</list>
<p>The compiler understands the module attribute
<c>-behaviour(Behaviour)</c> and issues warnings about
missing callback functions, for example:</p>
<code type="none">
-module(chs3).
-behaviour(gen_server).
...
3> c(chs3).
./chs3.erl:10: Warning: undefined call-back function handle_call/3
{ok,chs3}</code>
</section>
<section>
<title>Applications</title>
<p>Erlang/OTP comes with a number of components, each implementing
some specific functionality. Components are with Erlang/OTP
terminology called <em>applications</em>. Examples of Erlang/OTP
applications are Mnesia, which has everything needed for
programming database services, and Debugger, which is used
to debug Erlang programs. The minimal system based on Erlang/OTP
consists of the following two applications:</p>
<list type="bulleted">
<item>Kernel - Functionality necessary to run Erlang</item>
<item>STDLIB - Erlang standard libraries</item>
</list>
<p>The application concept applies both to program structure
(processes) and directory structure (modules).</p>
<p>The simplest applications do not have any processes,
but consist of a collection of functional modules. Such an
application is called a <em>library application</em>. An example
of a library application is STDLIB.</p>
<p>An application with processes is easiest implemented as a
supervision tree using the standard behaviours.</p>
<p>How to program applications is described in
<seealso marker="applications">Applications</seealso>.</p>
</section>
<section>
<title>Releases</title>
<p>A <em>release</em> is a complete system made out from a subset of
Erlang/OTP applications and a set of user-specific applications.</p>
<p>How to program releases is described in
<seealso marker="release_structure">Releases</seealso>.</p>
<p>How to install a release in a target environment is described
in the section about target systems in Section 2 System Principles.</p>
</section>
<section>
<title>Release Handling</title>
<p><em>Release handling</em> is upgrading and downgrading between
different versions of a release, in a (possibly) running system.
How to do this is described in
<seealso marker="release_handling">Release Handling</seealso>.</p>
</section>
</chapter>
