aboutsummaryrefslogblamecommitdiffstats
path: root/system/doc/design_principles/gen_server_concepts.xml
blob: 231333da0e9e0987464906bb633c41a78bffd8de (plain) (tree)









































































































































































































































































                                                                                                                




                                                             


            
<?xml version="1.0" encoding="latin1" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd">

<chapter>
  <header>
    <copyright>
      <year>1997</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>Gen_Server Behaviour</title>
    <prepared></prepared>
    <docno></docno>
    <date></date>
    <rev></rev>
    <file>gen_server_concepts.xml</file>
  </header>
  <marker id="gen_server"></marker>
  <p>This chapter should be read in conjunction with
    <seealso marker="stdlib:gen_server">gen_server(3)</seealso>,
    where all interface functions and callback
    functions are described in detail.</p>

  <section>
    <title>Client-Server Principles</title>
    <p>The client-server model is characterized by a central server
      and an arbitrary number of clients. The client-server model is
      generally used for resource management operations, where several
      different clients want to share a common resource. The server is
      responsible for managing this resource.</p>
    <marker id="clientserver"></marker>
    <image file="../design_principles/clientserver.gif">
      <icaption>Client-Server Model</icaption>
    </image>
  </section>

  <section>
    <title>Example</title>
    <p>An example of a simple server written in plain Erlang was
      given in <seealso marker="des_princ#ch1">Overview</seealso>.
      The server can be re-implemented using <c>gen_server</c>,
      resulting in this callback module:</p>
    <marker id="ex"></marker>
    <code type="none">
-module(ch3).
-behaviour(gen_server).

-export([start_link/0]).
-export([alloc/0, free/1]).
-export([init/1, handle_call/3, handle_cast/2]).

start_link() ->
    gen_server:start_link({local, ch3}, ch3, [], []).

alloc() ->
    gen_server:call(ch3, alloc).

free(Ch) ->
    gen_server:cast(ch3, {free, Ch}).

init(_Args) ->
    {ok, channels()}.

handle_call(alloc, _From, Chs) ->
    {Ch, Chs2} = alloc(Chs),
    {reply, Ch, Chs2}.

handle_cast({free, Ch}, Chs) ->
    Chs2 = free(Ch, Chs),
    {noreply, Chs2}.</code>
    <p>The code is explained in the next sections.</p>
  </section>

  <section>
    <title>Starting a Gen_Server</title>
    <p>In the example in the previous section, the gen_server is started
      by calling <c>ch3:start_link()</c>:</p>
    <code type="none">
start_link() ->
    gen_server:start_link({local, ch3}, ch3, [], []) => {ok, Pid}</code>
    <p><c>start_link</c> calls the function
      <c>gen_server:start_link/4</c>. This function spawns and links to
      a new process, a gen_server.</p>
    <list type="bulleted">
      <item>
        <p>The first argument <c>{local, ch3}</c> specifies the name. In
          this case, the gen_server will be locally registered as
          <c>ch3</c>.</p>
        <p>If the name is omitted, the gen_server is not registered.
          Instead its pid must be used. The name could also be given
          as <c>{global, Name}</c>, in which case the gen_server is
          registered using <c>global:register_name/2</c>.</p>
      </item>
      <item>
        <p>The second argument, <c>ch3</c>, is the name of the callback
          module, that is the module where the callback functions are
          located.</p>
        <p>In this case, the interface functions (<c>start_link</c>,
          <c>alloc</c> and <c>free</c>) are located in the same module
          as the callback functions (<c>init</c>, <c>handle_call</c> and
          <c>handle_cast</c>). This is normally good programming
          practice, to have the code corresponding to one process
          contained in one module.</p>
      </item>
      <item>
        <p>The third argument, [], is a term which is passed as-is to
          the callback function <c>init</c>. Here, <c>init</c> does not
          need any indata and ignores the argument.</p>
      </item>
      <item>
        <p>The fourth argument, [], is a list of options. See
          <c>gen_server(3)</c> for available options.</p>
      </item>
    </list>
    <p>If name registration succeeds, the new gen_server process calls
      the callback function <c>ch3:init([])</c>. <c>init</c> is expected
      to return <c>{ok, State}</c>, where <c>State</c> is the internal
      state of the gen_server. In this case, the state is the available
      channels.</p>
    <code type="none">
init(_Args) ->
    {ok, channels()}.</code>
    <p>Note that <c>gen_server:start_link</c> is synchronous. It does
      not return until the gen_server has been initialized and is ready
      to receive requests.</p>
    <p><c>gen_server:start_link</c> must be used if the gen_server is
      part of a supervision tree, i.e. is started by a supervisor.
      There is another function <c>gen_server:start</c> to start a
      stand-alone gen_server, i.e. a gen_server which is not part of a
      supervision tree.</p>
  </section>

  <section>
    <title>Synchronous Requests - Call</title>
    <p>The synchronous request <c>alloc()</c> is implemented using
      <c>gen_server:call/2</c>:</p>
    <code type="none">
alloc() ->
    gen_server:call(ch3, alloc).</code>
    <p><c>ch3</c> is the name of the gen_server and must agree with
      the name used to start it. <c>alloc</c> is the actual request.</p>
    <p>The request is made into a message and sent to the gen_server.
      When the request is received, the gen_server calls
      <c>handle_call(Request, From, State)</c> which is expected to
      return a tuple <c>{reply, Reply, State1}</c>. <c>Reply</c> is
      the reply which should be sent back to the client, and
      <c>State1</c> is a new value for the state of the gen_server.</p>
    <code type="none">
handle_call(alloc, _From, Chs) ->
    {Ch, Chs2} = alloc(Chs),
    {reply, Ch, Chs2}.</code>
    <p>In this case, the reply is the allocated channel <c>Ch</c> and
      the new state is the set of remaining available channels
      <c>Chs2</c>.</p>
    <p>Thus, the call <c>ch3:alloc()</c> returns the allocated channel
      <c>Ch</c> and the gen_server then waits for new requests, now
      with an updated list of available channels.</p>
  </section>

  <section>
    <title>Asynchronous Requests - Cast</title>
    <p>The asynchronous request <c>free(Ch)</c> is implemented using
      <c>gen_server:cast/2</c>:</p>
    <code type="none">
free(Ch) ->
    gen_server:cast(ch3, {free, Ch}).</code>
    <p><c>ch3</c> is the name of the gen_server. <c>{free, Ch}</c> is
      the actual request.</p>
    <p>The request is made into a message and sent to the gen_server.
      <c>cast</c>, and thus <c>free</c>, then returns <c>ok</c>.</p>
    <p>When the request is received, the gen_server calls
      <c>handle_cast(Request, State)</c> which is expected to
      return a tuple <c>{noreply, State1}</c>. <c>State1</c> is a new
      value for the state of the gen_server.</p>
    <code type="none">
handle_cast({free, Ch}, Chs) ->
    Chs2 = free(Ch, Chs),
    {noreply, Chs2}.</code>
    <p>In this case, the new state is the updated list of available
      channels <c>Chs2</c>. The gen_server is now ready for new
      requests.</p>
  </section>

  <section>
    <title>Stopping</title>

    <section>
      <title>In a Supervision Tree</title>
      <p>If the gen_server is part of a supervision tree, no stop
        function is needed. The gen_server will automatically be
        terminated by its supervisor. Exactly how this is done is
        defined by a <seealso marker="sup_princ#shutdown">shutdown strategy</seealso> set in the supervisor.</p>
      <p>If it is necessary to clean up before termination, the shutdown
        strategy must be a timeout value and the gen_server must be set
        to trap exit signals in the <c>init</c> function. When ordered
        to shutdown, the gen_server will then call the callback function
        <c>terminate(shutdown, State)</c>:</p>
      <code type="none">
init(Args) ->
    ...,
    process_flag(trap_exit, true),
    ...,
    {ok, State}.

...

terminate(shutdown, State) ->
    ..code for cleaning up here..
    ok.</code>
    </section>

    <section>
      <title>Stand-Alone Gen_Servers</title>
      <p>If the gen_server is not part of a supervision tree, a stop
        function may be useful, for example:</p>
      <code type="none">
...
export([stop/0]).
...

stop() ->
    gen_server:cast(ch3, stop).
...

handle_cast(stop, State) ->
    {stop, normal, State};
handle_cast({free, Ch}, State) ->
    ....

...

terminate(normal, State) ->
    ok.</code>
      <p>The callback function handling the <c>stop</c> request returns
        a tuple <c>{stop, normal, State1}</c>, where <c>normal</c>
        specifies that it is a normal termination and <c>State1</c> is
        a new value for the state of the gen_server. This will cause
        the gen_server to call <c>terminate(normal,State1)</c> and then
        terminate gracefully.</p>
    </section>
  </section>

  <section>
    <title>Handling Other Messages</title>
    <p>If the gen_server should be able to receive other messages than
      requests, the callback function <c>handle_info(Info, State)</c>
      must be implemented to handle them. Examples of other messages
      are exit messages, if the gen_server is linked to other processes
      (than the supervisor) and trapping exit signals.</p>
    <code type="none">
handle_info({'EXIT', Pid, Reason}, State) ->
    ..code to handle exits here..
    {noreply, State1}.</code>
    <p>The code_change method also has to be implemented.</p>
    <code type="none">
code_change(OldVsn, State, Extra) ->
    ..code to convert state (and more) during code change
    {ok, NewState}.</code>
  </section>
</chapter>