<?xml version="1.0" encoding="iso-8859-1" ?>
<!DOCTYPE erlref SYSTEM "erlref.dtd">

<erlref>
  <header>
    <copyright>
      <year>1996</year><year>2012</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>snmp_generic</title>
    <prepared></prepared>
    <responsible></responsible>
    <docno></docno>
    <approved></approved>
    <checked></checked>
    <date></date>
    <rev></rev>
    <file>snmp_generic.xml</file>
  </header>
  <module>snmp_generic</module>
  <modulesummary>Generic Functions for Implementing SNMP Objects in a Database</modulesummary>
  <description>
    <marker id="description"></marker>
    <p>The module <c>snmp_generic</c> contains generic functions for 
    implementing tables (and variables) using the SNMP built-in database 
    or Mnesia. These default functions are used if no instrumentation 
    function is provided for a managed object in a MIB. Sometimes, 
    it might be necessary to customize the behaviour of the default 
    functions. For example, in some situations a trap should be sent 
    if a row is deleted or modified, or some hardware is to be informed, 
    when information is changed. </p>

    <p>The overall structure is shown in the following figure:</p>
    <pre>
         +---------------+
         |   SNMP Agent  |
         +- - - - - - - -+
         |      MIB      |
         +---------------+
                 |
         Association file       (associates a MIB object with
                 |               snmp_generic:table_funct
                 |               snmp_generic:variable_func)
+--------------------------------------+
|           snmp_generic               |  Support for get-next,
|                                      |  RowStatus operations
+----------------------+---------------+
|    snmpa_local_db    |    Mnesia     |  Database
+--------------+-------+---------------+
|     dets     |  ets  | 
| (persistent) |       | 
+--------------+-------+     </pre>
    <p>Each function takes the argument <c>NameDb</c>, which is a
      tuple <c>{Name, Db}</c>, to identify which database the
      functions should use. <c>Name</c> is the symbolic name of the
      managed object as defined in the MIB, and <c>Db</c> is either
      <c>volatile</c>, <c>persistent</c>, or <c>mnesia</c>. If it is
      <c>mnesia</c>, all variables are stored in the Mnesia table
      <c>snmp_variables</c> which must be a table with two attributes
      (not a Mnesia SNMP table). The SNMP tables are stored in Mnesia
      tables with the same names as the SNMP tables.  All functions
      assume that a Mnesia table exists with the correct name and
      attributes. It is the programmer's responsibility to ensure
      this. Specifically, if variables are stored in Mnesia, the table
      <c>snmp_variables</c> must be created by the programmer.  The
      record definition for this table is defined in the file
      <c>snmp/include/snmp_types.hrl</c>.
      </p>
    <p>If an instrumentation function in the association file for a
      variable <c>myVar</c> does not have a name when compiling an
      MIB, the compiler generates an entry.
      </p>
    <pre>
{myVar, {snmp_generic, variable_func, [{myVar, Db]}}.
    </pre>
    <p>And for a table:</p>
    <pre>
{myTable, {snmp_generic, table_func, [{myTable, Db]}}.
    </pre>
  </description>

  <section>
    <marker id="data_types"></marker>
    <title>DATA TYPES</title>
    <p>In the functions defined below, the following types are used:</p>
    <code type="none">
name_db() = {name(), db()} 
name() = atom()
db() = volatile | persistent | mnesia
row_index() = [int()]
columns() = [column()] | [{column(), value()}]
column() = int()
value() = term()
    </code>
    <taglist>
      <tag><c>row_index()</c></tag>
      <item>
        <p>Denotes the last part of the OID which specifies the 
          index of the row in the table (see RFC1212, 4.1.6 for 
          more information about INDEX).  </p>
      </item>
      <tag><c>columns()</c></tag>
      <item>
        <p>Is a list of column numbers in the case of a <c>get</c> 
          operation, and a list of column numbers and values in the 
          case of a <c>set</c> operation. </p>
      </item>
    </taglist>

    <marker id="get_status_col2"></marker>
  </section>

  <funcs>
    <func>
      <name>get_status_col(Name, Cols)</name>
      <name>get_status_col(NameDb, Cols) -> {ok, StatusVal} | false</name>
      <fsummary>Get the value of the status column from <c>Cols</c></fsummary>
      <type>
        <v>Name = name()</v>
        <v>NameDb = name_db()</v>
        <v>Cols = columns()</v>
        <v>StatusVal = term()</v>
      </type>
      <desc>
        <p>Gets the value of the status column from <c>Cols</c>.
          </p>
        <p>This function can be used in instrumentation functions for
          <c>is_set_ok</c>, <c>undo</c> or <c>set</c> to check if the
          status column of a table is modified.</p>
	  
	  <marker id="get_index_types"></marker>
      </desc>
    </func>

    <func>
      <name>get_index_types(Name)</name>
      <fsummary>Get the index types of <c>Name</c></fsummary>
      <type>
        <v>Name = name()</v>
      </type>
      <desc>
        <p>Gets the index types of <c>Name</c></p>
        <p>This function can be used in instrumentation functions to
	retrieve the index types part of the table info.</p>
	
	<marker id="get_table_info"></marker>
      </desc>
    </func>

    <func>
      <name>get_table_info(Name, Item) -> table_info_result()</name>
      <fsummary>Get table info item of MIB table <c>Name</c></fsummary>
      <type>
        <v>Name = name()</v>
        <v>Item = table_item() | all</v>
        <v>table_item() = nbr_of_cols | defvals | status_col | not_accessible | 
	index_types | first_accessible | first_own_index</v>
        <v>table_info_result() = Value | [{table_item(), Value}]</v>
        <v>Value = term()</v>
      </type>
      <desc>
        <p>Get a specific table info item or, if <c>Item</c> has the
	value <c>all</c>, a two tuple list (property list) is instead 
	returned with all the items and their respctive values of the 
	given table. </p>

        <p>This function can be used in instrumentation functions to
	retrieve a given part of the table info.</p>
	
	<marker id="table_func"></marker>
      </desc>
    </func>

    <func>
      <name>table_func(Op1, NameDb)</name>
      <name>table_func(Op2, RowIndex, Cols, NameDb) -> Ret</name>
      <fsummary>Default instrumentation function for tables</fsummary>
      <type>
        <v>Op1 = new | delete </v>
        <v>Op2 = get | next | is_set_ok | set | undo</v>
        <v>NameDb = name_db()</v>
        <v>RowIndex = row_index()</v>
        <v>Cols = columns()</v>
        <v>Ret = term()</v>
      </type>
      <desc>
        <p>This is the default instrumentation function for tables.
          </p>
        <list type="bulleted">
          <item>The <c>new</c> function creates the table if it does
           not exist, but only if the database is the SNMP internal db.</item>
          <item>The <c>delete</c> function does not delete the table
           from the database since unloading an MIB does not
           necessarily mean that the table should be destroyed.</item>
          <item>The <c>is_set_ok</c> function checks that a row which
           is to be modified or deleted exists, and that a row which
           is to be created does not exist.</item>
          <item>The <c>undo</c> function does nothing.</item>
          <item>The <c>set</c> function checks if it has enough
           information to make the row change its status from
          <c>notReady</c> to <c>notInService</c> (when a row has
           been been set to <c>createAndWait</c>). If a row is set to
          <c>createAndWait</c>, columns without a value are set to
          <c>noinit</c>. If Mnesia is used, the set functionality is
           handled within a transaction.</item>
        </list>
        <p>If it is possible for a manager to create or delete rows in
          the table, there must be a <c>RowStatus</c> column for
          <c>is_set_ok</c>, <c>set</c> and <c>undo</c> to work properly.
          </p>
        <p>The function returns according to the specification of an
          instrumentation function.
          </p>
	
	<marker id="table_get_elements"></marker>
      </desc>
    </func>

    <func>
      <name>table_get_elements(NameDb, RowIndex, Cols) -> Values</name>
      <fsummary>Get elements in a table row</fsummary>
      <type>
        <v>NameDb = name_db()</v>
        <v>RowIndex = row_index()</v>
        <v>Cols = columns()</v>
        <v>Values = [value() | noinit]</v>
      </type>
      <desc>
        <p>Returns a list with values for all columns in <c>Cols</c>.
          If a column is undefined, its value is <c>noinit</c>.</p>
	
	<marker id="table_next"></marker>
      </desc>
    </func>

    <func>
      <name>table_next(NameDb, RestOid) -> RowIndex | endOfTable</name>
      <fsummary>Find the next row in the table</fsummary>
      <type>
        <v>NameDb = name_db()</v>
        <v>RestOid = [int()]</v>
        <v>RowIndex = row_index()</v>
      </type>
      <desc>
        <p>Finds the indices of the next row in the table.  <c>RestOid</c>
          does not have to specify an existing row.</p>
	
	<marker id="table_row_exists"></marker>
      </desc>
    </func>

    <func>
      <name>table_row_exists(NameDb, RowIndex) -> bool()</name>
      <fsummary>Check if a row in a table exists</fsummary>
      <type>
        <v>NameDb = name_db()</v>
        <v>RowIndex = row_index()</v>
      </type>
      <desc>
        <p>Checks if a row in a table exists.</p>
	
	<marker id="table_set_elements"></marker>
      </desc>
    </func>

    <func>
      <name>table_set_elements(NameDb, RowIndex, Cols) -> bool()</name>
      <fsummary>Set elements in a table row</fsummary>
      <type>
        <v>NameDb = name_db()</v>
        <v>RowIndex = row_index()</v>
        <v>Cols = columns()</v>
      </type>
      <desc>
        <p>Sets the elements in <c>Cols</c> to the row specified by
          <c>RowIndex</c>.  No checks are performed on the new values.
          </p>
        <p>If the Mnesia database is used, this function calls
          <c>mnesia:write</c> to store the values.  This means that
          this function must be called from within a transaction
          (<c>mnesia:transaction/1</c> or <c>mnesia:dirty/1</c>).</p>
	
	<marker id="variable_func"></marker>
      </desc>
    </func>

    <func>
      <name>variable_func(Op1, NameDb)</name>
      <name>variable_func(Op2, Val, NameDb) -> Ret</name>
      <fsummary>Default instrumentation function for tables</fsummary>
      <type>
        <v>Op1 = new | delete | get</v>
        <v>Op2 = is_set_ok | set | undo</v>
        <v>NameDb = name_db()</v>
        <v>Val = value()</v>
        <v>Ret = term()</v>
      </type>
      <desc>
        <p>This is the default instrumentation function for variables.</p>
        <p>The <c>new</c> function creates a new variable in the
          database with a default value as defined in the MIB, or a zero
          value (depending on the type).  </p>
        <p>The <c>delete</c> function does not delete the variable from 
          the database. </p>
        <p>The function returns according to the specification of an 
          instrumentation function. </p>
	
	<marker id="variable_get"></marker>
      </desc>
    </func>

    <func>
      <name>variable_get(NameDb) -> {value, Value} | undefined</name>
      <fsummary>Get the value of a variable</fsummary>
      <type>
        <v>NameDb = name_db()</v>
        <v>Value = value()</v>
      </type>
      <desc>
        <p>Gets the value of a variable.</p>
	
	<marker id="variable_set"></marker>
      </desc>
    </func>

    <func>
      <name>variable_set(NameDb, NewVal) -> true | false</name>
      <fsummary>Set a value for a variable</fsummary>
      <type>
        <v>NameDb = name_db()</v>
        <v>NewVal = value()</v>
      </type>
      <desc>
        <p>Sets a new value to a variable.  The variable is created if
          it does not exist.  No checks are made on the type of the
          new value.  </p>
        <p>Returns <c>false</c> if the <c>NameDb</c> argument
          is incorrectly specified, otherwise <c>true</c>.</p>
      </desc>
    </func>
  </funcs>

  <section>
    <marker id="example"></marker>
    <title>Example</title>
    <p>The following example shows an implementation of a table which is
      stored in Mnesia, but with some checks performed at set-request
      operations.
      </p>
    <pre>
myTable_func(new, NameDb) ->   % pass unchanged
  snmp_generic:table_func(new, NameDb).

myTable_func(delete, NameDb) ->   % pass unchanged
  snmp_generic:table_func(delete, NameDb).

%% change row
myTable_func(is_set_ok, RowIndex, Cols, NameDb) ->
  case snmp_generic:table_func(is_set_ok, RowIndex,
                               Cols, NameDb) of
    {noError, 0} -> 
      myApplication:is_set_ok(RowIndex, Cols);
    Err ->
      Err
  end;

myTable_func(set, RowIndex, Cols, NameDb) ->
  case snmp_generic:table_func(set, RowIndex, Cols,
                               NameDb),
    {noError, 0} ->
      % Now the row is updated, tell the application
      myApplication:update(RowIndex, Cols);
    Err ->
      Err
  end;

myTable_func(Op, RowIndex, Cols, NameDb) ->   % pass unchanged
  snmp_generic:table_func(Op, RowIndex, Cols, NameDb).
    </pre>
    <p>The <c>.funcs</c> file would look like:
      </p>
    <pre>
{myTable, {myModule, myTable_func, [{myTable, mnesia}]}}.
    </pre>
  </section>
  
</erlref>