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

<chapter>
  <header>
    <copyright>
      <year>2001</year><year>2010</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>Common Caveats</title>
    <prepared>Bjorn Gustavsson</prepared>
    <docno></docno>
    <date>2001-08-08</date>
    <rev></rev>
    <file>commoncaveats.xml</file>
  </header>

  <p>Here we list a few modules and BIFs to watch out for, and not only
  from a performance point of view.</p>

  <section>
     <title>The regexp module</title>

     <p>The regular expression functions in the
        <seealso marker="stdlib:regexp">regexp</seealso>
	module are written in Erlang, not in C, and were
	meant for occasional use on small amounts of data,
	for instance for validation of configuration files
	when starting an application.</p>

      <p>Use the <seealso marker="stdlib:re">re</seealso> module
      (introduced in R13A) instead, especially in time-critical code.</p>
  </section>

  <section>
     <title>The timer module</title>

     <p>Creating timers using <seealso
     marker="erts:erlang#erlang:send_after/3">erlang:send_after/3</seealso>
     and <seealso marker="erts:erlang#erlang:start_timer/3">erlang:start_timer/3</seealso>
     is much more efficient than using the timers provided by the
     <seealso marker="stdlib:timer">timer</seealso> module. The
     <c>timer</c> module uses a separate process to manage the timers,
     and that process can easily become overloaded if many processes
     create and cancel timers frequently (especially when using the
     SMP emulator).</p>

     <p>The functions in the <c>timer</c> module that do not manage timers (such as
     <c>timer:tc/3</c> or <c>timer:sleep/1</c>), do not call the timer-server process
     and are therefore harmless.</p>
  </section>

  <section>
    <title>list_to_atom/1</title>

    <p>Atoms are not garbage-collected. Once an atom is created, it will never
    be removed. The emulator will terminate if the limit for the number
    of atoms (1048576 by default) is reached.</p>

    <p>Therefore, converting arbitrary input strings to atoms could be
    dangerous in a system that will run continuously.
    If only certain well-defined atoms are allowed as input, you can use
    <seealso marker="erts:erlang#list_to_existing_atom/1">list_to_existing_atom/1</seealso>
    to guard against a denial-of-service attack. (All atoms that are allowed
    must have been created earlier, for instance by simply using all of them
    in a module and loading that module.)</p>

    <p>Using <c>list_to_atom/1</c> to construct an atom that is passed to
    <c>apply/3</c> like this</p>

    <code type="erl">
apply(list_to_atom("some_prefix"++Var), foo, Args)</code>    

    <p>is quite expensive and is not recommended in time-critical code.</p>
  </section>

  <section>
    <title>length/1</title>

    <p>The time for calculating the length of a list is proportional to the
    length of the list, as opposed to <c>tuple_size/1</c>, <c>byte_size/1</c>,
    and <c>bit_size/1</c>, which all execute in constant time.</p>

    <p>Normally you don't have to worry about the speed of <c>length/1</c>,
    because it is efficiently implemented in C. In time critical-code, though,
    you might want to avoid it if the input list could potentially be very long.</p>

    <p>Some uses of <c>length/1</c> can be replaced by matching.
    For instance, this code</p>

    <code type="erl">
foo(L) when length(L) >= 3 ->
    ...</code>    

    <p>can be rewritten to</p>
    <code type="erl">
foo([_,_,_|_]=L) ->
   ...</code>    

   <p>(One slight difference is that <c>length(L)</c> will fail if the <c>L</c>
   is an improper list, while the pattern in the second code fragment will
   accept an improper list.)</p>
  </section>

  <section>
    <title>setelement/3</title>

    <p><seealso marker="erts:erlang#setelement/3">setelement/3</seealso>
    copies the tuple it modifies. Therefore, updating a tuple in a loop
    using <c>setelement/3</c> will create a new copy of the tuple every time.</p>
    
    <p>There is one exception to the rule that the tuple is copied.
    If the compiler clearly can see that destructively updating the tuple would
    give exactly the same result as if the tuple was copied, the call to
    <c>setelement/3</c> will be replaced with a special destructive setelement
    instruction. In the following code sequence</p>

    <code type="erl">
multiple_setelement(T0) ->
    T1 = setelement(9, T0, bar),
    T2 = setelement(7, T1, foobar),
    setelement(5, T2, new_value).</code>    

    <p>the first <c>setelement/3</c> call will copy the tuple and modify the
    ninth element. The two following <c>setelement/3</c> calls will modify
    the tuple in place.</p>

    <p>For the optimization to be applied, <em>all</em> of the followings conditions
    must be true:</p>

    <list type="bulleted">
    <item>The indices must be integer literals, not variables or expressions.</item>
    <item>The indices must be given in descending order.</item>
    <item>There must be no calls to other function in between the calls to
    <c>setelement/3</c>.</item>
    <item>The tuple returned from one <c>setelement/3</c> call must only be used
    in the subsequent call to <c>setelement/3</c>.</item>
    </list>

    <p>If it is not possible to structure the code as in the <c>multiple_setelement/1</c>
    example, the best way to modify multiple elements in a large tuple is to
    convert the tuple to a list, modify the list, and convert the list back to
    a tuple.</p>
  </section>

  <section>
    <title>size/1</title>

    <p><c>size/1</c> returns the size for both tuples and binary.</p>

    <p>Using the new BIFs <c>tuple_size/1</c> and <c>byte_size/1</c> introduced
    in R12B gives the compiler and run-time system more opportunities for
    optimization. A further advantage is that the new BIFs could help Dialyzer
    find more bugs in your program.</p>
  </section>

  <section>
    <title>split_binary/2</title>
      <p>It is usually more efficient to split a binary using matching
      instead of calling the <c>split_binary/2</c> function.
      Furthermore, mixing bit syntax matching and <c>split_binary/2</c>
      may prevent some optimizations of bit syntax matching.</p>

        <p><em>DO</em></p>
        <code type="none"><![CDATA[
        <<Bin1:Num/binary,Bin2/binary>> = Bin,]]></code>
        <p><em>DO NOT</em></p>
        <code type="none">
        {Bin1,Bin2} = split_binary(Bin, Num)
        </code>
   </section>

  <section>
    <title>The '--' operator</title>
     <p>Note that the '<c>--</c>' operator has a complexity
     proportional to the product of the length of its operands,
     meaning that it will be very slow if both of its operands
     are long lists:</p>

        <p><em>DO NOT</em></p>
        <code type="none"><![CDATA[
        HugeList1 -- HugeList2]]></code>

     <p>Instead use the <seealso marker="stdlib:ordsets">ordsets</seealso>
     module:</p>

        <p><em>DO</em></p>
        <code type="none">
        HugeSet1 = ordsets:from_list(HugeList1),
        HugeSet2 = ordsets:from_list(HugeList2),
        ordsets:subtract(HugeSet1, HugeSet2)
        </code>

     <p>Obviously, that code will not work if the original order
     of the list is important. If the order of the list must be
     preserved, do like this:</p>

        <p><em>DO</em></p>
        <code type="none"><![CDATA[
        Set = gb_sets:from_list(HugeList2),
        [E || E <- HugeList1, not gb_sets:is_element(E, Set)]]]></code>

     <p>Subtle note 1: This code behaves differently from '<c>--</c>'
     if the lists contain duplicate elements. (One occurrence
     of an element in HugeList2 will remove <em>all</em>
     occurrences in HugeList1.)</p>

     <p>Subtle note 2: This code compares lists elements using the
     '<c>==</c>' operator, while '<c>--</c>' uses the '<c>=:=</c>'. If
     that difference is important, <c>sets</c> can be used instead of
     <c>gb_sets</c>, but note that <c>sets:from_list/1</c> is much
     slower than <c>gb_sets:from_list/1</c> for long lists.</p>

     <p>Using the '<c>--</c>' operator to delete an element
     from a list is not a performance problem:</p>

        <p><em>OK</em></p>
        <code type="none">
        HugeList1 -- [Element]
        </code>

   </section>

</chapter>