<?xml version="1.0" encoding="latin1" ?>
<!DOCTYPE erlref SYSTEM "erlref.dtd">
<erlref>
<header>
<copyright>
<year>1996</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>compile</title>
<prepared>Robert Virding</prepared>
<docno></docno>
<date>1996-11-04</date>
<rev>A</rev>
<file>compile.sgml</file>
</header>
<module>compile</module>
<modulesummary>Erlang Compiler</modulesummary>
<description>
<p>This module provides an interface to the standard Erlang
compiler. It can generate either a new file which contains
the object code, or return a binary which can be loaded directly.
</p>
</description>
<funcs>
<func>
<name>file(File)</name>
<fsummary>Compile a file</fsummary>
<desc>
<p>Is the same as
<c>file(File, [verbose,report_errors,report_warnings])</c>.
</p>
</desc>
</func>
<func>
<name>file(File, Options) -> CompRet</name>
<fsummary>Compile a file</fsummary>
<type>
<v>CompRet = ModRet | BinRet | ErrRet</v>
<v>ModRet = {ok,ModuleName} | {ok,ModuleName,Warnings}</v>
<v>BinRet = {ok,ModuleName,Binary} | {ok,ModuleName,Binary,Warnings}</v>
<v>ErrRet = error | {error,Errors,Warnings}</v>
</type>
<desc>
<p>Compiles the code in the file <c>File</c>, which is an
Erlang source code file without the <c>.erl</c> extension.
<c>Options</c> determine the behavior of the compiler.</p>
<p>Returns <c>{ok,ModuleName}</c> if successful, or <c>error</c>
if there are errors. An object code file is created if
the compilation succeeds with no errors. It is considered
to be an error if the module name in the source code is
not the same as the basename of the output file.</p>
<p>Here follows first all elements of <c>Options</c> that in
some way control the behavior of the compiler.</p>
<taglist>
<tag><c>basic_validation</c></tag>
<item>
<p>This option is fast way to test whether a module will
compile successfully (mainly useful for code generators
that want to verify the code they emit). No code will
generated. If warnings are enabled, warnings generated by
the <c>erl_lint</c> module (such as warnings for unused
variables and functions) will be returned too.</p>
<p>Use the <c>strong_validation</c> option to generate all
warnings that the compiler would generate.</p>
</item>
<tag><c>strong_validation</c></tag>
<item>
<p>Similar to the <c>basic_validation</c> option, no code
will be generated, but more compiler passes will be run
to ensure also warnings generated by the optimization
passes are generated (such as clauses that will not match
or expressions that are guaranteed to fail with an
exception at run-time).</p>
</item>
<tag><c>binary</c></tag>
<item>
<p>Causes the compiler to return the object code in a
binary instead of creating an object file. If successful,
the compiler returns <c>{ok,ModuleName,Binary}</c>.</p>
</item>
<tag><c>bin_opt_info</c></tag>
<item>
<p>The compiler will emit informational warnings about binary
matching optimizations (both successful and unsuccessful).
See the <em>Efficiency Guide</em> for further information.</p>
</item>
<tag><c>compressed</c></tag>
<item>
<p>The compiler will compress the generated object code,
which can be useful for embedded systems.</p>
</item>
<tag><c>debug_info</c></tag>
<item>
<marker id="debug_info"></marker>
<p>Include debug information in the form of abstract code
(see
<seealso marker="erts:absform">The Abstract Format</seealso>
in ERTS User's Guide) in the compiled beam module. Tools
such as Debugger, Xref and Cover require the debug
information to be included.</p>
<p><em>Warning</em>: Source code can be reconstructed from
the debug information. Use encrypted debug information
(see below) to prevent this.</p>
<p>See
<seealso marker="stdlib:beam_lib#debug_info">beam_lib(3)</seealso>
for details.</p>
</item>
<tag><c>{debug_info_key,KeyString}</c></tag>
<item></item>
<tag><c>{debug_info_key,{Mode,KeyString}}</c></tag>
<item>
<marker id="debug_info_key"></marker>
<p>Include debug information, but encrypt it, so that it
cannot be accessed without supplying the key. (To give
the <c>debug_info</c> option as well is allowed, but is
not necessary.) Using this option is a good way to always
have the debug information available during testing, yet
protect the source code.</p>
<p><c>Mode</c> is the type of crypto algorithm to be used
for encrypting the debug information. The default type --
and currently the only type -- is <c>des3_cbc</c>.</p>
<p>See
<seealso marker="stdlib:beam_lib#debug_info">beam_lib(3)</seealso>
for details.</p>
</item>
<tag><c>encrypt_debug_info</c></tag>
<item>
<marker id="encrypt_debug_info"></marker>
<p>Like the <c>debug_info_key</c> option above, except that
the key will be read from an <c>.erlang.crypt</c> file.
</p>
<p>See
<seealso marker="stdlib:beam_lib#debug_info">beam_lib(3)</seealso>
for details.</p>
</item>
<tag><c>'P'</c></tag>
<item>
<p>Produces a listing of the parsed code after preprocessing
and parse transforms, in the file
<c><![CDATA[<File>.P]]></c>. No object file is produced.
</p>
</item>
<tag><c>'E'</c></tag>
<item>
<p>Produces a listing of the code after all source code
transformations have been performed, in the file
<c><![CDATA[<File>.E]]></c>. No object file is produced.
</p>
</item>
<tag><c>'S'</c></tag>
<item>
<p>Produces a listing of the assembler code in the file
<c><![CDATA[<File>.S]]></c>. No object file is produced.
</p>
</item>
<tag><c>report_errors/report_warnings</c></tag>
<item>
<p>Causes errors/warnings to be printed as they occur.</p>
</item>
<tag><c>report</c></tag>
<item>
<p>This is a short form for both <c>report_errors</c> and
<c>report_warnings</c>.</p>
</item>
<tag><c>return_errors</c></tag>
<item>
<p>If this flag is set, then
<c>{error,ErrorList,WarningList}</c> is returned when
there are errors.</p>
</item>
<tag><c>return_warnings</c></tag>
<item>
<p>If this flag is set, then an extra field containing
<c>WarningList</c> is added to the tuples returned on
success.</p>
</item>
<tag><c>warnings_as_errors</c></tag>
<item>
<p>Causes warnings to be treated as errors. This option is supported
since R13B04.</p>
</item>
<tag><c>return</c></tag>
<item>
<p>This is a short form for both <c>return_errors</c> and
<c>return_warnings</c>.</p>
</item>
<tag><c>verbose</c></tag>
<item>
<p>Causes more verbose information from the compiler
describing what it is doing.</p>
</item>
<tag><c>{outdir,Dir}</c></tag>
<item>
<p>Sets a new directory for the object code. The current
directory is used for output, except when a directory
has been specified with this option.</p>
</item>
<tag><c>export_all</c></tag>
<item>
<p>Causes all functions in the module to be exported.</p>
</item>
<tag><c>{i,Dir}</c></tag>
<item>
<p>Add <c>Dir</c> to the list of directories to be searched
when including a file. When encountering an
<c>-include</c> or <c>-include_dir</c> directive,
the compiler searches for header files in the following
directories:</p>
<list type="ordered">
<item>
<p><c>"."</c>, the current working directory of
the file server;</p>
</item>
<item>
<p>the base name of the compiled file;</p>
</item>
<item>
<p>the directories specified using the <c>i</c> option.
The directory specified last is searched first.</p>
</item>
</list>
</item>
<tag><c>{d,Macro}</c></tag>
<item></item>
<tag><c>{d,Macro,Value}</c></tag>
<item>
<p>Defines a macro <c>Macro</c> to have the value
<c>Value</c>. The default is <c>true</c>).</p>
</item>
<tag><c>{parse_transform,Module}</c></tag>
<item>
<p>Causes the parse transformation function
<c>Module:parse_transform/2</c> to be applied to the
parsed code before the code is checked for errors.</p>
</item>
<tag><c>asm</c></tag>
<item>
<p>The input file is expected to be assembler code (default
file suffix ".S"). Note that the format of assembler files
is not documented, and may change between releases - this
option is primarily for internal debugging use.</p>
</item>
<tag><c>no_strict_record_tests</c></tag>
<item>
<p>This option is not recommended.</p>
<p>By default, the generated code for
the <c>Record#record_tag.field</c> operation verifies that
the tuple <c>Record</c> is of the correct size for
the record and that the first element is the tag
<c>record_tag</c>. Use this option to omit
the verification code.</p>
</item>
<tag><c>no_error_module_mismatch</c></tag>
<item>
<p>Normally the compiler verifies that the module name
given in the source code is the same as the base name
of the output file and refuses to generate an output file
if there is a mismatch. If you have a good reason (or
other reason) for having a module name unrelated to the
name of the output file, this option disables that verification
(there will not even be a warning if there is a mismatch).</p>
</item>
</taglist>
<p>If warnings are turned on (the <c>report_warnings</c> option
described above), the following options control what type of
warnings that will be generated.
<marker id="erl_lint_options"></marker>
With the exception of <c>{warn_format,Verbosity}</c> all
options below have two forms; one <c>warn_xxx</c> form to
turn on the warning and one <c>nowarn_xxx</c> form to turn off
the warning. In the description that follows, the form that
is used to change the default value is listed.</p>
<taglist>
<tag><c>{warn_format, Verbosity}</c></tag>
<item>
<p>Causes warnings to be emitted for malformed format
strings as arguments to <c>io:format</c> and similar
functions. <c>Verbosity</c> selects the amount of
warnings: 0 = no warnings; 1 = warnings for invalid
format strings and incorrect number of arguments; 2 =
warnings also when the validity could not be checked
(for example, when the format string argument is a
variable). The default verbosity is 1. Verbosity 0 can
also be selected by the option <c>nowarn_format</c>.</p>
</item>
<tag><c>nowarn_bif_clash</c></tag>
<item>
<p>By default, there will be a compilation error if a
module contains an exported function with the same name
as an auto-imported BIF (such as <c>size/1</c>) AND
there is a call to it without a qualifying module name.
The reason is that the BIF will be called, not
the function in the same module. The recommended way to
eliminate that warning is to use a call with a module
name - either <c>erlang</c> to call the BIF or
<c>?MODULE</c> to call the function in the same module.
The warning can also be turned off using this option,
but that is not recommended.</p>
<p><em>The use of this option is strongly discouraged,
as code that uses it will probably break in a future
major release (R14 or R15).</em></p>
</item>
<tag><c>{nowarn_bif_clash, FAs}</c></tag>
<item>
<p>Turns off warnings as <c>nowarn_bif_clash</c> but only
for the mentioned local functions. <c>FAs</c> is a tuple
<c>{Name,Arity}</c> or a list of such tuples.</p>
<p><em>The use of this option is strongly discouraged,
as code that uses it will probably break in a future
major release (R14 or R15).</em></p>
</item>
<tag><c>warn_export_all</c></tag>
<item>
<p>Causes a warning to be emitted if the <c>export_all</c>
option has also been given.</p>
</item>
<tag><c>warn_export_vars</c></tag>
<item>
<p>Causes warnings to be emitted for all implicitly
exported variables referred to after the primitives
where they were first defined. No warnings for exported
variables unless they are referred to in some pattern,
which is the default, can be selected by the option
<c>nowarn_export_vars</c>.</p>
</item>
<tag><c>warn_shadow_vars</c></tag>
<item>
<p>Causes warnings to be emitted for "fresh" variables
in functional objects or list comprehensions with the same
name as some already defined variable. The default is to
warn for such variables. No warnings for shadowed
variables can be selected by the option
<c>nowarn_shadow_vars</c>.</p>
</item>
<tag><c>nowarn_unused_function</c></tag>
<item>
<p>Turns off warnings for unused local functions.
By default (<c>warn_unused_function</c>), warnings are
emitted for all local functions that are not called
directly or indirectly by an exported function.
The compiler does not include unused local functions in
the generated beam file, but the warning is still useful
to keep the source code cleaner.</p>
</item>
<tag><c>{nowarn_unused_function, FAs}</c></tag>
<item>
<p>Turns off warnings for unused local functions as
<c>nowarn_unused_function</c> but only for the mentioned
local functions. <c>FAs</c> is a tuple <c>{Name,Arity}</c>
or a list of such tuples.</p>
</item>
<tag><c>nowarn_deprecated_function</c></tag>
<item>
<p>Turns off warnings for calls to deprecated functions. By
default (<c>warn_deprecated_function</c>), warnings are
emitted for every call to a function known by the compiler
to be deprecated. Note that the compiler does not know
about the <c>-deprecated()</c> attribute but uses an
assembled list of deprecated functions in Erlang/OTP. To
do a more general check the <c>Xref</c> tool can be used.
See also
<seealso marker="tools:xref#deprecated_function">xref(3)</seealso>
and the function
<seealso marker="tools:xref#m/1">xref:m/1</seealso> also
accessible through
the <seealso marker="stdlib:c#xm/1">c:xm/1</seealso>
function.</p>
</item>
<tag><c>{nowarn_deprecated_function, MFAs}</c></tag>
<item>
<p>Turns off warnings for calls to deprecated functions as
<c>nowarn_deprecated_function</c> but only for
the mentioned functions. <c>MFAs</c> is a tuple
<c>{Module,Name,Arity}</c> or a list of such tuples.</p>
</item>
<tag><c>warn_obsolete_guard</c></tag>
<item>
<p>Causes warnings to be emitted for calls to old type
testing BIFs such as <c>pid/1</c> and <c>list/1</c>. See
the
<seealso marker="doc/reference_manual:expressions#guards">Erlang Reference Manual</seealso>
for a complete list of type testing BIFs and their old
equivalents. No warnings for calls to old type testing
BIFs, which is the default, can be selected by the option
<c>nowarn_obsolete_guard</c>.</p>
</item>
<tag><c>warn_unused_import</c></tag>
<item>
<p>Causes warnings to be emitted for unused imported
functions. No warnings for unused imported functions,
which is the default, can be selected by the option
<c>nowarn_unused_import</c>. </p>
</item>
<tag><c>nowarn_unused_vars</c></tag>
<item>
<p>By default, warnings are emitted for variables which
are not used, with the exception of variables beginning
with an underscore ("Prolog style warnings").
Use this option to turn off this kind of warnings.</p>
</item>
<tag><c>nowarn_unused_record</c></tag>
<item>
<p>Turns off warnings for unused record types. By
default (<c>warn_unused_records</c>), warnings are
emitted for unused locally defined record types.</p>
</item>
</taglist>
<p>Another class of warnings is generated by the compiler
during optimization and code generation. They warn about
patterns that will never match (such as <c>a=b</c>), guards
that will always evaluate to false, and expressions that will
always fail (such as <c>atom+42</c>).</p>
<p>Note that the compiler does not warn for expressions that it
does not attempt to optimize. For instance, the compiler tries
to evaluate <c>1/0</c>, notices that it will cause an
exception and emits a warning. On the other hand,
the compiler is silent about the similar expression
<c>X/0</c>; because of the variable in it, the compiler does
not even try to evaluate and therefore it emits no warnings.
</p>
<p>Currently, those warnings cannot be disabled (except by
disabling all warnings).</p>
<warning>
<p>Obviously, the absence of warnings does not mean that
there are no remaining errors in the code.</p>
</warning>
<p>Note that all the options except the include path
(<c>{i,Dir}</c>) can also be given in the file with a
<c>-compile([Option,...])</c>. attribute.
The <c>-compile()</c> attribute is allowed after function
definitions.</p>
<p>Note also that the <c>{nowarn_unused_function, FAs}</c>,
<c>{nowarn_bif_clash, FAs}</c>, and
<c>{nowarn_deprecated_function, MFAs}</c> options are only
recognized when given in files. They are not affected by
the <c>warn_unused_function</c>, <c>warn_bif_clash</c>, or
<c>warn_deprecated_function</c> options.</p>
<p>For debugging of the compiler, or for pure curiosity,
the intermediate code generated by each compiler pass can be
inspected.
A complete list of the options to produce list files can be
printed by typing <c>compile:options()</c> at the Erlang
shell prompt.
The options will be printed in order that the passes are
executed. If more than one listing option is used, the one
representing the earliest pass takes effect.</p>
<p><em>Unrecognized options are ignored.</em></p>
<p>Both <c>WarningList</c> and <c>ErrorList</c> have
the following format:</p>
<code>
[{FileName,[ErrorInfo]}].
</code>
<p><c>ErrorInfo</c> is described below. The file name has been
included here as the compiler uses the Erlang pre-processor
<c>epp</c>, which allows the code to be included in other
files. For this reason, it is important to know to
<em>which</em> file an error or warning line number refers.
</p>
</desc>
</func>
<func>
<name>forms(Forms)</name>
<fsummary>Compile a list of forms</fsummary>
<desc>
<p>Is the same as
<c>forms(File, [verbose,report_errors,report_warnings])</c>.
</p>
</desc>
</func>
<func>
<name>forms(Forms, Options) -> CompRet</name>
<fsummary>Compile a list of forms</fsummary>
<type>
<v>Forms = [Form]</v>
<v>CompRet = BinRet | ErrRet</v>
<v>BinRet = {ok,ModuleName,BinaryOrCode} | {ok,ModuleName,BinaryOrCode,Warnings}</v>
<v>BinaryOrCode = binary() | term()</v>
<v>ErrRet = error | {error,Errors,Warnings}</v>
</type>
<desc>
<p>Analogous to <c>file/1</c>, but takes a list of forms (in
the Erlang abstract format representation) as first argument.
The option <c>binary</c> is implicit; i.e., no object code
file is produced. Options that would ordinarily produce a
listing file, such as 'E', will instead cause the internal
format for that compiler pass (an Erlang term; usually not a
binary) to be returned instead of a binary.</p>
</desc>
</func>
<func>
<name>format_error(ErrorDescriptor) -> chars()</name>
<fsummary>Format an error descriptor</fsummary>
<type>
<v>ErrorDescriptor = errordesc()</v>
</type>
<desc>
<p>Uses an <c>ErrorDescriptor</c> and returns a deep list of
characters which describes the error. This function is
usually called implicitly when an <c>ErrorInfo</c> structure
is processed. See below.</p>
</desc>
</func>
<func>
<name>output_generated(Options) -> true | false</name>
<fsummary>Determine whether the compile will generate an output file</fsummary>
<type>
<v>Options = [term()]</v>
</type>
<desc>
<p>Determines whether the compiler would generate a <c>beam</c>
file with the given options. <c>true</c> means that a <c>beam</c>
file would be generated; <c>false</c> means that the compiler
would generate some listing file, return a binary, or merely
check the syntax of the source code.</p>
</desc>
</func>
<func>
<name>noenv_file(File, Options) -> CompRet</name>
<fsummary>Compile a file (ignoring ERL_COMPILER_OPTIONS)</fsummary>
<desc>
<p>Works exactly like <seealso marker="#file/2">file/2</seealso>,
except that the environment variable <c>ERL_COMPILER_OPTIONS</c>
is not consulted.</p>
</desc>
</func>
<func>
<name>noenv_forms(Forms, Options) -> CompRet</name>
<fsummary>Compile a list of forms (ignoring ERL_COMPILER_OPTIONS)</fsummary>
<desc>
<p>Works exactly like <seealso marker="#forms/2">forms/2</seealso>,
except that the environment variable <c>ERL_COMPILER_OPTIONS</c>
is not consulted.</p>
</desc>
</func>
<func>
<name>noenv_output_generated(Options) -> true | false</name>
<fsummary>Determine whether the compile will generate an output file (ignoring ERL_COMPILER_OPTIONS)</fsummary>
<type>
<v>Options = [term()]</v>
</type>
<desc>
<p>Works exactly like
<seealso marker="#output_generated/1">output_generated/1</seealso>,
except that the environment variable <c>ERL_COMPILER_OPTIONS</c>
is not consulted.</p>
</desc>
</func>
</funcs>
<section>
<title>Default compiler options</title>
<p>The (host operating system) environment variable
<c>ERL_COMPILER_OPTIONS</c> can be used to give default compiler
options. Its value must be a valid Erlang term. If the value is a
list, it will be used as is. If it is not a list, it will be put
into a list.</p>
<p>The list will be appended to any options given to
<seealso marker="#file/2">file/2</seealso>,
<seealso marker="#forms/2">forms/2</seealso>, and
<seealso marker="#output_generated/1">output_generated/2</seealso>.
Use the alternative functions
<seealso marker="#noenv_file/2">noenv_file/2</seealso>,
<seealso marker="#noenv_forms/2">noenv_forms/2</seealso>, or
<seealso marker="#noenv_output_generated/1">noenv_output_generated/2</seealso>
if you don't want the environment variable to be consulted
(for instance, if you are calling the compiler recursively from
inside a parse transform).</p>
</section>
<section>
<title>Inlining</title>
<p>The compiler can do function inlining within an Erlang
module. Inlining means that a call to a function is replaced with
the function body with the arguments replaced with the actual
values. The semantics are preserved, except if exceptions are
generated in the inlined code. Exceptions will be reported as
occurring in the function the body was inlined into. Also,
<c>function_clause</c> exceptions will be converted to similar
<c>case_clause</c> exceptions.</p>
<p>When a function is inlined, the original function will be
kept if it is exported (either by an explicit export or if the
<c>export_all</c> option was given) or if not all calls to the
function were inlined.</p>
<p>Inlining does not necessarily improve running time.
For instance, inlining may increase Beam stack usage which will
probably be detrimental to performance for recursive functions.
</p>
<p>Inlining is never default; it must be explicitly enabled with a
compiler option or a <c>-compile()</c> attribute in the source
module.</p>
<p>To enable inlining, either use the <c>inline</c> option to
let the compiler decide which functions to inline or
<c>{inline,[{Name,Arity},...]}</c> to have the compiler inline
all calls to the given functions. If the option is given inside
a <c>compile</c> directive in an Erlang module, <c>{Name,Arity}</c>
may be written as <c>Name/Arity</c>.</p>
<p>Example of explicit inlining:</p>
<pre>
-compile({inline,[pi/0]}).
pi() -> 3.1416.
</pre>
<p>Example of implicit inlining:</p>
<pre>
-compile(inline).
</pre>
<p>The <c>{inline_size,Size}</c> option controls how large functions
that are allowed to be inlined. Default is <c>24</c>, which will
keep the size of the inlined code roughly the same as
the un-inlined version (only relatively small functions will be
inlined).</p>
<p>Example:</p>
<pre>
%% Aggressive inlining - will increase code size.
-compile(inline).
-compile({inline_size,100}).
</pre>
</section>
<section>
<title>Parse Transformations</title>
<p>Parse transformations are used when a programmer wants to use
Erlang syntax but with different semantics. The original Erlang
code is then transformed into other Erlang code.</p>
</section>
<section>
<title>Error Information</title>
<p>The <c>ErrorInfo</c> mentioned above is the standard
<c>ErrorInfo</c> structure which is returned from all IO modules.
It has the following format:</p>
<code>
{ErrorLine, Module, ErrorDescriptor}
</code>
<p>A string describing the error is obtained with the following
call:</p>
<code>
apply(Module, format_error, ErrorDescriptor)
</code>
</section>
<section>
<title>See Also</title>
<p>
<seealso marker="stdlib:epp">epp(3)</seealso>,
<seealso marker="stdlib:erl_id_trans">erl_id_trans(3)</seealso>,
<seealso marker="stdlib:erl_lint">erl_lint(3)</seealso>,
<seealso marker="stdlib:beam_lib">beam_lib(3)</seealso>
</p>
</section>
</erlref>
