<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
<meta name="generator" content="hevea 2.32">
<meta name="viewport" content="width=device-width, initial-scale=1.0, maximum-scale=1">
<link rel="stylesheet" type="text/css" href="manual.css">
<title>Chapter 12  Native-code compilation (ocamlopt)</title>
</head>
<body>
<a href="runtime.html"><img src="previous_motif.svg" alt="Previous"></a>
<a href="index.html"><img src="contents_motif.svg" alt="Up"></a>
<a href="lexyacc.html"><img src="next_motif.svg" alt="Next"></a>
<hr>
<h1 class="chapter" id="sec304">Chapter 12  Native-code compilation (ocamlopt)</h1>
<ul>
<li><a href="native.html#sec305">12.1  Overview of the compiler</a>
</li><li><a href="native.html#sec306">12.2  Options</a>
</li><li><a href="native.html#sec310">12.3  Common errors</a>
</li><li><a href="native.html#sec311">12.4  Running executables produced by ocamlopt</a>
</li><li><a href="native.html#sec312">12.5  Compatibility with the bytecode compiler</a>
</li></ul>
<p> <a id="c:nativecomp"></a>
</p><p>This chapter describes the OCaml high-performance
native-code compiler <span class="c003">ocamlopt</span>, which compiles OCaml source files to
native code object files and links these object files to produce
standalone executables.</p><p>The native-code compiler is only available on certain platforms.
It produces code that runs faster than the bytecode produced by
<span class="c003">ocamlc</span>, at the cost of increased compilation time and executable code
size. Compatibility with the bytecode compiler is extremely high: the
same source code should run identically when compiled with <span class="c003">ocamlc</span> and
<span class="c003">ocamlopt</span>.</p><p>It is not possible to mix native-code object files produced by <span class="c003">ocamlopt</span>
with bytecode object files produced by <span class="c003">ocamlc</span>: a program must be
compiled entirely with <span class="c003">ocamlopt</span> or entirely with <span class="c003">ocamlc</span>. Native-code
object files produced by <span class="c003">ocamlopt</span> cannot be loaded in the toplevel
system <span class="c003">ocaml</span>.</p>
<h2 class="section" id="sec305">12.1  Overview of the compiler</h2>
<p>The <span class="c003">ocamlopt</span> command has a command-line interface very close to that
of <span class="c003">ocamlc</span>. It accepts the same types of arguments, and processes them
sequentially, after all options have been processed:</p><ul class="itemize"><li class="li-itemize">
Arguments ending in <span class="c003">.mli</span> are taken to be source files for
compilation unit interfaces. Interfaces specify the names exported by
compilation units: they declare value names with their types, define
public data types, declare abstract data types, and so on. From the
file <span class="c009">x</span><span class="c003">.mli</span>, the <span class="c003">ocamlopt</span> compiler produces a compiled interface
in the file <span class="c009">x</span><span class="c003">.cmi</span>. The interface produced is identical to that
produced by the bytecode compiler <span class="c003">ocamlc</span>.</li><li class="li-itemize">Arguments ending in <span class="c003">.ml</span> are taken to be source files for compilation
unit implementations. Implementations provide definitions for the
names exported by the unit, and also contain expressions to be
evaluated for their side-effects. From the file <span class="c009">x</span><span class="c003">.ml</span>, the <span class="c003">ocamlopt</span>
compiler produces two files: <span class="c009">x</span><span class="c003">.o</span>, containing native object code,
and <span class="c009">x</span><span class="c003">.cmx</span>, containing extra information for linking and
optimization of the clients of the unit. The compiled implementation
should always be referred to under the name <span class="c009">x</span><span class="c003">.cmx</span> (when given
a <span class="c003">.o</span> or <span class="c003">.obj</span> file, <span class="c003">ocamlopt</span> assumes that it contains code compiled from C,
not from OCaml).<p>The implementation is checked against the interface file <span class="c009">x</span><span class="c003">.mli</span>
(if it exists) as described in the manual for <span class="c003">ocamlc</span>
(chapter <a href="comp.html#c%3Acamlc">9</a>).</p></li><li class="li-itemize">Arguments ending in <span class="c003">.cmx</span> are taken to be compiled object code. These
files are linked together, along with the object files obtained
by compiling <span class="c003">.ml</span> arguments (if any), and the OCaml standard
library, to produce a native-code executable program. The order in
which <span class="c003">.cmx</span> and <span class="c003">.ml</span> arguments are presented on the command line is
relevant: compilation units are initialized in that order at
run-time, and it is a link-time error to use a component of a unit
before having initialized it. Hence, a given <span class="c009">x</span><span class="c003">.cmx</span> file must come
before all <span class="c003">.cmx</span> files that refer to the unit <span class="c009">x</span>.</li><li class="li-itemize">Arguments ending in <span class="c003">.cmxa</span> are taken to be libraries of object code.
Such a library packs in two files (<span class="c009">lib</span><span class="c003">.cmxa</span> and <span class="c009">lib</span><span class="c003">.a</span>/<span class="c003">.lib</span>)
a set of object files (<span class="c003">.cmx</span> and <span class="c003">.o</span>/<span class="c003">.obj</span> files). Libraries are build with
<span class="c003">ocamlopt -a</span> (see the description of the <span class="c003">-a</span> option below). The object
files contained in the library are linked as regular <span class="c003">.cmx</span> files (see
above), in the order specified when the library was built. The only
difference is that if an object file contained in a library is not
referenced anywhere in the program, then it is not linked in.</li><li class="li-itemize">Arguments ending in <span class="c003">.c</span> are passed to the C compiler, which generates
a <span class="c003">.o</span>/<span class="c003">.obj</span> object file. This object file is linked with the program.</li><li class="li-itemize">Arguments ending in <span class="c003">.o</span>, <span class="c003">.a</span> or <span class="c003">.so</span> (<span class="c003">.obj</span>, <span class="c003">.lib</span> and <span class="c003">.dll</span>
under Windows) are assumed to be C object files and
libraries. They are linked with the program.</li></ul><p>The output of the linking phase is a regular Unix or Windows
executable file. It does not need <span class="c003">ocamlrun</span> to run.</p>
<h2 class="section" id="sec306">12.2  Options</h2>
<p>The following command-line options are recognized by <span class="c003">ocamlopt</span>.
The options <span class="c003">-pack</span>, <span class="c003">-a</span>, <span class="c003">-shared</span>, <span class="c003">-c</span> and <span class="c003">-output-obj</span> are mutually
exclusive.</p><dl class="description"><dt class="dt-description">
<span class="c006">-a</span></dt><dd class="dd-description">
Build a library(<span class="c003">.cmxa</span> and <span class="c003">.a</span>/<span class="c003">.lib</span> files)
with the object files (<span class="c003">.cmx</span> and <span class="c003">.o</span>/<span class="c003">.obj</span> files)
given on the command line, instead of linking them into an executable file.
The name of the library must be set with the <span class="c003">-o</span> option.<p>If <span class="c003">-cclib</span> or <span class="c003">-ccopt</span> options are passed on the command
line, these options are stored in the resulting <span class="c003">.cmxa</span>library. Then,
linking with this library automatically adds back the
<span class="c003">-cclib</span> and <span class="c003">-ccopt</span> options as if they had been provided on the
command line, unless the <span class="c003">-noautolink</span> option is given.
</p></dd><dt class="dt-description"><span class="c006">-absname</span></dt><dd class="dd-description">
Force error messages to show absolute paths for file names.</dd><dt class="dt-description"><span class="c006">-annot</span></dt><dd class="dd-description">
Dump detailed information about the compilation (types, bindings,
tail-calls, etc). The information for file <span class="c009">src</span><span class="c003">.ml</span>
is put into file <span class="c009">src</span><span class="c003">.annot</span>. In case of a type error, dump
all the information inferred by the type-checker before the error.
The <span class="c009">src</span><span class="c003">.annot</span> file can be used with the emacs commands given in
<span class="c003">emacs/caml-types.el</span> to display types and other annotations
interactively.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-args</span> <span class="c009">filename</span></span></dt><dd class="dd-description">
Read additional newline-terminated command line arguments from <span class="c009">filename</span>.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-args0</span> <span class="c009">filename</span></span></dt><dd class="dd-description">
Read additional null character terminated command line arguments from <span class="c009">filename</span>.
</dd><dt class="dt-description"><span class="c006">-bin-annot</span></dt><dd class="dd-description">
Dump detailed information about the compilation (types, bindings,
tail-calls, etc) in binary format. The information for file <span class="c009">src</span><span class="c003">.ml</span>
(resp. <span class="c009">src</span><span class="c003">.mli</span>) is put into file <span class="c009">src</span><span class="c003">.cmt</span>
(resp. <span class="c009">src</span><span class="c003">.cmti</span>). In case of a type error, dump
all the information inferred by the type-checker before the error.
The <span class="c003">*.cmt</span> and <span class="c003">*.cmti</span> files produced by <span class="c003">-bin-annot</span> contain
more information and are much more compact than the files produced by
<span class="c003">-annot</span>.
</dd><dt class="dt-description"><span class="c006">-c</span></dt><dd class="dd-description">
Compile only. Suppress the linking phase of the
compilation. Source code files are turned into compiled files, but no
executable file is produced. This option is useful to
compile modules separately.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-cc</span> <span class="c009">ccomp</span></span></dt><dd class="dd-description">
Use <span class="c009">ccomp</span> as the C linker called to build the final executable
and as the C compiler for compiling <span class="c003">.c</span> source files.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-cclib</span> <span class="c003">-l</span><span class="c009">libname</span></span></dt><dd class="dd-description">
Pass the <span class="c003">-l</span><span class="c009">libname</span> option to the linker
.
This causes the given C library to be linked with the program.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-ccopt</span> <span class="c009">option</span></span></dt><dd class="dd-description">
Pass the given option to the C compiler and linker.
For instance,<span class="c003">-ccopt -L</span><span class="c009">dir</span> causes the C linker to search for C libraries in
directory <span class="c009">dir</span>.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-color</span> <span class="c009">mode</span></span></dt><dd class="dd-description">
Enable or disable colors in compiler messages (especially warnings and errors).
The following modes are supported:
<dl class="description"><dt class="dt-description">
<span class="c006">auto</span></dt><dd class="dd-description"> use heuristics to enable colors only if the output supports them (an ANSI-compatible tty terminal);
</dd><dt class="dt-description"><span class="c006">always</span></dt><dd class="dd-description"> enable colors unconditionally;
</dd><dt class="dt-description"><span class="c006">never</span></dt><dd class="dd-description"> disable color output.
</dd></dl>
The default setting is ’auto’, and the current heuristic
checks that the <span class="c003">TERM</span> environment variable exists and is
not empty or <span class="c003">dumb</span>, and that ’isatty(stderr)’ holds.<p>The environment variable <span class="c003">OCAML_COLOR</span> is considered if <span class="c003">-color</span> is not
provided. Its values are auto/always/never as above.
</p></dd><dt class="dt-description"><span class="c006">-compact</span></dt><dd class="dd-description">
Optimize the produced code for space rather than for time. This
results in slightly smaller but slightly slower programs. The default is to
optimize for speed.
</dd><dt class="dt-description"><span class="c006">-config</span></dt><dd class="dd-description">
Print the version number of <span class="c003">ocamlopt</span> and a detailed
summary of its configuration, then exit.</dd><dt class="dt-description"><span class="c013"><span class="c003">-depend</span> <span class="c009">ocamldep-args</span></span></dt><dd class="dd-description">
Compute dependencies, as the <span class="c003">ocamldep</span> command would do. The remaining
arguments are interpreted as if they were given to the <span class="c003">ocamldep</span> command.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-for-pack</span> <span class="c009">module-path</span></span></dt><dd class="dd-description">
Generate an object file (<span class="c003">.cmx</span> and <span class="c003">.o</span>/<span class="c003">.obj</span> files)
that can later be included
as a sub-module (with the given access path) of a compilation unit
constructed with <span class="c003">-pack</span>. For instance,
<span class="c003">ocamlopt -for-pack P -c A.ml</span>
will generate <span class="c003">a..cmx</span> and <span class="c003">a.o</span> files that can
later be used with <span class="c003">ocamlopt -pack -o P.cmx a.cmx</span>.
Note: you can still pack a module that was compiled without
<span class="c003">-for-pack</span> but in this case exceptions will be printed with the wrong
names.
</dd><dt class="dt-description"><span class="c006">-g</span></dt><dd class="dd-description">
Add debugging information while compiling and linking. This option is
required in order to produce stack backtraces when
the program terminates on an uncaught exception (see
section <a href="runtime.html#ocamlrun-options">11.2</a>).
</dd><dt class="dt-description"><span class="c006">-i</span></dt><dd class="dd-description">
Cause the compiler to print all defined names (with their inferred
types or their definitions) when compiling an implementation (<span class="c003">.ml</span>
file). No compiled files (<span class="c003">.cmo</span> and <span class="c003">.cmi</span> files) are produced.
This can be useful to check the types inferred by the
compiler. Also, since the output follows the syntax of interfaces, it
can help in writing an explicit interface (<span class="c003">.mli</span> file) for a file:
just redirect the standard output of the compiler to a <span class="c003">.mli</span> file,
and edit that file to remove all declarations of unexported names.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-I</span> <span class="c009">directory</span></span></dt><dd class="dd-description">
Add the given directory to the list of directories searched for
compiled interface files (<span class="c003">.cmi</span>), compiled object code files (<span class="c003">.cmx</span>),
and libraries (<span class="c003">.cmxa</span>).
By default, the current directory is searched first, then the standard
library directory. Directories added with <span class="c003">-I</span> are searched after the
current directory, in the order in which they were given on the command line,
but before the standard library directory. See also option <span class="c003">-nostdlib</span>.<p>If the given directory starts with <span class="c003">+</span>, it is taken relative to the
standard library directory. For instance, <span class="c003">-I +unix</span> adds the
subdirectory <span class="c003">unix</span> of the standard library to the search path.</p></dd><dt class="dt-description"><span class="c013"><span class="c003">-impl</span> <span class="c009">filename</span></span></dt><dd class="dd-description">
Compile the file <span class="c009">filename</span> as an implementation file, even if its
extension is not <span class="c003">.ml</span>.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-inline</span> <span class="c009">n</span></span></dt><dd class="dd-description">
Set aggressiveness of inlining to <span class="c009">n</span>, where <span class="c009">n</span> is a positive
integer. Specifying <span class="c003">-inline 0</span> prevents all functions from being
inlined, except those whose body is smaller than the call site. Thus,
inlining causes no expansion in code size. The default aggressiveness,
<span class="c003">-inline 1</span>, allows slightly larger functions to be inlined, resulting
in a slight expansion in code size. Higher values for the <span class="c003">-inline</span>
option cause larger and larger functions to become candidate for
inlining, but can result in a serious increase in code size.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-intf</span> <span class="c009">filename</span></span></dt><dd class="dd-description">
Compile the file <span class="c009">filename</span> as an interface file, even if its
extension is not <span class="c003">.mli</span>.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-intf-suffix</span> <span class="c009">string</span></span></dt><dd class="dd-description">
Recognize file names ending with <span class="c009">string</span> as interface files
(instead of the default <span class="c003">.mli</span>).
</dd><dt class="dt-description"><span class="c006">-labels</span></dt><dd class="dd-description">
Labels are not ignored in types, labels may be used in applications,
and labelled parameters can be given in any order. This is the default.</dd><dt class="dt-description"><span class="c006">-linkall</span></dt><dd class="dd-description">
Force all modules contained in libraries to be linked in. If this
flag is not given, unreferenced modules are not linked in. When
building a library (option <span class="c003">-a</span>), setting the <span class="c003">-linkall</span> option forces all
subsequent links of programs involving that library to link all the
modules contained in the library. When compiling a module (option
<span class="c003">-c</span>), setting the <span class="c003">-linkall</span> option ensures that this module will
always be linked if it is put in a library and this library is linked.
</dd><dt class="dt-description"><span class="c006">-linscan</span></dt><dd class="dd-description">
Use linear scan register allocation. Compiling with this allocator is faster
than with the usual graph coloring allocator, sometimes quite drastically so for
long functions and modules. On the other hand, the generated code can be a bit
slower.
</dd><dt class="dt-description"><span class="c006">-no-alias-deps</span></dt><dd class="dd-description">
Do not record dependencies for module aliases. See
section <a href="extn.html#s%3Amodule-alias">8.12</a> for more information.
</dd><dt class="dt-description"><span class="c006">-no-app-funct</span></dt><dd class="dd-description">
Deactivates the applicative behaviour of functors. With this option,
each functor application generates new types in its result and
applying the same functor twice to the same argument yields two
incompatible structures.</dd><dt class="dt-description"><span class="c006">-noassert</span></dt><dd class="dd-description">
Do not compile assertion checks. Note that the special form
<span class="c003">assert false</span> is always compiled because it is typed specially.
This flag has no effect when linking already-compiled files.</dd><dt class="dt-description"><span class="c006">-noautolink</span></dt><dd class="dd-description">
When linking <span class="c003">.cmxa</span>libraries, ignore <span class="c003">-cclib</span> and <span class="c003">-ccopt</span>
options potentially contained in the libraries (if these options were
given when building the libraries). This can be useful if a library
contains incorrect specifications of C libraries or C options; in this
case, during linking, set <span class="c003">-noautolink</span> and pass the correct C
libraries and options on the command line.
</dd><dt class="dt-description"><span class="c006">-nodynlink</span></dt><dd class="dd-description">
Allow the compiler to use some optimizations that are valid only for code
that is never dynlinked.
</dd><dt class="dt-description"><span class="c006">-nolabels</span></dt><dd class="dd-description">
Ignore non-optional labels in types. Labels cannot be used in
applications, and parameter order becomes strict.</dd><dt class="dt-description"><span class="c006">-nostdlib</span></dt><dd class="dd-description">
Do not automatically add the standard library directory the list of
directories searched for compiled interface files (<span class="c003">.cmi</span>), compiled
object code files (<span class="c003">.cmx</span>), and libraries (<span class="c003">.cmxa</span>). See also option
<span class="c003">-I</span>.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-o</span> <span class="c009">exec-file</span></span></dt><dd class="dd-description">
Specify the name of the output file produced by the
linker. The
default output name is <span class="c003">a.out</span> under Unix and <span class="c003">camlprog.exe</span> under
Windows. If the <span class="c003">-a</span> option is given, specify the name of the library
produced. If the <span class="c003">-pack</span> option is given, specify the name of the
packed object file produced. If the <span class="c003">-output-obj</span> option is given,
specify the name of the output file produced.
If the <span class="c003">-shared</span> option is given, specify the name of plugin
file produced.
</dd><dt class="dt-description"><span class="c006">-opaque</span></dt><dd class="dd-description">
When the native compiler compiles an implementation, by default it
produces a <span class="c003">.cmx</span> file containing information for cross-module
optimization. It also expects <span class="c003">.cmx</span> files to be present for the
dependencies of the currently compiled source, and uses them for
optimization. Since OCaml 4.03, the compiler will emit a warning if it
is unable to locate the <span class="c003">.cmx</span> file of one of those dependencies.<p>The <span class="c003">-opaque</span> option, available since 4.04, disables cross-module
optimization information for the currently compiled unit. When
compiling <span class="c003">.mli</span> interface, using <span class="c003">-opaque</span> marks the compiled <span class="c003">.cmi</span>
interface so that subsequent compilations of modules that depend on it
will not rely on the corresponding <span class="c003">.cmx</span> file, nor warn if it is
absent. When the native compiler compiles a <span class="c003">.ml</span> implementation,
using <span class="c003">-opaque</span> generates a <span class="c003">.cmx</span> that does not contain any
cross-module optimization information.</p><p>Using this option may degrade the quality of generated code, but it
reduces compilation time, both on clean and incremental
builds. Indeed, with the native compiler, when the implementation of
a compilation unit changes, all the units that depend on it may need
to be recompiled – because the cross-module information may have
changed. If the compilation unit whose implementation changed was
compiled with <span class="c003">-opaque</span>, no such recompilation needs to occur. This
option can thus be used, for example, to get faster edit-compile-test
feedback loops.
</p></dd><dt class="dt-description"><span class="c013"><span class="c003">-open</span> <span class="c009">Module</span></span></dt><dd class="dd-description">
Opens the given module before processing the interface or
implementation files. If several <span class="c003">-open</span> options are given,
they are processed in order, just as if
the statements <span class="c003">open!</span> <span class="c009">Module1</span><span class="c003">;;</span> <span class="c003">...</span> <span class="c003">open!</span> <span class="c009">ModuleN</span><span class="c003">;;</span>
were added at the top of each file.
</dd><dt class="dt-description"><span class="c006">-output-obj</span></dt><dd class="dd-description">
Cause the linker to produce a C object file instead of
an executable file.
This is useful to wrap OCaml code as a C library,
callable from any C program. See chapter <a href="intfc.html#c%3Aintf-c">20</a>,
section <a href="intfc.html#s%3Aembedded-code">20.7.5</a>. The name of the output object file
must be set with the <span class="c003">-o</span> option.
This option can also be used to produce a compiled shared/dynamic library (<span class="c003">.so</span> extension, <span class="c003">.dll</span> under Windows).
</dd><dt class="dt-description"><span class="c006">-p</span></dt><dd class="dd-description">
Generate extra code to write profile information when the program is
executed. The profile information can then be examined with the
analysis program <span class="c003">gprof</span>. (See chapter <a href="profil.html#c%3Aprofiler">18</a> for more
information on profiling.) The <span class="c003">-p</span> option must be given both at
compile-time and at link-time. Linking object files not compiled with
<span class="c003">-p</span> is possible, but results in less precise profiling.<blockquote class="quote"><span class="c007">Unix:</span>   See the Unix manual page for <span class="c003">gprof(1)</span> for more
information about the profiles.<p>Full support for <span class="c003">gprof</span> is only available for certain platforms
(currently: Intel x86 32 and 64 bits under Linux, BSD and MacOS X).
On other platforms, the <span class="c003">-p</span> option will result in a less precise
profile (no call graph information, only a time profile).
</p></blockquote><blockquote class="quote"><span class="c007">Windows:</span>  
The <span class="c003">-p</span> option does not work under Windows.
</blockquote></dd><dt class="dt-description"><span class="c006">-pack</span></dt><dd class="dd-description">
Build an object file (<span class="c003">.cmx</span> and <span class="c003">.o</span>/<span class="c003">.obj</span> files) and its associated compiled
interface (<span class="c003">.cmi</span>) that combines the <span class="c003">.cmx</span> object
files given on the command line, making them appear as sub-modules of
the output <span class="c003">.cmx</span> file. The name of the output <span class="c003">.cmx</span> file must be
given with the <span class="c003">-o</span> option. For instance,
<pre> ocamlopt -pack -o P.cmx A.cmx B.cmx C.cmx
</pre>generates compiled files <span class="c003">P.cmx</span>, <span class="c003">P.o</span> and <span class="c003">P.cmi</span> describing a
compilation unit having three sub-modules <span class="c003">A</span>, <span class="c003">B</span> and <span class="c003">C</span>,
corresponding to the contents of the object files <span class="c003">A.cmx</span>, <span class="c003">B.cmx</span> and
<span class="c003">C.cmx</span>. These contents can be referenced as <span class="c003">P.A</span>, <span class="c003">P.B</span> and <span class="c003">P.C</span>
in the remainder of the program.<p>The <span class="c003">.cmx</span> object files being combined must have been compiled with
the appropriate <span class="c003">-for-pack</span> option. In the example above,
<span class="c003">A.cmx</span>, <span class="c003">B.cmx</span> and <span class="c003">C.cmx</span> must have been compiled with
<span class="c003">ocamlopt -for-pack P</span>.</p><p>Multiple levels of packing can be achieved by combining <span class="c003">-pack</span> with
<span class="c003">-for-pack</span>. Consider the following example:
</p><pre> ocamlopt -for-pack P.Q -c A.ml
ocamlopt -pack -o Q.cmx -for-pack P A.cmx
ocamlopt -for-pack P -c B.ml
ocamlopt -pack -o P.cmx Q.cmx B.cmx
</pre><p>The resulting <span class="c003">P.cmx</span> object file has sub-modules <span class="c003">P.Q</span>, <span class="c003">P.Q.A</span>
and <span class="c003">P.B</span>.
</p></dd><dt class="dt-description"><span class="c013"><span class="c003">-plugin</span> <span class="c009">plugin</span></span></dt><dd class="dd-description">
Dynamically load the code of the given <span class="c009">plugin</span>
(a <span class="c003">.cmo</span>, <span class="c003">.cma</span> or <span class="c003">.cmxs</span> file) in the compiler. <span class="c009">plugin</span> must exist in
the same kind of code as the compiler (<span class="c003">ocamlopt.byte</span> must load bytecode
plugins, while <span class="c003">ocamlopt.opt</span> must load native code plugins), and
extension adaptation is done automatically for <span class="c003">.cma</span> files (to <span class="c003">.cmxs</span> files
if the compiler is compiled in native code).
</dd><dt class="dt-description"><span class="c013"><span class="c003">-pp</span> <span class="c009">command</span></span></dt><dd class="dd-description">
Cause the compiler to call the given <span class="c009">command</span> as a preprocessor
for each source file. The output of <span class="c009">command</span> is redirected to
an intermediate file, which is compiled. If there are no compilation
errors, the intermediate file is deleted afterwards.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-ppx</span> <span class="c009">command</span></span></dt><dd class="dd-description">
After parsing, pipe the abstract syntax tree through the preprocessor
<span class="c009">command</span>. The module <span class="c003">Ast_mapper</span>, described in
chapter <a href="parsing.html#c%3Aparsinglib">27</a>: <a href="libref/Ast_mapper.html"> <span class="c003">Ast_mapper</span> </a>
,
implements the external interface of a preprocessor.</dd><dt class="dt-description"><span class="c006">-principal</span></dt><dd class="dd-description">
Check information path during type-checking, to make sure that all
types are derived in a principal way. When using labelled arguments
and/or polymorphic methods, this flag is required to ensure future
versions of the compiler will be able to infer types correctly, even
if internal algorithms change.
All programs accepted in <span class="c003">-principal</span> mode are also accepted in the
default mode with equivalent types, but different binary signatures,
and this may slow down type checking; yet it is a good idea to
use it once before publishing source code.</dd><dt class="dt-description"><span class="c006">-rectypes</span></dt><dd class="dd-description">
Allow arbitrary recursive types during type-checking. By default,
only recursive types where the recursion goes through an object type
are supported.Note that once you have created an interface using this
flag, you must use it again for all dependencies.</dd><dt class="dt-description"><span class="c013"><span class="c003">-runtime-variant</span> <span class="c009">suffix</span></span></dt><dd class="dd-description">
Add the <span class="c009">suffix</span> string to the name of the runtime library used by
the program. Currently, only one such suffix is supported: <span class="c003">d</span>, and
only if the OCaml compiler was configured with option
<span class="c003">-with-debug-runtime</span>. This suffix gives the debug version of the
runtime, which is useful for debugging pointer problems in low-level
code such as C stubs.
</dd><dt class="dt-description"><span class="c006">-S</span></dt><dd class="dd-description">
Keep the assembly code produced during the compilation. The assembly
code for the source file <span class="c009">x</span><span class="c003">.ml</span> is saved in the file <span class="c009">x</span><span class="c003">.s</span>.
</dd><dt class="dt-description"><span class="c006">-shared</span></dt><dd class="dd-description">
Build a plugin (usually <span class="c003">.cmxs</span>) that can be dynamically loaded with
the <span class="c003">Dynlink</span> module. The name of the plugin must be
set with the <span class="c003">-o</span> option. A plugin can include a number of OCaml
modules and libraries, and extra native objects (<span class="c003">.o</span>, <span class="c003">.obj</span>, <span class="c003">.a</span>,
<span class="c003">.lib</span> files). Building native plugins is only supported for some
operating system. Under some systems (currently,
only Linux AMD 64), all the OCaml code linked in a plugin must have
been compiled without the <span class="c003">-nodynlink</span> flag. Some constraints might also
apply to the way the extra native objects have been compiled (under
Linux AMD 64, they must contain only position-independent code).
</dd><dt class="dt-description"><span class="c006">-safe-string</span></dt><dd class="dd-description">
Enforce the separation between types <span class="c003">string</span> and <span class="c003">bytes</span>,
thereby making strings read-only. This is the default.</dd><dt class="dt-description"><span class="c006">-short-paths</span></dt><dd class="dd-description">
When a type is visible under several module-paths, use the shortest
one when printing the type’s name in inferred interfaces and error and
warning messages. Identifier names starting with an underscore <span class="c003">_</span> or
containing double underscores <span class="c003">__</span> incur a penalty of +10 when computing
their length.</dd><dt class="dt-description"><span class="c006">-strict-sequence</span></dt><dd class="dd-description">
Force the left-hand part of each sequence to have type unit.</dd><dt class="dt-description"><span class="c006">-strict-formats</span></dt><dd class="dd-description">
Reject invalid formats that were accepted in legacy format
implementations. You should use this flag to detect and fix such
invalid formats, as they will be rejected by future OCaml versions.</dd><dt class="dt-description"><span class="c006">-unboxed-types</span></dt><dd class="dd-description">
When a type is unboxable (i.e. a record with a single argument or a
concrete datatype with a single constructor of one argument) it will
be unboxed unless annotated with <span class="c003">[@@ocaml.boxed]</span>.
</dd><dt class="dt-description"><span class="c006">-no-unboxed-types</span></dt><dd class="dd-description">
When a type is unboxable it will be boxed unless annotated with
<span class="c003">[@@ocaml.unboxed]</span>. This is the default.
</dd><dt class="dt-description"><span class="c006">-unsafe</span></dt><dd class="dd-description">
Turn bound checking off for array and string accesses (the <span class="c003">v.(i)</span> and
<span class="c003">s.[i]</span> constructs). Programs compiled with <span class="c003">-unsafe</span> are therefore
faster, but unsafe: anything can happen if the program
accesses an array or string outside of its bounds.
Additionally, turn off the check for zero divisor in integer division
and modulus operations. With <span class="c003">-unsafe</span>, an integer division
(or modulus) by zero can halt the program or continue with an
unspecified result instead of raising a <span class="c003">Division_by_zero</span> exception.
</dd><dt class="dt-description"><span class="c006">-unsafe-string</span></dt><dd class="dd-description">
Identify the types <span class="c003">string</span> and <span class="c003">bytes</span>, thereby making strings writable.
This is intended for compatibility with old source code and should not
be used with new software.</dd><dt class="dt-description"><span class="c006">-v</span></dt><dd class="dd-description">
Print the version number of the compiler and the location of the
standard library directory, then exit.</dd><dt class="dt-description"><span class="c006">-verbose</span></dt><dd class="dd-description">
Print all external commands before they are executed,
in particular invocations of the assembler, C compiler, and linker.
Useful to debug C library problems.</dd><dt class="dt-description"><span class="c013"><span class="c003">-version</span> or <span class="c003">-vnum</span></span></dt><dd class="dd-description">
Print the version number of the compiler in short form (e.g. <span class="c003">3.11.0</span>),
then exit.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-w</span> <span class="c009">warning-list</span></span></dt><dd class="dd-description">
Enable, disable, or mark as fatal the warnings specified by the argument
<span class="c009">warning-list</span>.
Each warning can be <em>enabled</em> or <em>disabled</em>, and each warning
can be <em>fatal</em> or <em>non-fatal</em>.
If a warning is disabled, it isn’t displayed and doesn’t affect
compilation in any way (even if it is fatal). If a warning is
enabled, it is displayed normally by the compiler whenever the source
code triggers it. If it is enabled and fatal, the compiler will also
stop with an error after displaying it.<p>The <span class="c009">warning-list</span> argument is a sequence of warning specifiers,
with no separators between them. A warning specifier is one of the
following:</p><dl class="description"><dt class="dt-description">
<span class="c013"><span class="c003">+</span><span class="c009">num</span></span></dt><dd class="dd-description"> Enable warning number <span class="c009">num</span>.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-</span><span class="c009">num</span></span></dt><dd class="dd-description"> Disable warning number <span class="c009">num</span>.
</dd><dt class="dt-description"><span class="c013"><span class="c003">@</span><span class="c009">num</span></span></dt><dd class="dd-description"> Enable and mark as fatal warning number <span class="c009">num</span>.
</dd><dt class="dt-description"><span class="c013"><span class="c003">+</span><span class="c009">num1</span>..<span class="c009">num2</span></span></dt><dd class="dd-description"> Enable warnings in the given range.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-</span><span class="c009">num1</span>..<span class="c009">num2</span></span></dt><dd class="dd-description"> Disable warnings in the given range.
</dd><dt class="dt-description"><span class="c013"><span class="c003">@</span><span class="c009">num1</span>..<span class="c009">num2</span></span></dt><dd class="dd-description"> Enable and mark as fatal warnings in
the given range.
</dd><dt class="dt-description"><span class="c013"><span class="c003">+</span><span class="c009">letter</span></span></dt><dd class="dd-description"> Enable the set of warnings corresponding to
<span class="c009">letter</span>. The letter may be uppercase or lowercase.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-</span><span class="c009">letter</span></span></dt><dd class="dd-description"> Disable the set of warnings corresponding to
<span class="c009">letter</span>. The letter may be uppercase or lowercase.
</dd><dt class="dt-description"><span class="c013"><span class="c003">@</span><span class="c009">letter</span></span></dt><dd class="dd-description"> Enable and mark as fatal the set of warnings
corresponding to <span class="c009">letter</span>. The letter may be uppercase or
lowercase.
</dd><dt class="dt-description"><span class="c011">uppercase-letter</span></dt><dd class="dd-description"> Enable the set of warnings corresponding
to <span class="c009">uppercase-letter</span>.
</dd><dt class="dt-description"><span class="c011">lowercase-letter</span></dt><dd class="dd-description"> Disable the set of warnings corresponding
to <span class="c009">lowercase-letter</span>.
</dd></dl><p>Warning numbers and letters which are out of the range of warnings
that are currently defined are ignored. The warnings are as follows.
</p><dl class="description"><dt class="dt-description">
<span class="c013">1</span></dt><dd class="dd-description"> Suspicious-looking start-of-comment mark.
</dd><dt class="dt-description"><span class="c013">2</span></dt><dd class="dd-description"> Suspicious-looking end-of-comment mark.
</dd><dt class="dt-description"><span class="c013">3</span></dt><dd class="dd-description"> Deprecated feature.
</dd><dt class="dt-description"><span class="c013">4</span></dt><dd class="dd-description"> Fragile pattern matching: matching that will remain complete even
if additional constructors are added to one of the variant types
matched.
</dd><dt class="dt-description"><span class="c013">5</span></dt><dd class="dd-description"> Partially applied function: expression whose result has function
type and is ignored.
</dd><dt class="dt-description"><span class="c013">6</span></dt><dd class="dd-description"> Label omitted in function application.
</dd><dt class="dt-description"><span class="c013">7</span></dt><dd class="dd-description"> Method overridden.
</dd><dt class="dt-description"><span class="c013">8</span></dt><dd class="dd-description"> Partial match: missing cases in pattern-matching.
</dd><dt class="dt-description"><span class="c013">9</span></dt><dd class="dd-description"> Missing fields in a record pattern.
</dd><dt class="dt-description"><span class="c013">10</span></dt><dd class="dd-description"> Expression on the left-hand side of a sequence that doesn’t have type
<span class="c003">unit</span> (and that is not a function, see warning number 5).
</dd><dt class="dt-description"><span class="c013">11</span></dt><dd class="dd-description"> Redundant case in a pattern matching (unused match case).
</dd><dt class="dt-description"><span class="c013">12</span></dt><dd class="dd-description"> Redundant sub-pattern in a pattern-matching.
</dd><dt class="dt-description"><span class="c013">13</span></dt><dd class="dd-description"> Instance variable overridden.
</dd><dt class="dt-description"><span class="c013">14</span></dt><dd class="dd-description"> Illegal backslash escape in a string constant.
</dd><dt class="dt-description"><span class="c013">15</span></dt><dd class="dd-description"> Private method made public implicitly.
</dd><dt class="dt-description"><span class="c013">16</span></dt><dd class="dd-description"> Unerasable optional argument.
</dd><dt class="dt-description"><span class="c013">17</span></dt><dd class="dd-description"> Undeclared virtual method.
</dd><dt class="dt-description"><span class="c013">18</span></dt><dd class="dd-description"> Non-principal type.
</dd><dt class="dt-description"><span class="c013">19</span></dt><dd class="dd-description"> Type without principality.
</dd><dt class="dt-description"><span class="c013">20</span></dt><dd class="dd-description"> Unused function argument.
</dd><dt class="dt-description"><span class="c013">21</span></dt><dd class="dd-description"> Non-returning statement.
</dd><dt class="dt-description"><span class="c013">22</span></dt><dd class="dd-description"> Preprocessor warning.
</dd><dt class="dt-description"><span class="c013">23</span></dt><dd class="dd-description"> Useless record <span class="c003">with</span> clause.
</dd><dt class="dt-description"><span class="c013">24</span></dt><dd class="dd-description"> Bad module name: the source file name is not a valid OCaml module name.
</dd><dt class="dt-description"><span class="c013">25</span></dt><dd class="dd-description"> Deprecated: now part of warning 8.
</dd><dt class="dt-description"><span class="c013">26</span></dt><dd class="dd-description"> Suspicious unused variable: unused variable that is bound
with <span class="c003">let</span> or <span class="c003">as</span>, and doesn’t start with an underscore (<span class="c003">_</span>)
character.
</dd><dt class="dt-description"><span class="c013">27</span></dt><dd class="dd-description"> Innocuous unused variable: unused variable that is not bound with
<span class="c003">let</span> nor <span class="c003">as</span>, and doesn’t start with an underscore (<span class="c003">_</span>)
character.
</dd><dt class="dt-description"><span class="c013">28</span></dt><dd class="dd-description"> Wildcard pattern given as argument to a constant constructor.
</dd><dt class="dt-description"><span class="c013">29</span></dt><dd class="dd-description"> Unescaped end-of-line in a string constant (non-portable code).
</dd><dt class="dt-description"><span class="c013">30</span></dt><dd class="dd-description"> Two labels or constructors of the same name are defined in two
mutually recursive types.
</dd><dt class="dt-description"><span class="c013">31</span></dt><dd class="dd-description"> A module is linked twice in the same executable.
</dd><dt class="dt-description"><span class="c013">32</span></dt><dd class="dd-description"> Unused value declaration.
</dd><dt class="dt-description"><span class="c013">33</span></dt><dd class="dd-description"> Unused open statement.
</dd><dt class="dt-description"><span class="c013">34</span></dt><dd class="dd-description"> Unused type declaration.
</dd><dt class="dt-description"><span class="c013">35</span></dt><dd class="dd-description"> Unused for-loop index.
</dd><dt class="dt-description"><span class="c013">36</span></dt><dd class="dd-description"> Unused ancestor variable.
</dd><dt class="dt-description"><span class="c013">37</span></dt><dd class="dd-description"> Unused constructor.
</dd><dt class="dt-description"><span class="c013">38</span></dt><dd class="dd-description"> Unused extension constructor.
</dd><dt class="dt-description"><span class="c013">39</span></dt><dd class="dd-description"> Unused rec flag.
</dd><dt class="dt-description"><span class="c013">40</span></dt><dd class="dd-description"> Constructor or label name used out of scope.
</dd><dt class="dt-description"><span class="c013">41</span></dt><dd class="dd-description"> Ambiguous constructor or label name.
</dd><dt class="dt-description"><span class="c013">42</span></dt><dd class="dd-description"> Disambiguated constructor or label name (compatibility warning).
</dd><dt class="dt-description"><span class="c013">43</span></dt><dd class="dd-description"> Nonoptional label applied as optional.
</dd><dt class="dt-description"><span class="c013">44</span></dt><dd class="dd-description"> Open statement shadows an already defined identifier.
</dd><dt class="dt-description"><span class="c013">45</span></dt><dd class="dd-description"> Open statement shadows an already defined label or constructor.
</dd><dt class="dt-description"><span class="c013">46</span></dt><dd class="dd-description"> Error in environment variable.
</dd><dt class="dt-description"><span class="c013">47</span></dt><dd class="dd-description"> Illegal attribute payload.
</dd><dt class="dt-description"><span class="c013">48</span></dt><dd class="dd-description"> Implicit elimination of optional arguments.
</dd><dt class="dt-description"><span class="c013">49</span></dt><dd class="dd-description"> Absent cmi file when looking up module alias.
</dd><dt class="dt-description"><span class="c013">50</span></dt><dd class="dd-description"> Unexpected documentation comment.
</dd><dt class="dt-description"><span class="c013">51</span></dt><dd class="dd-description"> Warning on non-tail calls if @tailcall present.
</dd><dt class="dt-description"><span class="c013">52 (see </span><a href="comp.html#ss%3Awarn52"><span class="c013">9.5.2</span></a><span class="c013">)</span></dt><dd class="dd-description"> Fragile constant pattern.
</dd><dt class="dt-description"><span class="c013">53</span></dt><dd class="dd-description"> Attribute cannot appear in this context
</dd><dt class="dt-description"><span class="c013">54</span></dt><dd class="dd-description"> Attribute used more than once on an expression
</dd><dt class="dt-description"><span class="c013">55</span></dt><dd class="dd-description"> Inlining impossible
</dd><dt class="dt-description"><span class="c013">56</span></dt><dd class="dd-description"> Unreachable case in a pattern-matching (based on type information).
</dd><dt class="dt-description"><span class="c013">57 (see </span><a href="comp.html#ss%3Awarn57"><span class="c013">9.5.3</span></a><span class="c013">)</span></dt><dd class="dd-description"> Ambiguous or-pattern variables under guard
</dd><dt class="dt-description"><span class="c013">58</span></dt><dd class="dd-description"> Missing cmx file
</dd><dt class="dt-description"><span class="c013">59</span></dt><dd class="dd-description"> Assignment to non-mutable value
</dd><dt class="dt-description"><span class="c013">60</span></dt><dd class="dd-description"> Unused module declaration
</dd><dt class="dt-description"><span class="c013">61</span></dt><dd class="dd-description"> Unboxable type in primitive declaration
</dd><dt class="dt-description"><span class="c013">62</span></dt><dd class="dd-description"> Type constraint on GADT type declaration
</dd><dt class="dt-description"><span class="c013">A</span></dt><dd class="dd-description"> all warnings
</dd><dt class="dt-description"><span class="c013">C</span></dt><dd class="dd-description"> warnings 1, 2.
</dd><dt class="dt-description"><span class="c013">D</span></dt><dd class="dd-description"> Alias for warning 3.
</dd><dt class="dt-description"><span class="c013">E</span></dt><dd class="dd-description"> Alias for warning 4.
</dd><dt class="dt-description"><span class="c013">F</span></dt><dd class="dd-description"> Alias for warning 5.
</dd><dt class="dt-description"><span class="c013">K</span></dt><dd class="dd-description"> warnings 32, 33, 34, 35, 36, 37, 38, 39.
</dd><dt class="dt-description"><span class="c013">L</span></dt><dd class="dd-description"> Alias for warning 6.
</dd><dt class="dt-description"><span class="c013">M</span></dt><dd class="dd-description"> Alias for warning 7.
</dd><dt class="dt-description"><span class="c013">P</span></dt><dd class="dd-description"> Alias for warning 8.
</dd><dt class="dt-description"><span class="c013">R</span></dt><dd class="dd-description"> Alias for warning 9.
</dd><dt class="dt-description"><span class="c013">S</span></dt><dd class="dd-description"> Alias for warning 10.
</dd><dt class="dt-description"><span class="c013">U</span></dt><dd class="dd-description"> warnings 11, 12.
</dd><dt class="dt-description"><span class="c013">V</span></dt><dd class="dd-description"> Alias for warning 13.
</dd><dt class="dt-description"><span class="c013">X</span></dt><dd class="dd-description"> warnings 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 30.
</dd><dt class="dt-description"><span class="c013">Y</span></dt><dd class="dd-description"> Alias for warning 26.
</dd><dt class="dt-description"><span class="c013">Z</span></dt><dd class="dd-description"> Alias for warning 27.
</dd></dl><p>The default setting is <span class="c003">-w +a-4-6-7-9-27-29-32..42-44-45-48-50-60</span>.
It is displayed by <span class="c003">ocamlopt -help</span>.
Note that warnings 5 and 10 are not always triggered, depending on
the internals of the type checker.</p></dd><dt class="dt-description"><span class="c013"><span class="c003">-warn-error</span> <span class="c009">warning-list</span></span></dt><dd class="dd-description">
Mark as fatal the warnings specified in the argument <span class="c009">warning-list</span>.
The compiler will stop with an error when one of these warnings is
emitted. The <span class="c009">warning-list</span> has the same meaning as for
the <span class="c003">-w</span> option: a <span class="c003">+</span> sign (or an uppercase letter) marks the
corresponding warnings as fatal, a <span class="c003">-</span>
sign (or a lowercase letter) turns them back into non-fatal warnings,
and a <span class="c003">@</span> sign both enables and marks as fatal the corresponding
warnings.<p>Note: it is not recommended to use warning sets (i.e. letters) as
arguments to <span class="c003">-warn-error</span>
in production code, because this can break your build when future versions
of OCaml add some new warnings.</p><p>The default setting is <span class="c003">-warn-error -a+31</span> (only warning 31 is fatal).</p></dd><dt class="dt-description"><span class="c006">-warn-help</span></dt><dd class="dd-description">
Show the description of all available warning numbers.</dd><dt class="dt-description"><span class="c006">-where</span></dt><dd class="dd-description">
Print the location of the standard library, then exit.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-</span> <span class="c009">file</span></span></dt><dd class="dd-description">
Process <span class="c009">file</span> as a file name, even if it starts with a dash (<span class="c003">-</span>)
character.
</dd><dt class="dt-description"><span class="c013"><span class="c003">-help</span> or <span class="c003">--help</span></span></dt><dd class="dd-description">
Display a short usage summary and exit.</dd></dl>
<h5 class="paragraph" id="sec307">Options for the IA32 architecture</h5>
<p>
The IA32 code generator (Intel Pentium, AMD Athlon) supports the
following additional option:</p><dl class="description"><dt class="dt-description">
<span class="c006">-ffast-math</span></dt><dd class="dd-description"> Use the IA32 instructions to compute
trigonometric and exponential functions, instead of calling the
corresponding library routines. The functions affected are:
<span class="c003">atan</span>, <span class="c003">atan2</span>, <span class="c003">cos</span>, <span class="c003">log</span>, <span class="c003">log10</span>, <span class="c003">sin</span>, <span class="c003">sqrt</span> and <span class="c003">tan</span>.
The resulting code runs faster, but the range of supported arguments
and the precision of the result can be reduced. In particular,
trigonometric operations <span class="c003">cos</span>, <span class="c003">sin</span>, <span class="c003">tan</span> have their range reduced to
[−2<sup>64</sup>, 2<sup>64</sup>].
</dd></dl>
<h5 class="paragraph" id="sec308">Options for the AMD64 architecture</h5>
<p>
The AMD64 code generator (64-bit versions of Intel Pentium and AMD
Athlon) supports the following additional options:</p><dl class="description"><dt class="dt-description">
<span class="c006">-fPIC</span></dt><dd class="dd-description"> Generate position-independent machine code. This is
the default.
</dd><dt class="dt-description"><span class="c006">-fno-PIC</span></dt><dd class="dd-description"> Generate position-dependent machine code.
</dd></dl>
<h5 class="paragraph" id="sec309">Contextual control of command-line options</h5>
<p>The compiler command line can be modified “from the outside”
with the following mechanisms. These are experimental
and subject to change. They should be used only for experimental and
development work, not in released packages.</p><dl class="description"><dt class="dt-description">
<span class="c006">OCAMLPARAM</span> (environment variable)</dt><dd class="dd-description">
A set of arguments that will be inserted before or after the arguments from
the command line. Arguments are specified in a comma-separated list
of <span class="c003">name=value</span> pairs. A <span class="c003">_</span> is used to specify the position of
the command line arguments, i.e. <span class="c003">a=x,_,b=y</span> means that <span class="c003">a=x</span> should be
executed before parsing the arguments, and <span class="c003">b=y</span> after. Finally,
an alternative separator can be specified as the
first character of the string, within the set <span class="c003">:|; ,</span>.
</dd><dt class="dt-description"><span class="c006">ocaml_compiler_internal_params</span> (file in the stdlib directory)</dt><dd class="dd-description">
A mapping of file names to lists of arguments that
will be added to the command line (and <span class="c003">OCAMLPARAM</span>) arguments.
</dd><dt class="dt-description"><span class="c006">OCAML_FLEXLINK</span> (environment variable)</dt><dd class="dd-description">
Alternative executable to use on native
Windows for <span class="c003">flexlink</span> instead of the
configured value. Primarily used for bootstrapping.
</dd></dl>
<h2 class="section" id="sec310">12.3  Common errors</h2>
<p>The error messages are almost identical to those of <span class="c003">ocamlc</span>.
See section <a href="comp.html#s%3Acomp-errors">9.4</a>.</p>
<h2 class="section" id="sec311">12.4  Running executables produced by ocamlopt</h2>
<p>Executables generated by <span class="c003">ocamlopt</span> are native, stand-alone executable
files that can be invoked directly. They do
not depend on the <span class="c003">ocamlrun</span> bytecode runtime system nor on
dynamically-loaded C/OCaml stub libraries.</p><p>During execution of an <span class="c003">ocamlopt</span>-generated executable,
the following environment variables are also consulted:
</p><dl class="description"><dt class="dt-description">
<span class="c006">OCAMLRUNPARAM</span></dt><dd class="dd-description"> Same usage as in <span class="c003">ocamlrun</span>
(see section <a href="runtime.html#ocamlrun-options">11.2</a>), except that option <span class="c003">l</span>
is ignored (the operating system’s stack size limit
is used instead).
</dd><dt class="dt-description"><span class="c006">CAMLRUNPARAM</span></dt><dd class="dd-description"> If <span class="c003">OCAMLRUNPARAM</span> is not found in the
environment, then <span class="c003">CAMLRUNPARAM</span> will be used instead. If
<span class="c003">CAMLRUNPARAM</span> is not found, then the default values will be used.
</dd></dl>
<h2 class="section" id="sec312">12.5  Compatibility with the bytecode compiler</h2>
<p>
<a id="s:compat-native-bytecode"></a></p><p>This section lists the known incompatibilities between the bytecode
compiler and the native-code compiler. Except on those points, the two
compilers should generate code that behave identically.</p><ul class="itemize"><li class="li-itemize">Signals are detected only when the program performs an
allocation in the heap. That is, if a signal is delivered while in a
piece of code that does not allocate, its handler will not be called
until the next heap allocation.</li><li class="li-itemize">Stack overflow, typically caused by excessively deep recursion,
is not always turned into a <span class="c003">Stack_overflow</span> exception like the
bytecode compiler does. The runtime system makes a best effort to
trap stack overflows and raise the <span class="c003">Stack_overflow</span> exception, but
sometimes it fails and a “segmentation fault” or another system fault
occurs instead.</li><li class="li-itemize">On ARM and PowerPC processors (32 and 64 bits), fused
multiply-add (FMA) instructions can be generated for a
floating-point multiplication followed by a floating-point addition
or subtraction, as in <span class="c003">x *. y +. z</span>. The FMA instruction avoids
rounding the intermediate result <span class="c003">x *. y</span>, which is generally
beneficial, but produces floating-point results that differ slightly
from those produced by the bytecode interpreter.</li><li class="li-itemize">On IA32 processors only (Intel and AMD x86 processors in 32-bit
mode), some intermediate results in floating-point computations are
kept in extended precision rather than being rounded to double
precision like the bytecode compiler always does. Floating-point
results can therefore differ slightly between bytecode and native code.</li><li class="li-itemize">The native-code compiler performs a number of optimizations that
the bytecode compiler does not perform, especially when the Flambda
optimizer is active. In particular, the native-code compiler
identifies and eliminates “dead code”, i.e. computations that do
not contribute to the results of the program. For example,
<pre> let _ = ignore M.f
</pre>contains a reference to compilation unit <span class="c003">M</span> when compiled to
bytecode. This reference forces <span class="c003">M</span> to be linked and its
initialization code to be executed. The native-code compiler
eliminates the reference to <span class="c003">M</span>, hence the compilation unit <span class="c003">M</span> may
not be linked and executed. A workaround is to compile <span class="c003">M</span> with the
<span class="c003">-linkall</span> flag so that it will always be linked and executed, even if
not referenced. See also the <span class="c003">Sys.opaque_identity</span> function from the
<span class="c003">Sys</span> standard library module.</li></ul>
<hr>
<a href="runtime.html"><img src="previous_motif.svg" alt="Previous"></a>
<a href="index.html"><img src="contents_motif.svg" alt="Up"></a>
<a href="lexyacc.html"><img src="next_motif.svg" alt="Next"></a>
</body>
</html>
distrib
>
Mageia
>
7
>
armv7hl
>
media
>
core-release
>
by-pkgid
>
fb18813323b88f9a6e869238ab603257
>
files
>
510
