Chapter 12 Native-code compilation (ocamlopt)

This chapter describes the OCaml high-performancenative-code compiler ocamlopt, which compiles OCaml source files tonative code object files and links these object files to producestandalone executables.

The native-code compiler is only available on certain platforms.It produces code that runs faster than the bytecode produced byocamlc, at the cost of increased compilation time and executable codesize. Compatibility with the bytecode compiler is extremely high: thesame source code should run identically when compiled with ocamlc andocamlopt.

It is not possible to mix native-code object files produced by ocamloptwith bytecode object files produced by ocamlc: a program must becompiled entirely with ocamlopt or entirely with ocamlc. Native-codeobject files produced by ocamlopt cannot be loaded in the toplevelsystem ocaml.

12.1 Overview of the compiler

The ocamlopt command has a command-line interface very close to thatof ocamlc. It accepts the same types of arguments, and processes themsequentially, after all options have been processed:

  • Arguments ending in .mli are taken to be source files forcompilation unit interfaces. Interfaces specify the names exported bycompilation units: they declare value names with their types, definepublic data types, declare abstract data types, and so on. From thefile x.mli, the ocamlopt compiler produces a compiled interfacein the file x.cmi. The interface produced is identical to thatproduced by the bytecode compiler ocamlc.

  • Arguments ending in .ml are taken to be source files for compilationunit implementations. Implementations provide definitions for thenames exported by the unit, and also contain expressions to beevaluated for their side-effects. From the file x.ml, the ocamloptcompiler produces two files: x.o, containing native object code,and x.cmx, containing extra information for linking andoptimization of the clients of the unit. The compiled implementationshould always be referred to under the name x.cmx (when givena .o or .obj file, ocamlopt assumes that it contains code compiled from C,not from OCaml).The implementation is checked against the interface file x.mli(if it exists) as described in the manual for ocamlc(chapter 9).

  • Arguments ending in .cmx are taken to be compiled object code. Thesefiles are linked together, along with the object files obtainedby compiling .ml arguments (if any), and the OCaml standardlibrary, to produce a native-code executable program. The order inwhich .cmx and .ml arguments are presented on the command line isrelevant: compilation units are initialized in that order atrun-time, and it is a link-time error to use a component of a unitbefore having initialized it. Hence, a given x.cmx file must comebefore all .cmx files that refer to the unit x.

  • Arguments ending in .cmxa are taken to be libraries of object code.Such a library packs in two files (lib.cmxa and lib.a/.lib)a set of object files (.cmx and .o/.obj files). Libraries are build withocamlopt -a (see the description of the -a option below). The objectfiles contained in the library are linked as regular .cmx files (seeabove), in the order specified when the library was built. The onlydifference is that if an object file contained in a library is notreferenced anywhere in the program, then it is not linked in.
  • Arguments ending in .c are passed to the C compiler, which generatesa .o/.obj object file. This object file is linked with the program.
  • Arguments ending in .o, .a or .so (.obj, .lib and .dllunder Windows) are assumed to be C object files andlibraries. They are linked with the program.The output of the linking phase is a regular Unix or Windowsexecutable file. It does not need ocamlrun to run.

12.2 Options

The following command-line options are recognized by ocamlopt.The options -pack, -a, -shared, -c and -output-obj are mutuallyexclusive.

  • -a

  • Build a library(.cmxa and .a/.lib files)with the object files (.cmx and .o/.obj files)given on the command line, instead of linking them into an executable file.The name of the library must be set with the -o option.If -cclib or -ccopt options are passed on the commandline, these options are stored in the resulting .cmxalibrary. Then,linking with this library automatically adds back the -cclib and -ccopt options as if they had been provided on thecommand line, unless the -noautolink option is given.

  • -absname

  • Force error messages to show absolute paths for file names.
  • -annot
  • Dump detailed information about the compilation (types, bindings,tail-calls, etc). The information for file src.mlis put into file src.annot. In case of a type error, dumpall the information inferred by the type-checker before the error.The src.annot file can be used with the emacs commands given inemacs/caml-types.el to display types and other annotationsinteractively.
  • -argsfilename
  • Read additional newline-terminated command line arguments from filename.
  • -args0filename
  • Read additional null character terminated command line arguments fromfilename.
  • -bin-annot
  • Dump detailed information about the compilation (types, bindings,tail-calls, etc) in binary format. The information for file src.ml(resp. src.mli) is put into file src.cmt(resp. src.cmti). In case of a type error, dumpall the information inferred by the type-checker before the error.The .cmt and .cmti files produced by -bin-annot containmore information and are much more compact than the files produced by-annot.
  • -c
  • Compile only. Suppress the linking phase of thecompilation. Source code files are turned into compiled files, but noexecutable file is produced. This option is useful tocompile modules separately.
  • -ccccomp
  • Use ccomp as the C linker called to build the final executable and as the C compiler for compiling .c source files.
  • -cclib-llibname
  • Pass the -llibname option to the linker.This causes the given C library to be linked with the program.
  • -ccoptoption
  • Pass the given option to the C compiler and linker.For instance,-ccopt -Ldir causes the C linker to search for C libraries indirectory dir.
  • -colormode

  • Enable or disable colors in compiler messages (especially warnings and errors).The following modes are supported:

    • auto
    • use heuristics to enable colors only if the output supports them(an ANSI-compatible tty terminal);
    • always
    • enable colors unconditionally;
    • never
    • disable color output.The default setting is ’auto’, and the current heuristicchecks that the TERM environment variable exists and isnot empty or dumb, and that ’isatty(stderr)’ holds.The environment variable OCAML_COLOR is considered if -color is notprovided. Its values are auto/always/never as above.

  • -error-stylemode

  • Control the way error messages and warnings are printed.The following modes are supported:

    • short
    • only print the error and its location;
    • contextual
    • like short, but also display the source code snippetcorresponding to the location of the error.The default setting is contextual.The environment variable OCAML_ERROR_STYLE is considered if -error-style isnot provided. Its values are short/contextual as above.

  • -compact

  • Optimize the produced code for space rather than for time. Thisresults in slightly smaller but slightly slower programs. The default is tooptimize for speed.
  • -config
  • Print the version number of ocamlopt and a detailedsummary of its configuration, then exit.
  • -config-varvar
  • Print the value of a specific configuration variable from the-config output, then exit. If the variable does not exist, the exitcode is non-zero. This option is only available since OCaml 4.08,so script authors should have a fallback for older versions.
  • -dependocamldep-args
  • Compute dependencies, as the ocamldep command would do. The remainingarguments are interpreted as if they were given to the ocamldep command.
  • -for-packmodule-path
  • Generate an object file (.cmx and .o/.obj files)that can later be includedas a sub-module (with the given access path) of a compilation unitconstructed with -pack. For instance,ocamlopt -for-pack P -c A.mlwill generate a..cmx and a.o files that canlater be used with ocamlopt -pack -o P.cmx a.cmx.Note: you can still pack a module that was compiled without-for-pack but in this case exceptions will be printed with the wrongnames.
  • -g
  • Add debugging information while compiling and linking. This option isrequired in order to produce stack backtraces whenthe program terminates on an uncaught exception (seesection 11.2).
  • -i
  • Cause the compiler to print all defined names (with their inferredtypes or their definitions) when compiling an implementation (.mlfile). No compiled files (.cmo and .cmi files) are produced.This can be useful to check the types inferred by thecompiler. Also, since the output follows the syntax of interfaces, itcan help in writing an explicit interface (.mli file) for a file:just redirect the standard output of the compiler to a .mli file,and edit that file to remove all declarations of unexported names.
  • -Idirectory

  • Add the given directory to the list of directories searched forcompiled interface files (.cmi), compiled object code files (.cmx),and libraries (.cmxa).By default, the current directory is searched first, then the standardlibrary directory. Directories added with -I are searched after thecurrent directory, in the order in which they were given on the command line,but before the standard library directory. See also option -nostdlib.If the given directory starts with +, it is taken relative to thestandard library directory. For instance, -I +unix adds thesubdirectory unix of the standard library to the search path.

  • -implfilename

  • Compile the file filename as an implementation file, even if itsextension is not .ml.
  • -inlinen
  • Set aggressiveness of inlining to n, where n is a positiveinteger. Specifying -inline 0 prevents all functions from beinginlined, except those whose body is smaller than the call site. Thus,inlining causes no expansion in code size. The default aggressiveness,-inline 1, allows slightly larger functions to be inlined, resultingin a slight expansion in code size. Higher values for the -inlineoption cause larger and larger functions to become candidate forinlining, but can result in a serious increase in code size.
  • -intffilename
  • Compile the file filename as an interface file, even if itsextension is not .mli.
  • -intf-suffixstring
  • Recognize file names ending with string as interface files(instead of the default .mli).
  • -labels
  • 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.
  • -linkall
  • Force all modules contained in libraries to be linked in. If thisflag is not given, unreferenced modules are not linked in. Whenbuilding a library (option -a), setting the -linkall option forces allsubsequent links of programs involving that library to link all themodules contained in the library. When compiling a module (option-c), setting the -linkall option ensures that this module willalways be linked if it is put in a library and this library is linked.
  • -linscan
  • Use linear scan register allocation. Compiling with this allocator is fasterthan with the usual graph coloring allocator, sometimes quite drastically so forlong functions and modules. On the other hand, the generated code can be a bitslower.
  • -match-context-rows
  • Set the number of rows of context used for optimization duringpattern matching compilation. The default value is 32. Lower valuescause faster compilation, but less optimized code. This advancedoption is meant for use in the event that a pattern-match-heavyprogram leads to significant increases in compilation time.
  • -no-alias-deps
  • Do not record dependencies for module aliases. Seesection 8.9 for more information.
  • -no-app-funct
  • Deactivates the applicative behaviour of functors. With this option,each functor application generates new types in its result andapplying the same functor twice to the same argument yields twoincompatible structures.
  • -noassert
  • Do not compile assertion checks. Note that the special formassert false is always compiled because it is typed specially.This flag has no effect when linking already-compiled files.
  • -noautolink
  • When linking .cmxalibraries, ignore -cclib and -ccoptoptions potentially contained in the libraries (if these options weregiven when building the libraries). This can be useful if a librarycontains incorrect specifications of C libraries or C options; in thiscase, during linking, set -noautolink and pass the correct Clibraries and options on the command line.
  • -nodynlink
  • Allow the compiler to use some optimizations that are valid only for codethat is never dynlinked.
  • -nolabels
  • Ignore non-optional labels in types. Labels cannot be used inapplications, and parameter order becomes strict.
  • -nostdlib
  • Do not automatically add the standard library directory to the list ofdirectories searched for compiled interface files (.cmi), compiledobject code files (.cmx), and libraries (.cmxa). See also option-I.
  • -oexec-file
  • Specify the name of the output file produced by thelinker. Thedefault output name is a.out under Unix and camlprog.exe underWindows. If the -a option is given, specify the name of the libraryproduced. If the -pack option is given, specify the name of thepacked object file produced. If the -output-obj option is given,specify the name of the output file produced.If the -shared option is given, specify the name of pluginfile produced.
  • -opaque
  • When the native compiler compiles an implementation, by default itproduces a .cmx file containing information for cross-moduleoptimization. It also expects .cmx files to be present for thedependencies of the currently compiled source, and uses them foroptimization. Since OCaml 4.03, the compiler will emit a warning if itis unable to locate the .cmx file of one of those dependencies.The -opaque option, available since 4.04, disables cross-moduleoptimization information for the currently compiled unit. Whencompiling .mli interface, using -opaque marks the compiled .cmiinterface so that subsequent compilations of modules that depend on itwill not rely on the corresponding .cmx file, nor warn if it isabsent. When the native compiler compiles a .ml implementation,using -opaque generates a .cmx that does not contain anycross-module optimization information.

Using this option may degrade the quality of generated code, but itreduces compilation time, both on clean and incrementalbuilds. Indeed, with the native compiler, when the implementation ofa compilation unit changes, all the units that depend on it may needto be recompiled – because the cross-module information may havechanged. If the compilation unit whose implementation changed wascompiled with -opaque, no such recompilation needs to occur. Thisoption can thus be used, for example, to get faster edit-compile-testfeedback loops.

  • -openModule
  • Opens the given module before processing the interface orimplementation files. If several -open options are given,they are processed in order, just as ifthe statements open!Module1;;…open!ModuleN;;were added at the top of each file.
  • -output-obj
  • Cause the linker to produce a C object file instead ofan executable file.This is useful to wrap OCaml code as a C library,callable from any C program. See chapter 19,section 19.7.5. The name of the output object filemust be set with the -o option.This option can also be used to produce a compiled shared/dynamic library (.so extension, .dll under Windows).
  • -p
  • Generate extra code to write profile information when the program isexecuted. The profile information can then be examined with theanalysis program gprof. (See chapter 17 for moreinformation on profiling.) The -p option must be given both atcompile-time and at link-time. Linking object files not compiled with-p is possible, but results in less precise profiling.
Unix: See the Unix manual page for gprof(1) for moreinformation about the profiles.

Full support for gprof is only available for certain platforms(currently: Intel x86 32 and 64 bits under Linux, BSD and MacOS X).On other platforms, the -p option will result in a less preciseprofile (no call graph information, only a time profile).

Windows:The -p option does not work under Windows.
  • -pack
  • Build an object file (.cmx and .o/.obj files) and its associated compiledinterface (.cmi) that combines the .cmx objectfiles given on the command line, making them appear as sub-modules ofthe output .cmx file. The name of the output .cmx file must begiven with the -o option. For instance,
  1.         ocamlopt -pack -o P.cmx A.cmx B.cmx C.cmx

generates compiled files P.cmx, P.o and P.cmi describing acompilation unit having three sub-modules A, B and C,corresponding to the contents of the object files A.cmx, B.cmx andC.cmx. These contents can be referenced as P.A, P.B and P.Cin the remainder of the program.The .cmx object files being combined must have been compiled withthe appropriate -for-pack option. In the example above,A.cmx, B.cmx and C.cmx must have been compiled withocamlopt -for-pack P.

Multiple levels of packing can be achieved by combining -pack with-for-pack. Consider the following example:

  1.         ocamlopt -for-pack P.Q -c A.ml
  2.         ocamlopt -pack -o Q.cmx -for-pack P A.cmx
  3.         ocamlopt -for-pack P -c B.ml
  4.         ocamlopt -pack -o P.cmx Q.cmx B.cmx

The resulting P.cmx object file has sub-modules P.Q, P.Q.Aand P.B.

  • -pluginplugin
  • Dynamically load the code of the given plugin(a .cmo, .cma or .cmxs file) in the compiler. plugin must exist inthe same kind of code as the compiler (ocamlopt.byte must loadbytecode plugins, while ocamlopt.opt must load native code plugins),and extension adaptation is done automatically for .cma files (to .cmxsfiles if the compiler is compiled in native code).
  • -ppcommand
  • Cause the compiler to call the given command as a preprocessorfor each source file. The output of command is redirected toan intermediate file, which is compiled. If there are no compilationerrors, the intermediate file is deleted afterwards.
  • -ppxcommand
  • After parsing, pipe the abstract syntax tree through the preprocessorcommand. The module Ast_mapper, described inchapter 24:Ast_mapper,implements the external interface of a preprocessor.
  • -principal
  • Check information path during type-checking, to make sure that alltypes are derived in a principal way. When using labelled argumentsand/or polymorphic methods, this flag is required to ensure futureversions of the compiler will be able to infer types correctly, evenif internal algorithms change.All programs accepted in -principal mode are also accepted in thedefault mode with equivalent types, but different binary signatures,and this may slow down type checking; yet it is a good idea touse it once before publishing source code.
  • -rectypes
  • Allow arbitrary recursive types during type-checking. By default,only recursive types where the recursion goes through an object typeare supported.Note that once you have created an interface using thisflag, you must use it again for all dependencies.
  • -runtime-variantsuffix
  • Add the suffix string to the name of the runtime library used bythe program. Currently, only one such suffix is supported: d, andonly if the OCaml compiler was configured with option-with-debug-runtime. This suffix gives the debug version of theruntime, which is useful for debugging pointer problems in low-levelcode such as C stubs.
  • -stop-afterpass
  • Stop compilation after the given compilation pass. The currentlysupported passes are: parsing, typing.
  • -S
  • Keep the assembly code produced during the compilation. The assemblycode for the source file x.ml is saved in the file x.s.
  • -shared
  • Build a plugin (usually .cmxs) that can be dynamically loaded withthe Dynlink module. The name of the plugin must beset with the -o option. A plugin can include a number of OCamlmodules and libraries, and extra native objects (.o, .obj, .a,.lib files). Building native plugins is only supported for someoperating system. Under some systems (currently,only Linux AMD 64), all the OCaml code linked in a plugin must havebeen compiled without the -nodynlink flag. Some constraints might alsoapply to the way the extra native objects have been compiled (underLinux AMD 64, they must contain only position-independent code).
  • -safe-string
  • Enforce the separation between types string and bytes,thereby making strings read-only. This is the default.
  • -short-paths
  • When a type is visible under several module-paths, use the shortestone when printing the type’s name in inferred interfaces and error andwarning messages. Identifier names starting with an underscore orcontaining double underscores _ incur a penalty of +10 when computingtheir length.
  • -strict-sequence
  • Force the left-hand part of each sequence to have type unit.
  • -strict-formats
  • Reject invalid formats that were accepted in legacy formatimplementations. You should use this flag to detect and fix suchinvalid formats, as they will be rejected by future OCaml versions.
  • -unboxed-types
  • When a type is unboxable (i.e. a record with a single argument or aconcrete datatype with a single constructor of one argument) it willbe unboxed unless annotated with [@@ocaml.boxed].
  • -no-unboxed-types
  • When a type is unboxable it will be boxed unless annotated with[@@ocaml.unboxed]. This is the default.
  • -unsafe
  • Turn bound checking off for array and string accesses (the v.(i) ands.[i] constructs). Programs compiled with -unsafe are therefore faster, but unsafe: anything can happen if the programaccesses an array or string outside of its bounds.Additionally, turn off the check for zero divisor in integer divisionand modulus operations. With -unsafe, an integer division(or modulus) by zero can halt the program or continue with anunspecified result instead of raising a Division_by_zero exception.
  • -unsafe-string
  • Identify the types string and bytes, thereby making strings writable.This is intended for compatibility with old source code and should notbe used with new software.
  • -v
  • Print the version number of the compiler and the location of thestandard library directory, then exit.
  • -verbose
  • Print all external commands before they are executed,in particular invocations of the assembler, C compiler, and linker.Useful to debug C library problems.
  • -version or -vnum
  • Print the version number of the compiler in short form (e.g. 3.11.0),then exit.
  • -wwarning-list

  • Enable, disable, or mark as fatal the warnings specified by the argumentwarning-list.Each warning can be enabled or disabled, and each warningcan be fatal or non-fatal.If a warning is disabled, it isn’t displayed and doesn’t affectcompilation in any way (even if it is fatal). If a warning isenabled, it is displayed normally by the compiler whenever the sourcecode triggers it. If it is enabled and fatal, the compiler will alsostop with an error after displaying it.The warning-list argument is a sequence of warning specifiers,with no separators between them. A warning specifier is one of thefollowing:

    • +num
    • Enable warning number num.
    • -num
    • Disable warning number num.
    • @num
    • Enable and mark as fatal warning number num.
    • +num1..num2
    • Enable warnings in the given range.
    • -num1..num2
    • Disable warnings in the given range.
    • @num1..num2
    • Enable and mark as fatal warnings inthe given range.
    • +letter
    • Enable the set of warnings corresponding toletter. The letter may be uppercase or lowercase.
    • -letter
    • Disable the set of warnings corresponding toletter. The letter may be uppercase or lowercase.
    • @letter
    • Enable and mark as fatal the set of warningscorresponding to letter. The letter may be uppercase orlowercase.
    • uppercase-letter
    • Enable the set of warnings correspondingto uppercase-letter.
    • lowercase-letter

    • Disable the set of warnings correspondingto lowercase-letter.Warning numbers and letters which are out of the range of warningsthat are currently defined are ignored. The warnings are as follows.

    • 1

    • Suspicious-looking start-of-comment mark.
    • 2
    • Suspicious-looking end-of-comment mark.
    • 3
    • Deprecated synonym for the ’deprecated’ alert
    • 4
    • Fragile pattern matching: matching that will remain complete evenif additional constructors are added to one of the variant typesmatched.
    • 5
    • Partially applied function: expression whose result has functiontype and is ignored.
    • 6
    • Label omitted in function application.
    • 7
    • Method overridden.
    • 8
    • Partial match: missing cases in pattern-matching.
    • 9
    • Missing fields in a record pattern.
    • 10
    • Expression on the left-hand side of a sequence that doesn’t have typeunit (and that is not a function, see warning number 5).
    • 11
    • Redundant case in a pattern matching (unused match case).
    • 12
    • Redundant sub-pattern in a pattern-matching.
    • 13
    • Instance variable overridden.
    • 14
    • Illegal backslash escape in a string constant.
    • 15
    • Private method made public implicitly.
    • 16
    • Unerasable optional argument.
    • 17
    • Undeclared virtual method.
    • 18
    • Non-principal type.
    • 19
    • Type without principality.
    • 20
    • Unused function argument.
    • 21
    • Non-returning statement.
    • 22
    • Preprocessor warning.
    • 23
    • Useless record with clause.
    • 24
    • Bad module name: the source file name is not a valid OCaml module name.
    • 25
    • Deprecated: now part of warning 8.
    • 26
    • Suspicious unused variable: unused variable that is boundwith let or as, and doesn’t start with an underscore (_)character.
    • 27
    • Innocuous unused variable: unused variable that is not bound withlet nor as, and doesn’t start with an underscore (_)character.
    • 28
    • Wildcard pattern given as argument to a constant constructor.
    • 29
    • Unescaped end-of-line in a string constant (non-portable code).
    • 30
    • Two labels or constructors of the same name are defined in twomutually recursive types.
    • 31
    • A module is linked twice in the same executable.
    • 32
    • Unused value declaration.
    • 33
    • Unused open statement.
    • 34
    • Unused type declaration.
    • 35
    • Unused for-loop index.
    • 36
    • Unused ancestor variable.
    • 37
    • Unused constructor.
    • 38
    • Unused extension constructor.
    • 39
    • Unused rec flag.
    • 40
    • Constructor or label name used out of scope.
    • 41
    • Ambiguous constructor or label name.
    • 42
    • Disambiguated constructor or label name (compatibility warning).
    • 43
    • Nonoptional label applied as optional.
    • 44
    • Open statement shadows an already defined identifier.
    • 45
    • Open statement shadows an already defined label or constructor.
    • 46
    • Error in environment variable.
    • 47
    • Illegal attribute payload.
    • 48
    • Implicit elimination of optional arguments.
    • 49
    • Absent cmi file when looking up module alias.
    • 50
    • Unexpected documentation comment.
    • 51
    • Warning on non-tail calls if @tailcall present.
    • 52 (see 9.5.2)
    • Fragile constant pattern.
    • 53
    • Attribute cannot appear in this context
    • 54
    • Attribute used more than once on an expression
    • 55
    • Inlining impossible
    • 56
    • Unreachable case in a pattern-matching (based on type information).
    • 57 (see 9.5.3)
    • Ambiguous or-pattern variables under guard
    • 58
    • Missing cmx file
    • 59
    • Assignment to non-mutable value
    • 60
    • Unused module declaration
    • 61
    • Unboxable type in primitive declaration
    • 62
    • Type constraint on GADT type declaration
    • 63
    • Erroneous printed signature
    • 64
    • -unsafe used with a preprocessor returning a syntax tree
    • 65
    • Type declaration defining a new ’()’ constructor
    • 66
    • Unused open! statement
    • A
    • all warnings
    • C
    • warnings 1, 2.
    • D
    • Alias for warning 3.
    • E
    • Alias for warning 4.
    • F
    • Alias for warning 5.
    • K
    • warnings 32, 33, 34, 35, 36, 37, 38, 39.
    • L
    • Alias for warning 6.
    • M
    • Alias for warning 7.
    • P
    • Alias for warning 8.
    • R
    • Alias for warning 9.
    • S
    • Alias for warning 10.
    • U
    • warnings 11, 12.
    • V
    • Alias for warning 13.
    • X
    • warnings 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 30.
    • Y
    • Alias for warning 26.
    • Z
    • Alias for warning 27.The default setting is -w +a-4-6-7-9-27-29-32..42-44-45-48-50-60.It is displayed by ocamlopt -help.Note that warnings 5 and 10 are not always triggered, depending onthe internals of the type checker.

  • -warn-errorwarning-list

  • Mark as fatal the warnings specified in the argument warning-list.The compiler will stop with an error when one of these warnings isemitted. The warning-list has the same meaning as forthe -w option: a + sign (or an uppercase letter) marks thecorresponding warnings as fatal, a -sign (or a lowercase letter) turns them back into non-fatal warnings,and a @ sign both enables and marks as fatal the correspondingwarnings.Note: it is not recommended to use warning sets (i.e. letters) asarguments to -warn-errorin production code, because this can break your build when future versionsof OCaml add some new warnings.

The default setting is -warn-error -a+31 (only warning 31 is fatal).

  • -warn-help
  • Show the description of all available warning numbers.
  • -where
  • Print the location of the standard library, then exit.
  • -file
  • Process file as a file name, even if it starts with a dash (-)character.
  • -help or —help
  • Display a short usage summary and exit.
Options for the IA32 architecture

The IA32 code generator (Intel Pentium, AMD Athlon) supports thefollowing additional option:

  • -ffast-math
  • Use the IA32 instructions to computetrigonometric and exponential functions, instead of calling thecorresponding library routines. The functions affected are:atan, atan2, cos, log, log10, sin, sqrt and tan.The resulting code runs faster, but the range of supported argumentsand the precision of the result can be reduced. In particular,trigonometric operations cos, sin, tan have their range reduced to[−264, 264].
Options for the AMD64 architecture

The AMD64 code generator (64-bit versions of Intel Pentium and AMDAthlon) supports the following additional options:

  • -fPIC
  • Generate position-independent machine code. This isthe default.
  • -fno-PIC
  • Generate position-dependent machine code.
Contextual control of command-line options

The compiler command line can be modified “from the outside”with the following mechanisms. These are experimentaland subject to change. They should be used only for experimental anddevelopment work, not in released packages.

  • OCAMLPARAM (environment variable)
  • A set of arguments that will be inserted before or after the arguments fromthe command line. Arguments are specified in a comma-separated listof name=value pairs. A is used to specify the position ofthe command line arguments, i.e. a=x,,b=y means that a=x should beexecuted before parsing the arguments, and b=y after. Finally,an alternative separator can be specified as thefirst character of the string, within the set :|; ,.
  • ocaml_compiler_internal_params (file in the stdlib directory)
  • A mapping of file names to lists of arguments thatwill be added to the command line (and OCAMLPARAM) arguments.
  • OCAML_FLEXLINK (environment variable)
  • Alternative executable to use on nativeWindows for flexlink instead of theconfigured value. Primarily used for bootstrapping.

12.3 Common errors

The error messages are almost identical to those of ocamlc.See section 9.4.

12.4 Running executables produced by ocamlopt

Executables generated by ocamlopt are native, stand-alone executablefiles that can be invoked directly. They donot depend on the ocamlrun bytecode runtime system nor ondynamically-loaded C/OCaml stub libraries.

During execution of an ocamlopt-generated executable,the following environment variables are also consulted:

  • OCAMLRUNPARAM
  • Same usage as in ocamlrun(see section 11.2), except that option lis ignored (the operating system’s stack size limitis used instead).
  • CAMLRUNPARAM
  • If OCAMLRUNPARAM is not found in theenvironment, then CAMLRUNPARAM will be used instead. IfCAMLRUNPARAM is not found, then the default values will be used.

12.5 Compatibility with the bytecode compiler

This section lists the known incompatibilities between the bytecodecompiler and the native-code compiler. Except on those points, the twocompilers should generate code that behave identically.

  • Signals are detected only when the program performs anallocation in the heap. That is, if a signal is delivered while in apiece of code that does not allocate, its handler will not be calleduntil the next heap allocation.
  • Stack overflow, typically caused by excessively deep recursion,is not always turned into a Stack_overflow exception like thebytecode compiler does. The runtime system makes a best effort totrap stack overflows and raise the Stack_overflow exception, butsometimes it fails and a “segmentation fault” or another system faultoccurs instead.
  • On ARM and PowerPC processors (32 and 64 bits), fusedmultiply-add (FMA) instructions can be generated for afloating-point multiplication followed by a floating-point additionor subtraction, as in x . y +. z. The FMA instruction avoidsrounding the intermediate result x . y, which is generallybeneficial, but produces floating-point results that differ slightlyfrom those produced by the bytecode interpreter.
  • On IA32 processors only (Intel and AMD x86 processors in 32-bitmode), some intermediate results in floating-point computations arekept in extended precision rather than being rounded to doubleprecision like the bytecode compiler always does. Floating-pointresults can therefore differ slightly between bytecode and native code.
  • The native-code compiler performs a number of optimizations thatthe bytecode compiler does not perform, especially when the Flambdaoptimizer is active. In particular, the native-code compileridentifies and eliminates “dead code”, i.e. computations that donot contribute to the results of the program. For example,
  1.         let _ = ignore M.f

contains a reference to compilation unit M when compiled tobytecode. This reference forces M to be linked and itsinitialization code to be executed. The native-code compilereliminates the reference to M, hence the compilation unit M maynot be linked and executed. A workaround is to compile M with the-linkall flag so that it will always be linked and executed, even ifnot referenced. See also the Sys.opaque_identity function from theSys standard library module.