1 <chapter id="using-ghc">
2 <title>Using GHC</title>
4 <indexterm><primary>GHC, using</primary></indexterm>
5 <indexterm><primary>using GHC</primary></indexterm>
7 <para>GHC can work in one of three “modes”:</para>
11 <term><cmdsynopsis><command>ghc</command>
12 <arg choice=plain>––interactive</arg>
14 <indexterm><primary>interactive mode</primary>
16 <indexterm><primary>ghci</primary>
19 <para>Interactive mode, which is also available as
20 <command>ghci</command>. Interactive mode is described in
21 more detail in <xref linkend="ghci">.</para>
26 <term><cmdsynopsis><command>ghc</command>
27 <arg choice=plain>––make</arg>
29 <indexterm><primary>make mode</primary>
31 <indexterm><primary><option>––make</option></primary>
34 <para>In this mode, GHC will build a multi-module Haskell
35 program automatically, figuring out dependencies for itself.
36 If you have a straightforward Haskell program, this is likely
37 to be much easier, and faster, than using
38 <command>make</command>.</para>
44 <command>ghc</command>
52 <indexterm><primary><option>-E</option></primary></indexterm>
53 <indexterm><primary><option>-C</option></primary></indexterm>
54 <indexterm><primary><option>-S</option></primary></indexterm>
55 <indexterm><primary><option>-c</option></primary></indexterm>
57 <para>This is the traditional batch-compiler mode, in which
58 GHC can compile source files one at a time, or link objects
59 together into an executable.</para>
65 <title>Options overview</title>
67 <para>GHC's behaviour is controlled by
68 <firstterm>options</firstterm>, which for historical reasons are
69 also sometimes referred to as command-line flags or arguments.
70 Options can be specified in three ways:</para>
73 <title>Command-line arguments</title>
75 <indexterm><primary>structure, command-line</primary></indexterm>
76 <indexterm><primary>command-line</primary><secondary>arguments</secondary></indexterm>
77 <indexterm><primary>arguments</primary><secondary>command-line</secondary></indexterm>
79 <para>An invocation of GHC takes the following form:</para>
85 <para>Command-line arguments are either options or file names.</para>
87 <para>Command-line options begin with <literal>-</literal>.
88 They may <emphasis>not</emphasis> be grouped:
89 <option>-vO</option> is different from <option>-v -O</option>.
90 Options need not precede filenames: e.g., <literal>ghc *.o -o
91 foo</literal>. All options are processed and then applied to
92 all files; you cannot, for example, invoke <literal>ghc -c -O1
93 Foo.hs -O2 Bar.hs</literal> to apply different optimisation
94 levels to the files <filename>Foo.hs</filename> and
95 <filename>Bar.hs</filename>.</para>
98 <Sect2 id="source-file-options">
99 <title>Command line options in source files</title>
101 <indexterm><primary>source-file options</primary></indexterm>
103 <para>Sometimes it is useful to make the connection between a
104 source file and the command-line options it requires quite
105 tight. For instance, if a Haskell source file uses GHC
106 extensions, it will always need to be compiled with the
107 <option>-fglasgow-exts</option> option. Rather than maintaining
108 the list of per-file options in a <filename>Makefile</filename>,
109 it is possible to do this directly in the source file using the
110 <literal>OPTIONS</literal> pragma <indexterm><primary>OPTIONS
111 pragma</primary></indexterm>:</para>
114 {-# OPTIONS -fglasgow-exts #-}
119 <para><literal>OPTIONS</literal> pragmas are only looked for at
120 the top of your source files, upto the first
121 (non-literate,non-empty) line not containing
122 <literal>OPTIONS</literal>. Multiple <literal>OPTIONS</literal>
123 pragmas are recognised. Note that your command shell does not
124 get to the source file options, they are just included literally
125 in the array of command-line arguments the compiler driver
126 maintains internally, so you'll be desperately disappointed if
127 you try to glob etc. inside <literal>OPTIONS</literal>.</para>
129 <para>NOTE: the contents of OPTIONS are prepended to the
130 command-line options, so you <emphasis>do</emphasis> have the
131 ability to override OPTIONS settings via the command
134 <para>It is not recommended to move all the contents of your
135 Makefiles into your source files, but in some circumstances, the
136 <literal>OPTIONS</literal> pragma is the Right Thing. (If you
137 use <option>-keep-hc-file-too</option> and have OPTION flags in
138 your module, the OPTIONS will get put into the generated .hc
143 <title>Setting options in GHCi</title>
145 <para>Options may also be modified from within GHCi, using the
146 <literal>:set</literal> command. See <xref linkend="ghci-set">
147 for more details.</para>
151 <sect1 id="static-dynamic-flags">
152 <title>Static vs. Dynamic options</title>
153 <indexterm><primary>static</primary><secondary>options</secondary>
155 <indexterm><primary>dynamic</primary><secondary>options</secondary>
158 <para>Each of GHC's command line options is classified as either
159 <firstterm>static</firstterm> or <firstterm>dynamic</firstterm>.
160 A static flag may only be specified on the command line, whereas a
161 dynamic flag may also be given in an <literal>OPTIONS</literal>
162 pragma in a source file or set from the GHCi command-line with
163 <literal>:set</literal>.</para>
165 <para>As a rule of thumb, all the language options are dynamic, as
166 are the warning options and the debugging options. The rest are
167 static, with the notable exceptions of <option>-v</option>,
168 <option>-cpp</option>, <option>-fasm</option>,
169 <option>-fvia-C</option>, and <option>-#include</option>.
171 The flag reference tables (<xref linkend="flag-reference">) lists
172 the status of each flag.</para>
175 <sect1 id="file-suffixes">
176 <title>Meaningful file suffixes</title>
178 <indexterm><primary>suffixes, file</primary></indexterm>
179 <indexterm><primary>file suffixes for GHC</primary></indexterm>
181 <para>File names with “meaningful” suffixes (e.g.,
182 <filename>.lhs</filename> or <filename>.o</filename>) cause the
183 “right thing” to happen to those files.</para>
188 <term><filename>.lhs</filename></term>
189 <indexterm><primary><literal>lhs</literal> suffix</primary></indexterm>
191 <para>A “literate Haskell” module.</para>
196 <term><filename>.hs</filename></term>
198 <para>A not-so-literate Haskell module.</para>
203 <term><filename>.hi</filename></term>
205 <para>A Haskell interface file, probably
206 compiler-generated.</para>
211 <term><filename>.hc</filename></term>
213 <para>Intermediate C file produced by the Haskell
219 <term><filename>.c</filename></term>
221 <para>A C file not produced by the Haskell
227 <term><filename>.s</filename></term>
229 <para>An assembly-language source file, usually produced by
235 <term><filename>.o</filename></term>
237 <para>An object file, produced by an assembler.</para>
242 <para>Files with other suffixes (or without suffixes) are passed
243 straight to the linker.</para>
247 <sect1 id="options-help">
248 <title>Help and verbosity options</title>
250 <IndexTerm><Primary>help options</Primary></IndexTerm>
251 <IndexTerm><Primary>verbosity options</Primary></IndexTerm>
255 <term><option>––help</option></term>
256 <term><option>-?</option></term>
257 <indexterm><primary><option>-?</option></primary></indexterm>
258 <indexterm><primary><option>––help</option></primary></indexterm>
260 <para>Cause GHC to spew a long usage message to standard
261 output and then exit.</para>
266 <term><option>-v</option></term>
267 <indexterm><primary><option>-v</option></primary></indexterm>
269 <para>The <option>-v</option> option makes GHC
270 <emphasis>verbose</emphasis>: it reports its version number
271 and shows (on stderr) exactly how it invokes each phase of
272 the compilation system. Moreover, it passes the
273 <option>-v</option> flag to most phases; each reports its
274 version number (and possibly some other information).</para>
276 <para>Please, oh please, use the <option>-v</option> option
277 when reporting bugs! Knowing that you ran the right bits in
278 the right order is always the first thing we want to
284 <term><option>-v</option><replaceable>n</replaceable></term>
285 <indexterm><primary><option>-v</option></primary></indexterm>
287 <para>To provide more control over the compiler's verbosity,
288 the <option>-v</option> flag takes an optional numeric
289 argument. Specifying <option>-v</option> on its own is
290 equivalent to <option>-v3</option>, and the other levels
291 have the following meanings:</para>
295 <term><option>-v0</option></term>
297 <para>Disable all non-essential messages (this is the
303 <term><option>-v1</option></term>
305 <para>Minimal verbosity: print one line per
306 compilation (this is the default when
307 <option>––make</option> or
308 <option>––interactive</option> is on).</para>
313 <term><option>-v2</option></term>
315 <para>Print the name of each compilation phase as it
316 is executed. (equivalent to
317 <option>-dshow-passes</option>).</para>
322 <term><option>-v3</option></term>
324 <para>The same as <option>-v2</option>, except that in
325 addition the full command line (if appropriate) for
326 each compilation phase is also printed.</para>
331 <term><option>-v4</option></term>
333 <para>The same as <option>-v3</option> except that the
334 intermediate program representation after each
335 compilation phase is also printed (excluding
336 preprocessed and C/assembly files).</para>
344 <term><option>-V</option></term>
345 <term><option>––version</option></term>
346 <indexterm><primary><option>-V</option></primary></indexterm>
347 <indexterm><primary><option>––version</option></primary></indexterm>
349 <para>Print a one-line string including GHC's version number.</para>
354 <term><option>––numeric-version</option></term>
355 <indexterm><primary><option>––numeric-version</option></primary></indexterm>
357 <para>Print GHC's numeric version number only.</para>
362 <term><option>––print-libdir</option></term>
363 <indexterm><primary><option>––print-libdir</option></primary></indexterm>
365 <para>Print the path to GHC's library directory. This is
366 the top of the directory tree containing GHC's libraries,
367 interfaces, and include files (usually something like
368 <literal>/usr/local/lib/ghc-5.04</literal> on Unix). This
370 <literal>$libdir</literal><indexterm><primary><literal>libdir</literal></primary>
371 </indexterm>in the package configuration file (see <xref
372 linkend="packages">).</para>
379 <sect1 id="make-mode">
380 <title>Using <command>ghc</command> <option>––make</option></title>
382 <indexterm><primary><option>––make</option></primary>
384 <indexterm><primary>separate compilation</primary>
387 <para>When given the <option>––make</option> option, GHC will
388 build a multi-module Haskell program by following dependencies
389 from a single root module (usually <literal>Main</literal>). For
390 example, if your <literal>Main</literal> module is in a file
391 called <filename>Main.hs</filename>, you could compile and link
392 the program like this:</para>
395 ghc ––make Main.hs
398 <para>The command line may contain any number of source file names
399 or module names; GHC will figure out all the modules in the
400 program by following the imports from these initial modules. It
401 will then attempt to compile each module which is out of date, and
402 finally if there is a <literal>Main</literal> module, the program
403 will also be linked into an executable.</para>
405 <para>The main advantages to using <literal>ghc ––make</literal>
406 over traditional <literal>Makefile</literal>s are:</para>
410 <para>GHC doesn't have to be restarted for each compilation,
411 which means it can cache information between compilations.
412 Compiling a muli-module program with <literal>ghc
413 ––make</literal> can be up to twice as fast as running
414 <literal>ghc</literal> individually on each source
418 <para>You don't have to write a
419 <literal>Makefile</literal>.</para>
421 <indexterm><primary><literal>Makefile</literal>s</primary><secondary>avoiding</secondary>
424 <para>GHC re-calculates the dependencies each time it is
425 invoked, so the dependencies never get out of sync with the
430 <para>Any of the command-line options described in the rest of
431 this chapter can be used with <option>––make</option>, but note
432 that any options you give on the command line will apply to all
433 the source files compiled, so if you want any options to apply to
434 a single source file only, you'll need to use an
435 <literal>OPTIONS</literal> pragma (see <xref
436 linkend="source-file-options">).</para>
438 <para>If the program needs to be linked with additional objects
439 (say, some auxilliary C code), then the object files can be
440 given on the command line and GHC will include them when linking
441 the executable.</para>
443 <para>Note that GHC can only follow dependencies if it has the
444 source file available, so if your program includes a module for
445 which there is no source file, even if you have an object and an
446 interface file for the module, then GHC will complain. The
447 exception to this rule is for package modules, which may or may
448 not have source files.</para>
450 <para>The source files for the program don't all need to be in the
451 same directory; the <option>-i</option> option can be used to add
452 directories to the search path (see <xref
453 linkend="search-path">).</para>
457 <Sect1 id="options-order">
458 <title>GHC without <option>––make</option></title>
460 <para>Without <option>––make</option>, GHC will compile one or
461 more source files given on the command line.</para>
463 <para>The first phase to run is determined by each input-file
464 suffix, and the last phase is determined by a flag. If no
465 relevant flag is present, then go all the way through linking.
466 This table summarises:</para>
470 <colspec align="left">
471 <colspec align="left">
472 <colspec align="left">
473 <colspec align="left">
477 <entry>Phase of the compilation system</entry>
478 <entry>Suffix saying “start here”</entry>
479 <entry>Flag saying “stop after”</entry>
480 <entry>(suffix of) output file</entry>
485 <entry>literate pre-processor</entry>
486 <entry><literal>.lhs</literal></entry>
488 <entry><literal>.hs</literal></entry>
492 <entry>C pre-processor (opt.)
494 <entry><literal>.hs</literal> (with
495 <option>-cpp</option>)</entry>
496 <entry><option>-E</option></entry>
497 <entry><literal>.hspp</literal></entry>
501 <entry>Haskell compiler</entry>
502 <entry><literal>.hs</literal></entry>
503 <entry><option>-C</option>, <option>-S</option></entry>
504 <entry><literal>.hc</literal>, <literal>.s</literal></entry>
508 <entry>C compiler (opt.)</entry>
509 <entry><literal>.hc</literal> or <literal>.c</literal></entry>
510 <entry><option>-S</option></entry>
511 <entry><literal>.s</literal></entry>
515 <entry>assembler</entry>
516 <entry><literal>.s</literal></entry>
517 <entry><option>-c</option></entry>
518 <entry><literal>.o</literal></entry>
522 <entry>linker</entry>
523 <entry><replaceable>other</replaceable></entry>
525 <entry><filename>a.out</filename></entry>
531 <indexterm><primary><option>-C</option></primary></indexterm>
532 <indexterm><primary><option>-E</option></primary></indexterm>
533 <indexterm><primary><option>-S</option></primary></indexterm>
534 <indexterm><primary><option>-c</option></primary></indexterm>
536 <para>Thus, a common invocation would be: <literal>ghc -c
537 Foo.hs</literal></para>
539 <para>Note: What the Haskell compiler proper produces depends on
540 whether a native-code generator<indexterm><primary>native-code
541 generator</primary></indexterm> is used (producing assembly
542 language) or not (producing C). See <xref
543 linkend="options-codegen"> for more details.</para>
545 <para>Note: C pre-processing is optional, the
546 <option>-ccp</option><indexterm><primary><option>-cpp</option></primary>
547 </indexterm>flag turns it on. See <xref
548 linkend="c-pre-processor"> for more details.</para>
550 <para>Note: The option <option>-E</option><IndexTerm><Primary>-E
551 option</Primary></IndexTerm> runs just the pre-processing passes
552 of the compiler, dumping the result in a file. Note that this
553 differs from the previous behaviour of dumping the file to
554 standard output.</para>
559 <sect1 id="options-sanity">
560 <title>Warnings and sanity-checking</title>
562 <indexterm><primary>sanity-checking options</primary></indexterm>
563 <indexterm><primary>warnings</primary></indexterm>
566 <para>GHC has a number of options that select which types of
567 non-fatal error messages, otherwise known as warnings, can be
568 generated during compilation. By default, you get a standard set
569 of warnings which are generally likely to indicate bugs in your
571 <option>-fwarn-overlapping-patterns</option>,
572 <option>-fwarn-deprecations</option>,
573 <option>-fwarn-duplicate-exports</option>,
574 <option>-fwarn-missing-fields</option>, and
575 <option>-fwarn-missing-methods</option>. The following flags are
576 simple ways to select standard “packages” of warnings:
582 <term><option>-W</option>:</term>
584 <IndexTerm><Primary>-W option</Primary></IndexTerm>
585 <para>Provides the standard warnings plus
586 <option>-fwarn-incomplete-patterns</option>,
587 <option>-fwarn-unused-matches</option>,
588 <option>-fwarn-unused-imports</option>,
589 <option>-fwarn-misc</option>, and
590 <option>-fwarn-unused-binds</option>.</para>
595 <term><option>-w</option>:</term>
597 <IndexTerm><Primary><option>-w</option></Primary></IndexTerm>
598 <para>Turns off all warnings, including the standard ones.</para>
603 <term><option>-Wall</option>:</term>
605 <indexterm><primary><option>-Wall</option></primary></indexterm>
606 <para>Turns on all warning options.</para>
611 <term><option>-Werror</option>:</term>
613 <indexterm><primary><option>-Werror</option></primary></indexterm>
614 <para>Makes any warning into a fatal error. Useful so that you don't
615 miss warnings when doing batch compilation. </para>
621 <para>The full set of warning options is described below. To turn
622 off any warning, simply give the corresponding
623 <option>-fno-warn-...</option> option on the command line.</para>
628 <term><option>-fwarn-deprecations</option>:</term>
630 <indexterm><primary><option>-fwarn-deprecations</option></primary>
632 <indexterm><primary>deprecations</primary></indexterm>
633 <para>Causes a warning to be emitted when a deprecated
634 function or type is used. Entities can be marked as
635 deprecated using a pragma, see <xref
636 linkend="deprecated-pragma">.</para>
641 <term><option>-fwarn-duplicate-exports</option>:</term>
643 <indexterm><primary><option>-fwarn-duplicate-exports</option></primary></indexterm>
644 <indexterm><primary>duplicate exports, warning</primary></indexterm>
645 <indexterm><primary>export lists, duplicates</primary></indexterm>
647 <para>Have the compiler warn about duplicate entries in
648 export lists. This is useful information if you maintain
649 large export lists, and want to avoid the continued export
650 of a definition after you've deleted (one) mention of it in
651 the export list.</para>
653 <para>This option is on by default.</para>
658 <term><option>-fwarn-hi-shadowing</option>:</term>
660 <indexterm><primary><option>-fwarn-hi-shadowing</option></primary></indexterm>
661 <indexterm><primary>shadowing</primary>
662 <secondary>interface files</secondary></indexterm>
664 <para>Causes the compiler to emit a warning when a module or
665 interface file in the current directory is shadowing one
666 with the same module name in a library or other
672 <term><option>-fwarn-incomplete-patterns</option>:</term>
674 <indexterm><primary><option>-fwarn-incomplete-patterns</option></primary></indexterm>
675 <indexterm><primary>incomplete patterns, warning</primary></indexterm>
676 <indexterm><primary>patterns, incomplete</primary></indexterm>
678 <para>Similarly for incomplete patterns, the function
679 <function>g</function> below will fail when applied to
680 non-empty lists, so the compiler will emit a warning about
681 this when <option>-fwarn-incomplete-patterns</option> is
688 <para>This option isn't enabled be default because it can be
689 a bit noisy, and it doesn't always indicate a bug in the
690 program. However, it's generally considered good practice
691 to cover all the cases in your functions.</para>
696 <term><option>-fwarn-misc</option>:</term>
697 <indexterm><primary><option>-fwarn-misc</option></primary></indexterm>
699 <para>Turns on warnings for various harmless but untidy
700 things. This currently includes: importing a type with
701 <literal>(..)</literal> when the export is abstract, and
702 listing duplicate class assertions in a qualified type.</para>
707 <term><option>-fwarn-missing-fields</option>:</term>
709 <indexterm><primary><option>-fwarn-missing-fields</option></primary></indexterm>
710 <indexterm><primary>missing fields, warning</primary></indexterm>
711 <indexterm><primary>fields, missing</primary></indexterm>
713 <para>This option is on by default, and warns you whenever
714 the construction of a labelled field constructor isn't
715 complete, missing initializers for one or more fields. While
716 not an error (the missing fields are initialised with
717 bottoms), it is often an indication of a programmer error.</para>
722 <term><option>-fwarn-missing-methods</option>:</term>
724 <indexterm><primary><option>-fwarn-missing-methods</option></primary></indexterm>
725 <indexterm><primary>missing methods, warning</primary></indexterm>
726 <indexterm><primary>methods, missing</primary></indexterm>
728 <para>This option is on by default, and warns you whenever
729 an instance declaration is missing one or more methods, and
730 the corresponding class declaration has no default
731 declaration for them.</para>
732 <para>The warning is suppressed if the method name
733 begins with an underscore. Here's an example where this is useful:
736 _simpleFn :: a -> String
737 complexFn :: a -> a -> String
738 complexFn x y = ... _simpleFn ...
740 The idea is that: (a) users of the class will only call <literal>complexFn</literal>;
741 never <literal>_simpleFn</literal>; and (b)
742 instance declarations can define either <literal>complexFn</literal> or <literal>_simpleFn</literal>.
748 <term><option>-fwarn-missing-signatures</option>:</term>
750 <indexterm><primary><option>-fwarn-missing-signatures</option></primary></indexterm>
751 <indexterm><primary>type signatures, missing</primary></indexterm>
753 <para>If you would like GHC to check that every top-level
754 function/value has a type signature, use the
755 <option>-fwarn-missing-signatures</option> option. This
756 option is off by default.</para>
761 <term><option>-fwarn-name-shadowing</option>:</term>
763 <indexterm><primary><option>-fwarn-name-shadowing</option></primary></indexterm>
764 <indexterm><primary>shadowing, warning</primary></indexterm>
766 <para>This option causes a warning to be emitted whenever an
767 inner-scope value has the same name as an outer-scope value,
768 i.e. the inner value shadows the outer one. This can catch
769 typographical errors that turn into hard-to-find bugs, e.g.,
770 in the inadvertent cyclic definition <literal>let x = ... x
771 ... in</literal>.</para>
773 <para>Consequently, this option does
774 <emphasis>will</emphasis> complain about cyclic recursive
780 <term><option>-fwarn-overlapping-patterns</option>:</term>
781 <indexterm><primary><option>-fwarn-overlapping-patterns</option></primary></indexterm>
782 <indexterm><primary>overlapping patterns, warning</primary></indexterm>
783 <indexterm><primary>patterns, overlapping</primary></indexterm>
785 <para>By default, the compiler will warn you if a set of
786 patterns are overlapping, i.e.,</para>
789 f :: String -> Int
795 <para>where the last pattern match in <Function>f</Function>
796 won't ever be reached, as the second pattern overlaps
797 it. More often than not, redundant patterns is a programmer
798 mistake/error, so this option is enabled by default.</para>
803 <term><option>-fwarn-simple-patterns</option>:</term>
805 <indexterm><primary><option>-fwarn-simple-patterns</option></primary>
807 <para>Causes the compiler to warn about lambda-bound
808 patterns that can fail, eg. <literal>\(x:xs)->...</literal>.
809 Normally, these aren't treated as incomplete patterns by
810 <option>-fwarn-incomplete-patterns</option>.</para>
811 <para>``Lambda-bound patterns'' includes all places where there is a single pattern,
812 including list comprehensions and do-notation. In these cases, a pattern-match
813 failure is quite legitimate, and triggers filtering (list comprehensions) or
814 the monad <literal>fail</literal> operation (monads). For example:
816 f :: [Maybe a] -> [a]
817 f xs = [y | Just y <- xs]
819 Switching on <option>-fwarn-simple-patterns</option> will elicit warnings about
820 these probably-innocent cases, which is why the flag is off by default. </para>
821 <para> The <literal>deriving( Read )</literal> mechanism produces monadic code with
822 pattern matches, so you will also get misleading warnings about the compiler-generated
823 code. (This is arguably a Bad Thing, but it's awkward to fix.)</para>
829 <term><option>-fwarn-type-defaults</option>:</term>
831 <indexterm><primary><option>-fwarn-type-defaults</option></primary></indexterm>
832 <indexterm><primary>defaulting mechanism, warning</primary></indexterm>
833 <para>Have the compiler warn/inform you where in your source
834 the Haskell defaulting mechanism for numeric types kicks
835 in. This is useful information when converting code from a
836 context that assumed one default into one with another,
837 e.g., the `default default' for Haskell 1.4 caused the
838 otherwise unconstrained value <Constant>1</Constant> to be
839 given the type <literal>Int</literal>, whereas Haskell 98
840 defaults it to <literal>Integer</literal>. This may lead to
841 differences in performance and behaviour, hence the
842 usefulness of being non-silent about this.</para>
844 <para>This warning is off by default.</para>
849 <term><option>-fwarn-unused-binds</option>:</term>
851 <indexterm><primary><option>-fwarn-unused-binds</option></primary></indexterm>
852 <indexterm><primary>unused binds, warning</primary></indexterm>
853 <indexterm><primary>binds, unused</primary></indexterm>
854 <para>Report any function definitions (and local bindings)
855 which are unused. For top-level functions, the warning is
856 only given if the binding is not exported.</para>
861 <term><option>-fwarn-unused-imports</option>:</term>
863 <indexterm><primary><option>-fwarn-unused-imports</option></primary></indexterm>
864 <indexterm><primary>unused imports, warning</primary></indexterm>
865 <indexterm><primary>imports, unused</primary></indexterm>
867 <para>Report any objects that are explicitly imported but
873 <term><option>-fwarn-unused-matches</option>:</term>
875 <indexterm><primary><option>-fwarn-unused-matches</option></primary></indexterm>
876 <indexterm><primary>unused matches, warning</primary></indexterm>
877 <indexterm><primary>matches, unused</primary></indexterm>
879 <para>Report all unused variables which arise from pattern
880 matches, including patterns consisting of a single variable.
881 For instance <literal>f x y = []</literal> would report
882 <VarName>x</VarName> and <VarName>y</VarName> as unused. The
883 warning is suppressed if the variable name begins with an underscore, thus:
893 <para>If you're feeling really paranoid, the
894 <option>-dcore-lint</option>
895 option<indexterm><primary><option>-dcore-lint</option></primary></indexterm>
896 is a good choice. It turns on heavyweight intra-pass
897 sanity-checking within GHC. (It checks GHC's sanity, not
904 <sect1 id="options-optimise">
905 <title>Optimisation (code improvement)</title>
907 <indexterm><primary>optimisation</primary></indexterm>
908 <indexterm><primary>improvement, code</primary></indexterm>
910 <para>The <option>-O*</option> options specify convenient
911 “packages” of optimisation flags; the
912 <option>-f*</option> options described later on specify
913 <emphasis>individual</emphasis> optimisations to be turned on/off;
914 the <option>-m*</option> options specify
915 <emphasis>machine-specific</emphasis> optimisations to be turned
918 <sect2 id="optimise-pkgs">
919 <title><option>-O*</option>: convenient “packages” of optimisation flags.</title>
921 <para>There are <emphasis>many</emphasis> options that affect
922 the quality of code produced by GHC. Most people only have a
923 general goal, something like “Compile quickly” or
924 “Make my program run like greased lightning.” The
925 following “packages” of optimisations (or lack
926 thereof) should suffice.</para>
928 <para>Once you choose a <option>-O*</option>
929 “package,” stick with it—don't chop and
930 change. Modules' interfaces <emphasis>will</emphasis> change
931 with a shift to a new <option>-O*</option> option, and you may
932 have to recompile a large chunk of all importing modules before
933 your program can again be run safely (see <XRef
934 LinkEnd="recomp">).</para>
939 <term>No <option>-O*</option>-type option specified:</term>
940 <indexterm><primary>-O* not specified</primary></indexterm>
942 <para>This is taken to mean: “Please compile
943 quickly; I'm not over-bothered about compiled-code
944 quality.” So, for example: <command>ghc -c
945 Foo.hs</command></para>
950 <term><option>-O0</option>:</term>
951 <indexterm><primary><option>-O0</option></primary></indexterm>
953 <para>Means “turn off all optimisation”,
954 reverting to the same settings as if no
955 <option>-O</option> options had been specified. Saying
956 <option>-O0</option> can be useful if
957 eg. <command>make</command> has inserted a
958 <option>-O</option> on the command line already.</para>
963 <term><option>-O</option> or <option>-O1</option>:</term>
964 <indexterm><primary>-O option</primary></indexterm>
965 <indexterm><primary>-O1 option</primary></indexterm>
966 <indexterm><primary>optimise</primary><secondary>normally</secondary></indexterm>
968 <para>Means: “Generate good-quality code without
969 taking too long about it.” Thus, for example:
970 <command>ghc -c -O Main.lhs</command></para>
975 <term><option>-O2</option>:</term>
976 <indexterm><primary>-O2 option</primary></indexterm>
977 <indexterm><primary>optimise</primary><secondary>aggressively</secondary></indexterm>
979 <para>Means: “Apply every non-dangerous
980 optimisation, even if it means significantly longer
981 compile times.”</para>
983 <para>The avoided “dangerous” optimisations
984 are those that can make runtime or space
985 <emphasis>worse</emphasis> if you're unlucky. They are
986 normally turned on or off individually.</para>
988 <para>At the moment, <option>-O2</option> is
989 <emphasis>unlikely</emphasis> to produce better code than
990 <option>-O</option>.</para>
995 <term><option>-Ofile <file></option>:</term>
996 <indexterm><primary>-Ofile <file> option</primary></indexterm>
997 <indexterm><primary>optimising, customised</primary></indexterm>
999 <para>(NOTE: not supported yet in GHC 5.x. Please ask if
1000 you're interested in this.)</para>
1002 <para>For those who need <emphasis>absolute</emphasis>
1003 control over <emphasis>exactly</emphasis> what options are
1004 used (e.g., compiler writers, sometimes :-), a list of
1005 options can be put in a file and then slurped in with
1006 <option>-Ofile</option>.</para>
1008 <para>In that file, comments are of the
1009 <literal>#</literal>-to-end-of-line variety; blank
1010 lines and most whitespace is ignored.</para>
1012 <para>Please ask if you are baffled and would like an
1013 example of <option>-Ofile</option>!</para>
1018 <para>We don't use a <option>-O*</option> flag for day-to-day
1019 work. We use <option>-O</option> to get respectable speed;
1020 e.g., when we want to measure something. When we want to go for
1021 broke, we tend to use <option>-O -fvia-C</option> (and we go for
1022 lots of coffee breaks).</para>
1024 <para>The easiest way to see what <option>-O</option> (etc.)
1025 “really mean” is to run with <option>-v</option>,
1026 then stand back in amazement.</para>
1029 <sect2 id="options-f">
1030 <title><option>-f*</option>: platform-independent flags</title>
1032 <indexterm><primary>-f* options (GHC)</primary></indexterm>
1033 <indexterm><primary>-fno-* options (GHC)</primary></indexterm>
1035 <para>These flags turn on and off individual optimisations.
1036 They are normally set via the <option>-O</option> options
1037 described above, and as such, you shouldn't need to set any of
1038 them explicitly (indeed, doing so could lead to unexpected
1039 results). However, there are one or two that may be of
1044 <term><option>-fexcess-precision</option>:</term>
1046 <indexterm><primary><option>-fexcess-precision</option></primary></indexterm>
1047 <para>When this option is given, intermediate floating
1048 point values can have a <emphasis>greater</emphasis>
1049 precision/range than the final type. Generally this is a
1050 good thing, but some programs may rely on the exact
1052 <literal>Float</literal>/<literal>Double</literal> values
1053 and should not use this option for their compilation.</para>
1058 <term><option>-fignore-asserts</option>:</term>
1060 <indexterm><primary><option>-fignore-asserts</option></primary></indexterm>
1061 <para>Causes GHC to ignore uses of the function
1062 <literal>Exception.assert</literal> in source code (in
1063 other words, rewriting <literal>Exception.assert p
1064 e</literal> to <literal>e</literal> (see <xref
1065 linkend="sec-assertions">). This flag is turned on by
1066 <option>-O</option>.
1072 <term><option>-fno-strictness</option></term>
1073 <indexterm><primary><option>-fno-strictness</option></primary>
1076 <para>Turns off the strictness analyser; sometimes it eats
1077 too many cycles.</para>
1082 <term><option>-fno-cpr-analyse</option></term>
1083 <indexterm><primary><option>-fno-cpr-analyse</option></primary>
1086 <para>Turns off the CPR (constructed product result)
1087 analysis; it is somewhat experimental.</para>
1092 <term><option>-funbox-strict-fields</option>:</term>
1094 <indexterm><primary><option>-funbox-strict-fields</option></primary></indexterm>
1095 <indexterm><primary>strict constructor fields</primary></indexterm>
1096 <indexterm><primary>constructor fields, strict</primary></indexterm>
1098 <para>This option causes all constructor fields which are
1099 marked strict (i.e. “!”) to be unboxed or
1100 unpacked if possible. For example:</para>
1103 data T = T !Float !Float
1106 <para>will create a constructor <literal>T</literal>
1107 containing two unboxed floats if the
1108 <option>-funbox-strict-fields</option> flag is given.
1109 This may not always be an optimisation: if the
1110 <Function>T</Function> constructor is scrutinised and the
1111 floats passed to a non-strict function for example, they
1112 will have to be reboxed (this is done automatically by the
1115 <para>This option should only be used in conjunction with
1116 <option>-O</option>, in order to expose unfoldings to the
1117 compiler so the reboxing can be removed as often as
1118 possible. For example:</para>
1122 f (T f1 f2) = f1 + f2
1125 <para>The compiler will avoid reboxing
1126 <Function>f1</Function> and <Function>f2</Function> by
1127 inlining <Function>+</Function> on floats, but only when
1128 <option>-O</option> is on.</para>
1130 <para>Any single-constructor data is eligible for
1131 unpacking; for example</para>
1134 data T = T !(Int,Int)
1137 <para>will store the two <literal>Int</literal>s directly
1138 in the <Function>T</Function> constructor, by flattening
1139 the pair. Multi-level unpacking is also supported:</para>
1143 data S = S !Int !Int
1146 <para>will store two unboxed <literal>Int#</literal>s
1147 directly in the <Function>T</Function> constructor.</para>
1152 <term><option>-funfolding-update-in-place<n></option></term>
1153 <indexterm><primary><option>-funfolding-update-in-place</option></primary></indexterm>
1155 <para>Switches on an experimental "optimisation".
1156 Switching it on makes the compiler a little keener to
1157 inline a function that returns a constructor, if the
1158 context is that of a thunk.
1162 If we inlined plusInt we might get an opportunity to use
1163 update-in-place for the thunk 'x'.</para>
1168 <term><option>-funfolding-creation-threshold<n></option>:</term>
1170 <indexterm><primary><option>-funfolding-creation-threshold</option></primary></indexterm>
1171 <indexterm><primary>inlining, controlling</primary></indexterm>
1172 <indexterm><primary>unfolding, controlling</primary></indexterm>
1174 <para>(Default: 45) Governs the maximum size that GHC will
1175 allow a function unfolding to be. (An unfolding has a
1176 “size” that reflects the cost in terms of
1177 “code bloat” of expanding that unfolding at
1178 at a call site. A bigger function would be assigned a
1179 bigger cost.) </para>
1181 <para> Consequences: (a) nothing larger than this will be
1182 inlined (unless it has an INLINE pragma); (b) nothing
1183 larger than this will be spewed into an interface
1187 <para> Increasing this figure is more likely to result in longer
1188 compile times than faster code. The next option is more
1194 <term><option>-funfolding-use-threshold<n></option>:</term>
1196 <indexterm><primary><option>-funfolding-use-threshold</option></primary></indexterm>
1197 <indexterm><primary>inlining, controlling</primary></indexterm>
1198 <indexterm><primary>unfolding, controlling</primary></indexterm>
1200 <para>(Default: 8) This is the magic cut-off figure for
1201 unfolding: below this size, a function definition will be
1202 unfolded at the call-site, any bigger and it won't. The
1203 size computed for a function depends on two things: the
1204 actual size of the expression minus any discounts that
1205 apply (see <option>-funfolding-con-discount</option>).</para>
1216 <sect1 id="sec-using-concurrent">
1217 <title>Using Concurrent Haskell</title>
1219 <indexterm><primary>Concurrent Haskell—use</primary></indexterm>
1222 GHC supports Concurrent Haskell by default, without requiring a
1223 special option or libraries compiled in a certain way. To get access
1224 to the support libraries for Concurrent Haskell, just import
1225 <literal>Control.Concurrent</literal> (details are in the accompanying
1226 library documentation).</para>
1229 RTS options are provided for modifying the behaviour of the threaded
1230 runtime system. See <XRef LinkEnd="parallel-rts-opts">.
1234 Concurrent Haskell is described in more detail in the documentation
1235 for the <literal>Control.Concurrent</literal> module.
1240 <Sect1 id="sec-using-parallel">
1241 <title>Using Parallel Haskell</title>
1244 <indexterm><primary>Parallel Haskell—use</primary></indexterm>
1248 [You won't be able to execute parallel Haskell programs unless PVM3
1249 (Parallel Virtual Machine, version 3) is installed at your site.]
1253 To compile a Haskell program for parallel execution under PVM, use the
1254 <Option>-parallel</Option> option,<IndexTerm><Primary>-parallel
1255 option</Primary></IndexTerm> both when compiling <Emphasis>and
1256 linking</Emphasis>. You will probably want to <Literal>import
1257 Parallel</Literal> into your Haskell modules.
1261 To run your parallel program, once PVM is going, just invoke it
1262 “as normal”. The main extra RTS option is
1263 <Option>-qp<n></Option>, to say how many PVM
1264 “processors” your program to run on. (For more details of
1265 all relevant RTS options, please see <XRef
1266 LinkEnd="parallel-rts-opts">.)
1270 In truth, running Parallel Haskell programs and getting information
1271 out of them (e.g., parallelism profiles) is a battle with the vagaries of
1272 PVM, detailed in the following sections.
1275 <Sect2 id="pvm-dummies">
1276 <Title>Dummy's guide to using PVM</Title>
1279 <indexterm><primary>PVM, how to use</primary></indexterm>
1280 <indexterm><primary>Parallel Haskell—PVM use</primary></indexterm>
1281 Before you can run a parallel program under PVM, you must set the
1282 required environment variables (PVM's idea, not ours); something like,
1283 probably in your <filename>.cshrc</filename> or equivalent:
1286 setenv PVM_ROOT /wherever/you/put/it
1287 setenv PVM_ARCH `$PVM_ROOT/lib/pvmgetarch`
1288 setenv PVM_DPATH $PVM_ROOT/lib/pvmd
1294 Creating and/or controlling your “parallel machine” is a purely-PVM
1295 business; nothing specific to Parallel Haskell. The following paragraphs
1296 describe how to configure your parallel machine interactively.
1300 If you use parallel Haskell regularly on the same machine configuration it
1301 is a good idea to maintain a file with all machine names and to make the
1302 environment variable PVM_HOST_FILE point to this file. Then you can avoid
1303 the interactive operations described below by just saying
1311 You use the <Command>pvm</Command><IndexTerm><Primary>pvm command</Primary></IndexTerm> command to start PVM on your
1312 machine. You can then do various things to control/monitor your
1313 “parallel machine;” the most useful being:
1319 <ColSpec Align="Left">
1323 <entry><KeyCombo><KeyCap>Control</KeyCap><KeyCap>D</KeyCap></KeyCombo></entry>
1324 <entry>exit <command>pvm</command>, leaving it running</entry>
1328 <entry><command>halt</command></entry>
1329 <entry>kill off this “parallel machine” & exit</entry>
1333 <entry><command>add <host></command></entry>
1334 <entry>add <command><host></command> as a processor</entry>
1338 <entry><command>delete <host></command></entry>
1339 <entry>delete <command><host></command></entry>
1343 <entry><command>reset</command></entry>
1344 <entry>kill what's going, but leave PVM up</entry>
1348 <entry><command>conf</command></entry>
1349 <entry>list the current configuration</entry>
1353 <entry><command>ps</command></entry>
1354 <entry>report processes' status</entry>
1358 <entry><command>pstat <pid></command></entry>
1359 <entry>status of a particular process</entry>
1368 The PVM documentation can tell you much, much more about <command>pvm</command>!
1373 <Sect2 id="par-profiles">
1374 <Title>Parallelism profiles</Title>
1377 <indexterm><primary>parallelism profiles</primary></indexterm>
1378 <indexterm><primary>profiles, parallelism</primary></indexterm>
1379 <indexterm><primary>visualisation tools</primary></indexterm>
1383 With Parallel Haskell programs, we usually don't care about the
1384 results—only with “how parallel” it was! We want pretty pictures.
1388 Parallelism profiles (à la <Command>hbcpp</Command>) can be generated with the
1389 <Option>-qP</Option><IndexTerm><Primary>-qP RTS option (concurrent, parallel)</Primary></IndexTerm> RTS option. The
1390 per-processor profiling info is dumped into files named
1391 <Filename><full-path><program>.gr</Filename>. These are then munged into a PostScript picture,
1392 which you can then display. For example, to run your program
1393 <Filename>a.out</Filename> on 8 processors, then view the parallelism profile, do:
1399 <prompt>$</prompt> ./a.out +RTS -qP -qp8
1400 <prompt>$</prompt> grs2gr *.???.gr > temp.gr # combine the 8 .gr files into one
1401 <prompt>$</prompt> gr2ps -O temp.gr # cvt to .ps; output in temp.ps
1402 <prompt>$</prompt> ghostview -seascape temp.ps # look at it!
1408 The scripts for processing the parallelism profiles are distributed
1409 in <filename>ghc/utils/parallel/</filename>.
1415 <Title>Other useful info about running parallel programs</Title>
1418 The “garbage-collection statistics” RTS options can be useful for
1419 seeing what parallel programs are doing. If you do either
1420 <Option>+RTS -Sstderr</Option><IndexTerm><Primary>-Sstderr RTS option</Primary></IndexTerm> or <Option>+RTS -sstderr</Option>, then
1421 you'll get mutator, garbage-collection, etc., times on standard
1422 error. The standard error of all PE's other than the `main thread'
1423 appears in <filename>/tmp/pvml.nnn</filename>, courtesy of PVM.
1427 Whether doing <option>+RTS -Sstderr</option> or not, a handy way to watch
1428 what's happening overall is: <command>tail -f /tmp/pvml.nnn</command>.
1433 <Sect2 id="parallel-rts-opts">
1434 <title>RTS options for Concurrent/Parallel Haskell
1438 <indexterm><primary>RTS options, concurrent</primary></indexterm>
1439 <indexterm><primary>RTS options, parallel</primary></indexterm>
1440 <indexterm><primary>Concurrent Haskell—RTS options</primary></indexterm>
1441 <indexterm><primary>Parallel Haskell—RTS options</primary></indexterm>
1445 Besides the usual runtime system (RTS) options
1446 (<XRef LinkEnd="runtime-control">), there are a few options particularly
1447 for concurrent/parallel execution.
1454 <Term><Option>-qp<N></Option>:</Term>
1457 <IndexTerm><Primary>-qp<N> RTS option</Primary></IndexTerm>
1458 (PARALLEL ONLY) Use <Literal><N></Literal> PVM processors to run this program;
1464 <term><option>-C[<us>]</option>:</term>
1467 <indexterm><primary>-C<us> RTS option</primary></indexterm> Sets
1468 the context switch interval to <literal><s></literal> seconds.
1469 A context switch will occur at the next heap block allocation after
1470 the timer expires (a heap block allocation occurs every 4k of
1471 allocation). With <option>-C0</option> or <option>-C</option>,
1472 context switches will occur as often as possible (at every heap block
1473 allocation). By default, context switches occur every 20ms
1474 milliseconds. Note that GHC's internal timer ticks every 20ms, and
1475 the context switch timer is always a multiple of this timer, so 20ms
1476 is the maximum granularity available for timed context switches.
1481 <term><option>-q[v]</option>:</term>
1484 <indexterm><primary>-q RTS option</primary></indexterm>
1485 (PARALLEL ONLY) Produce a quasi-parallel profile of thread activity,
1486 in the file <FIlename><program>.qp</FIlename>. In the style of <command>hbcpp</command>, this profile
1487 records the movement of threads between the green (runnable) and red
1488 (blocked) queues. If you specify the verbose suboption (<option>-qv</option>), the
1489 green queue is split into green (for the currently running thread
1490 only) and amber (for other runnable threads). We do not recommend
1491 that you use the verbose suboption if you are planning to use the
1492 <Command>hbcpp</Command> profiling tools or if you are context switching at every heap
1493 check (with <Option>-C</Option>).
1499 <Term><Option>-qt<num></Option>:</Term>
1502 <IndexTerm><Primary>-qt<num> RTS option</Primary></IndexTerm>
1503 (PARALLEL ONLY) Limit the thread pool size, i.e. the number of concurrent
1504 threads per processor to <Literal><num></Literal>. The default is
1505 32. Each thread requires slightly over 1K <Emphasis>words</Emphasis> in
1506 the heap for thread state and stack objects. (For 32-bit machines, this
1507 translates to 4K bytes, and for 64-bit machines, 8K bytes.)
1513 <Term><Option>-d</Option>:</Term>
1516 <IndexTerm><Primary>-d RTS option (parallel)</Primary></IndexTerm>
1517 (PARALLEL ONLY) Turn on debugging. It pops up one xterm (or GDB, or
1518 something…) per PVM processor. We use the standard <Command>debugger</Command>
1519 script that comes with PVM3, but we sometimes meddle with the
1520 <Command>debugger2</Command> script. We include ours in the GHC distribution,
1521 in <Filename>ghc/utils/pvm/</Filename>.
1527 <Term><Option>-qe<num></Option>:</Term>
1530 <IndexTerm><Primary>-qe<num> RTS option
1531 (parallel)</Primary></IndexTerm> (PARALLEL ONLY) Limit the spark pool size
1532 i.e. the number of pending sparks per processor to
1533 <Literal><num></Literal>. The default is 100. A larger number may be
1534 appropriate if your program generates large amounts of parallelism
1540 <Term><Option>-qQ<num></Option>:</Term>
1543 <IndexTerm><Primary>-qQ<num> RTS option (parallel)</Primary></IndexTerm>
1544 (PARALLEL ONLY) Set the size of packets transmitted between processors
1545 to <Literal><num></Literal>. The default is 1024 words. A larger number may be
1546 appropriate if your machine has a high communication cost relative to
1552 <Term><Option>-qh<num></Option>:</Term>
1555 <IndexTerm><Primary>-qh<num> RTS option (parallel)</Primary></IndexTerm>
1556 (PARALLEL ONLY) Select a packing scheme. Set the number of non-root thunks to pack in one packet to
1557 <num>-1 (0 means infinity). By default GUM uses full-subgraph
1558 packing, i.e. the entire subgraph with the requested closure as root is
1559 transmitted (provided it fits into one packet). Choosing a smaller value
1560 reduces the amount of pre-fetching of work done in GUM. This can be
1561 advantageous for improving data locality but it can also worsen the balance
1562 of the load in the system.
1567 <Term><Option>-qg<num></Option>:</Term>
1570 <IndexTerm><Primary>-qg<num> RTS option
1571 (parallel)</Primary></IndexTerm> (PARALLEL ONLY) Select a globalisation
1572 scheme. This option affects the
1573 generation of global addresses when transferring data. Global addresses are
1574 globally unique identifiers required to maintain sharing in the distributed
1575 graph structure. Currently this is a binary option. With <num>=0 full globalisation is used
1576 (default). This means a global address is generated for every closure that
1577 is transmitted. With <num>=1 a thunk-only globalisation scheme is
1578 used, which generated global address only for thunks. The latter case may
1579 lose sharing of data but has a reduced overhead in packing graph structures
1580 and maintaining internal tables of global addresses.
1591 <sect1 id="options-platform">
1592 <title>Platform-specific Flags</title>
1594 <indexterm><primary>-m* options</primary></indexterm>
1595 <indexterm><primary>platform-specific options</primary></indexterm>
1596 <indexterm><primary>machine-specific options</primary></indexterm>
1598 <para>Some flags only make sense for particular target
1604 <term><option>-mv8</option>:</term>
1606 <para>(SPARC machines)<indexterm><primary>-mv8 option (SPARC
1607 only)</primary></indexterm> Means to pass the like-named
1608 option to GCC; it says to use the Version 8 SPARC
1609 instructions, notably integer multiply and divide. The
1610 similiar <option>-m*</option> GCC options for SPARC also
1611 work, actually.</para>
1616 <term><option>-monly-[32]-regs</option>:</term>
1618 <para>(iX86 machines)<indexterm><primary>-monly-N-regs
1619 option (iX86 only)</primary></indexterm> GHC tries to
1620 “steal” four registers from GCC, for performance
1621 reasons; it almost always works. However, when GCC is
1622 compiling some modules with four stolen registers, it will
1623 crash, probably saying:
1626 Foo.hc:533: fixed or forbidden register was spilled.
1627 This may be due to a compiler bug or to impossible asm
1628 statements or clauses.
1631 Just give some registers back with
1632 <option>-monly-N-regs</option>. Try `3' first, then `2'.
1633 If `2' doesn't work, please report the bug to us.</para>
1642 <sect1 id="ext-core">
1643 <title>Generating and compiling External Core Files</title>
1645 <indexterm><primary>intermediate code generation</primary></indexterm>
1647 <para>GHC can dump its optimized intermediate code (said to be in “Core” format)
1648 to a file as a side-effect of compilation. Core files, which are given the suffix
1649 <filename>.hcr</filename>, can be read and processed by non-GHC back-end
1650 tools. The Core format is formally described in <ulink url="http://www.haskell.org/ghc/docs/papers/core.ps.gz"
1651 <citetitle>An External Representation for the GHC Core Language</citetitle></ulink>,
1652 and sample tools (in Haskell)
1653 for manipulating Core files are available in the GHC source distribution
1654 directory <literal>/fptools/ghc/utils/ext-core</literal>.
1655 Note that the format of <literal>.hcr</literal>
1656 files is <emphasis>different</emphasis> (though similar) to the Core output format generated
1657 for debugging purposes (<xref linkend="options-debugging">).</para>
1659 <para>The Core format natively supports notes which you can add to
1660 your source code using the <literal>CORE</literal> pragma (see <xref
1661 linkend="pragmas">).</para>
1666 <term><option>-fext-core</option></term>
1668 <primary><option>-fext-core</option></primary>
1671 <para>Generate <literal>.hcr</literal> files.</para>
1677 <para>GHC can also read in External Core files as source; just give the <literal>.hcr</literal> file on
1678 the command line, instead of the <literal>.hs</literal> or <literal>.lhs</literal> Haskell source.
1679 A current infelicity is that you need to give teh <literal>-fglasgow-exts</literal> flag too, because
1680 ordinary Haskell 98, when translated to External Core, uses things like rank-2 types.</para>
1689 ;;; Local Variables: ***
1691 ;;; sgml-parent-document: ("users_guide.sgml" "book" "chapter") ***