4 <nidx>GHC, using</nidx>
7 GHC is a command-line compiler: in order to compile a Haskell program,
8 GHC must be invoked on the source file(s) by typing a command to the
9 shell. The steps involved in compiling a program can be automated
10 using the @make@ tool (this is especially useful if the program
11 consists of multiple source files which depend on each other). This
12 section describes how to use GHC from the command-line.
14 %************************************************************************
16 <sect1> Overall command-line structure
17 <label id="command-line-structure">
19 <nidx>structure, command-line</nidx>
20 <nidx>command-line structure</nidx>
22 %************************************************************************
24 An invocation of GHC takes the following form:
30 Command-line arguments are either options or file names.
32 Command-line options begin with @-@. They may <em>not</em> be
33 grouped: @-vO@ is different from @-v -O@. Options need not
34 precede filenames: e.g., @ghc *.o -o foo@. All options are
35 processed and then applied to all files; you cannot, for example, invoke
36 @ghc -c -O1 Foo.hs -O2 Bar.hs@ to apply different optimisation
37 levels to the files @Foo.hs@ and @Bar.hs@. For conflicting
38 options, e.g., @-c -S@, we reserve the right to do anything we
39 want. (Usually, the last one applies.)
41 %************************************************************************
43 <sect1>Meaningful file suffixes
44 <label id="file-suffixes">
46 <nidx>suffixes, file</nidx>
47 <nidx>file suffixes for GHC</nidx>
49 %************************************************************************
51 File names with ``meaningful'' suffixes (e.g., @.lhs@ or @.o@)
52 cause the ``right thing'' to happen to those files.
56 <nidx>lhs suffix</nidx>
57 A ``literate Haskell'' module.
60 A not-so-literate Haskell module.
63 A Haskell interface file, probably compiler-generated.
66 Intermediate C file produced by the Haskell compiler.
69 A C~file not produced by the Haskell compiler.
72 % C code after it has be preprocessed by the C compiler (using the
76 An assembly-language source file, usually
77 produced by the compiler.
80 An object file, produced by an assembler.
83 Files with other suffixes (or without suffixes) are passed straight
86 %************************************************************************
88 <sect1>Help and verbosity options
89 <label id="options-help">
91 <nidx>help options (GHC)</nidx>
92 <nidx>verbose option (GHC)</nidx>
94 %************************************************************************
96 A good option to start with is the @-help@ (or @-?@) option.
97 <nidx>-help option</nidx>
98 <nidx>-? option</nidx>
99 GHC spews a long message to standard output and then exits.
101 The @-v@<nidx>-v option</nidx> option makes GHC <em>verbose</em>: it
102 reports its version number and shows (on stderr) exactly how it invokes each
103 phase of the compilation system. Moreover, it passes
104 the @-v@ flag to most phases; each reports
105 its version number (and possibly some other information).
107 Please, oh please, use the @-v@ option when reporting bugs!
108 Knowing that you ran the right bits in the right order is always the
109 first thing we want to verify.
111 If you're just interested in the compiler version number, the
112 @--version@<nidx>--version option</nidx> option prints out a
113 one-line string containing the requested info.
115 %************************************************************************
117 <sect1>Running the right phases in the right order
118 <label id="options-order">
120 <nidx>order of passes in GHC</nidx>
121 <nidx>pass ordering in GHC</nidx>
123 %************************************************************************
125 The basic task of the @ghc@ driver is to run each input file
126 through the right phases (parsing, linking, etc.).
128 The first phase to run is determined by the input-file suffix, and the
129 last phase is determined by a flag. If no relevant flag is present,
130 then go all the way through linking. This table summarises:
133 Phase of the | Suffix saying | Flag saying | (suffix of) @@
134 compilation system | ``start here''| ``stop after''| output file @@
136 literate pre-processor | .lhs | - | - @@
137 C pre-processor (opt.) | - | - | - @@
138 Haskell compiler | .hs | -C, -S | .hc, .s @@
139 C compiler (opt.) | .hc or .c | -S | .s @@
140 assembler | .s | -c | .o @@
141 linker | other | - | a.out @@
143 <nidx>-C option</nidx>
144 <nidx>-S option</nidx>
145 <nidx>-c option</nidx>
147 Thus, a common invocation would be: @ghc -c Foo.hs@
149 Note: What the Haskell compiler proper produces depends on whether a
150 native-code generator is used (producing assembly language) or not
153 The option @-cpp@<nidx>-cpp option</nidx> must be given for the C
154 pre-processor phase to be run, that is, the pre-processor will be run
155 over your Haskell source file before continuing.
157 The option @-E@<nidx>-E option</nidx> runs just the pre-processing
158 passes of the compiler, outputting the result on stdout before
159 stopping. If used in conjunction with -cpp, the output is the
160 code blocks of the original (literal) source after having put it
161 through the grinder that is the C pre-processor. Sans @-cpp@, the
162 output is the de-litted version of the original source.
164 The option @-optcpp-E@<nidx>-optcpp-E option</nidx> runs just the
165 pre-processing stage of the C-compiling phase, sending the result to
166 stdout. (For debugging or obfuscation contests, usually.)
168 %************************************************************************
170 <sect1>Re-directing the compilation output(s)
171 <label id="options-output">
173 <nidx>output-directing options</nidx>
175 %************************************************************************
177 GHC's compiled output normally goes into a @.hc@, @.o@, etc., file,
178 depending on the last-run compilation phase. The option @-o
179 foo@<nidx>-o option</nidx> re-directs the output of that last-run
182 Note: this ``feature'' can be counterintuitive:
183 @ghc -C -o foo.o foo.hs@ will put the intermediate C code in the
184 file @foo.o@, name notwithstanding!
186 EXOTICA: But the @-o@ option isn't of much use if you have
187 <em>several</em> input files... Non-interface output files are
188 normally put in the same directory as their corresponding input file
189 came from. You may specify that they be put in another directory
190 using the @-odir <dir>@<nidx>-odir <dir> option</nidx> (the
191 ``Oh, dear'' option). For example:
194 % ghc -c parse/Foo.hs parse/Bar.hs gurgle/Bumble.hs -odir `arch`
197 The output files, @Foo.o@, @Bar.o@, and @Bumble.o@ would be
198 put into a subdirectory named after the architecture of the executing
199 machine (@sun4@, @mips@, etc). The directory must already
200 exist; it won't be created.
202 Note that the @-odir@ option does <em>not</em> affect where the
203 interface files are put. In the above example, they would still be
204 put in @parse/Foo.hi@, @parse/Bar.hi@, and @gurgle/Bumble.hi@.
206 MORE EXOTICA: The @-osuf <suffix>@<nidx>-osuf <suffix>
207 option</nidx> will change the @.o@ file suffix for object files to
208 whatever you specify. (We use this in compiling the prelude.).
209 Similarly, the @-hisuf <suffix>@<nidx>-hisuf <suffix>
210 option</nidx> will change the @.hi@ file suffix for non-system
211 interface files (see Section <ref name="Other options related to
212 interface files" id="hi-options">).
214 The @-hisuf@/@-osuf@ game is useful if you want to compile a program
215 with both GHC and HBC (say) in the same directory. Let HBC use the
216 standard @.hi@/@.o@ suffixes; add @-hisuf g_hi -osuf g_o@ to your
217 @make@ rule for GHC compiling...
219 FURTHER EXOTICA: If you are doing a normal @.hs@-to-@.o@ compilation
220 but would like to hang onto the intermediate @.hc@ C file, just
221 throw in a @-keep-hc-file-too@ option<nidx>-keep-hc-file-too option</nidx>.
222 If you would like to look at the assembler output, toss in a
223 @-keep-s-file-too@,<nidx>-keep-s-file-too option</nidx> too.
225 <sect2> Saving GHC's standard error output
226 <label id="saving-ghc-stderr">
228 <nidx>standard error, saving</nidx>
230 Sometimes, you may cause GHC to be rather chatty on standard error;
231 with @-dshow-rn-trace@, for example. You can instruct GHC to
232 <em>append</em> this output to a particular log file with a @-odump
233 <blah>@<nidx>-odump <blah> option</nidx> option.
235 <sect2> Redirecting temporary files
236 <label id="temp-files">
238 <nidx>temporary files, redirecting</nidx>
240 If you have trouble because of running out of space in @/tmp@ (or
241 wherever your installation thinks temporary files should go), you may
242 use the @-tmpdir <dir>@<nidx>-tmpdir <dir> option</nidx> option
243 to specify an alternate directory. For example, @-tmpdir .@ says to
244 put temporary files in the current working directory.
246 Alternatively, use your @TMPDIR@ environment variable.<nidx>TMPDIR
247 environment variable</nidx> Set it to the name of the directory where
248 temporary files should be put. GCC and other programs will honour the
249 @TMPDIR@ variable as well.
251 Even better idea: Set the @TMPDIR@ variable when building GHC, and
252 never worry about @TMPDIR@ again. (see the build documentation).
254 %************************************************************************
256 <sect1>Warnings and sanity-checking
257 <label id="options-sanity">
259 <nidx>sanity-checking options</nidx>
260 <nidx>warnings</nidx>
262 %************************************************************************
264 GHC has a number of options that select which types of non-fatal error
265 messages, otherwise known as warnings, can be generated during
266 compilation. By default, you get a standard set of warnings which are
267 generally likely to indicate bugs in your program. These are:
268 @-fwarn-overlapping-patterns@, @-fwarn-duplicate-exports@, and
269 @-fwarn-missing-methods@. The following flags are simple ways to
270 select standard ``packages'' of warnings:
275 <nidx>-Wnot option</nidx>
277 Turns off all warnings, including the standard ones.
280 <nidx>-w option</nidx>
285 <nidx>-W option</nidx>
287 Provides the standard warnings plus @-fwarn-incomplete-patterns@,
288 @-fwarn-unused-imports@ and @-fwarn-unused-binds@.
291 <nidx>-Wall option</nidx>
293 Turns on all warning options.
297 The full set of warning options is described below. To turn off any
298 warning, simply give the corresponding @-fno-warn-...@ option on
303 <tag>@-fwarn-name-shadowing@:</tag>
304 <nidx>-fwarn-name-shadowing option</nidx>
305 <nidx>shadowing, warning</nidx>
307 This option causes a warning to be emitted whenever an inner-scope
308 value has the same name as an outer-scope value, i.e. the inner value
309 shadows the outer one. This can catch typographical errors that turn
310 into hard-to-find bugs, e.g., in the inadvertent cyclic definition
311 @let x = ... x ... in@.
313 Consequently, this option does <em>not</em> allow cyclic recursive
316 <tag>@-fwarn-hi-shadowing@:</tag>
317 <nidx>-fwarn-hi-shadowing option</nidx>
318 <nidx>interface files, shadowing</nidx>
320 Warns you about shadowing of interface files along the supplied import path.
321 For instance, assuming you invoke @ghc@ with the import path
322 @-iutils:src@ and @Utils.hi@ exist in both the @utils@ and @src@
323 directories, @-fwarn-hi-shadowing@ will warn you that @utils/Utils.hi@
324 shadows @src/Utils.hi@.
326 <tag>@-fwarn-overlapping-patterns@:</tag>
327 <nidx>-fwarn-overlapping-patterns option</nidx>
328 <nidx>overlapping patterns, warning</nidx>
329 <nidx>patterns, overlapping</nidx>
331 By default, the compiler will warn you if a set of patterns are either
332 incomplete (i.e., you're only matching on a subset of an algebraic
333 data type's constructors), or overlapping, i.e.,
344 where the last pattern match in @f@ won't ever be reached, as the
345 second pattern overlaps it. More often than not, redundant patterns
346 is a programmer mistake/error, so this option is enabled by default.
348 <tag>@-fwarn-incomplete-patterns@:</tag>
349 <nidx>-fwarn-incomplete-patterns option</nidx>
350 <nidx>incomplete patterns, warning</nidx>
351 <nidx>patterns, incomplete</nidx>
353 Similarly for incomplete patterns, the function @g@ will fail when
354 applied to non-empty lists, so the compiler will emit a warning about
355 this when this option is enabled.
357 <tag>@-fwarn-missing-methods@:</tag>
358 <nidx>-fwarn-missing-methods option</nidx>
359 <nidx>missing methods, warning</nidx>
360 <nidx>methods, missing</nidx>
362 This option is on by default, and warns you whenever an instance
363 declaration is missing one or more methods, and the corresponding
364 class declaration has no default declaration for them.
366 <tag>@-fwarn-unused-imports@:</tag>
367 <nidx>-fwarn-unused-imports option</nidx>
368 <nidx>unused imports, warning</nidx>
369 <nidx>imports, unused</nidx>
371 Report any objects that are explicitly imported but never used.
373 <tag>@-fwarn-unused-binds@:</tag>
374 <nidx>-fwarn-unused-binds option</nidx>
375 <nidx>unused binds, warning</nidx>
376 <nidx>binds, unused</nidx>
378 Report any function definitions (and local bindings) which are unused.
379 For top-level functions, the warning is only given if the binding is
382 <tag>@-fwarn-unused-matches@:</tag>
383 <nidx>-fwarn-unused-matches option</nidx>
384 <nidx>unused matches, warning</nidx>
385 <nidx>matches, unused</nidx>
387 Report all unused variables which arise from pattern matches,
388 including patterns consisting of a single variable. For instance @f x
389 y = []@ would report @x@ and @y@ as unused. To eliminate the warning,
390 all unused variables can be replaced with wildcards.
392 <tag>@-fwarn-duplicate-exports@:</tag>
393 <nidx>-fwarn-duplicate-exports option</nidx>
394 <nidx>duplicate exports, warning</nidx>
395 <nidx>export lists, duplicates</nidx>
397 Have the compiler warn about duplicate entries in export lists. This
398 is useful information if you maintain large export lists, and want to
399 avoid the continued export of a definition after you've deleted (one)
400 mention of it in the export list.
402 This option is on by default.
406 If you would like GHC to check that every top-level value has a type
407 signature, use the @-fsignatures-required@
408 option.<nidx>-fsignatures-required option</nidx>
410 If you're feeling really paranoid, the @-dcore-lint@
411 option<nidx>-dcore-lint option</nidx> is a good choice. It turns on
412 heavyweight intra-pass sanity-checking within GHC. (It checks GHC's
415 %************************************************************************
417 <sect1>Separate compilation
418 <label id="separate-compilation">
420 <nidx>separate compilation</nidx>
421 <nidx>recompilation checker</nidx>
422 <nidx>make and recompilation</nidx>
424 %************************************************************************
426 This section describes how GHC supports separate compilation.
428 <sect2>Interface files
429 <label id="hi-files">
431 <nidx>interface files</nidx>
432 <nidx>.hi files</nidx>
434 When GHC compiles a source file @F@ which contains a module @A@, say,
435 it generates an object @F.o@, <em>and</em> a companion <em>interface
438 NOTE: Having the name of the interface file follow the module name and
439 not the file name, means that working with tools such as @make(1)@
440 become harder. @make@ implicitly assumes that any output files
441 produced by processing a translation unit will have file names that
442 can be derived from the file name of the translation unit. For
443 instance, pattern rules becomes unusable. For this reason, we
444 recommend you stick to using the same file name as the module name.
446 The interface file for @A@ contains information needed by the compiler
447 when it compiles any module @B@ that imports @A@, whether directly or
448 indirectly. When compiling @B@, GHC will read @A.hi@ to find the
449 details that it needs to know about things defined in @A@.
451 Furthermore, when compiling module @C@ which imports @B@, GHC may
452 decide that it needs to know something about @A@ --- for example, @B@
453 might export a function that involves a type defined in @A@. In this
454 case, GHC will go and read @A.hi@ even though @C@ does not explicitly
457 The interface file may contain all sorts of things that aren't
458 explicitly exported from @A@ by the programmer. For example, even
459 though a data type is exported abstractly, @A.hi@ will contain the
460 full data type definition. For small function definitions, @A.hi@
461 will contain the complete definition of the function. For bigger
462 functions, @A.hi@ will contain strictness information about the
463 function. And so on. GHC puts much more information into @.hi@ files
464 when optimisation is turned on with the @-O@ flag. Without @-O@ it
465 puts in just the minimum; with @-O@ it lobs in a whole pile of stuff.
466 <nidx>optimsation, effect on .hi files</nidx>
468 @A.hi@ should really be thought of as a compiler-readable version of
469 @A.o@. If you use a @.hi@ file that wasn't generated by the same
470 compilation run that generates the @.o@ file the compiler may assume
471 all sorts of incorrect things about @A@, resulting in core dumps and
472 other unpleasant happenings.
474 %************************************************************************
476 <sect2>Finding interface files
477 <label id="options-finding-imports">
479 <nidx>interface files, finding them</nidx>
480 <nidx>finding interface files</nidx>
482 %************************************************************************
484 In your program, you import a module @Foo@ by saying
485 @import Foo@. GHC goes looking for an interface file, @Foo.hi@.
486 It has a builtin list of directories (notably including @.@) where
491 <tag>@-i<dirs>@</tag><nidx>-i<dirs> option</nidx> This flag
492 prepends a colon-separated list of @dirs@ to the ``import
495 <tag>@-i@</tag> resets the ``import directories'' list back to nothing.
497 <tag>@-fno-implicit-prelude@</tag>
498 <nidx>-fno-implicit-prelude option</nidx>
499 GHC normally imports @Prelude.hi@ files for you. If you'd rather it
500 didn't, then give it a @-fno-implicit-prelude@ option. You are
501 unlikely to get very far without a Prelude, but, hey, it's a free
504 <tag>@-syslib <lib>@</tag>
505 <nidx>-syslib <lib> option</nidx>
507 If you are using a system-supplied non-Prelude library (e.g., the
508 POSIX library), just use a @-syslib posix@ option (for example). The
509 right interface files should then be available. Section <ref
510 name="The GHC Prelude and Libraries" id="ghc-prelude"> lists the
511 libraries available by this mechanism.
514 <nidx>-I<dir> option</nidx>
516 Once a Haskell module has been compiled to C (@.hc@ file), you may
517 wish to specify where GHC tells the C compiler to look for @.h@ files.
518 (Or, if you are using the @-cpp@ option<nidx>-cpp option</nidx>, where
519 it tells the C pre-processor to look...) For this purpose, use a @-I@
520 option in the usual C-ish way.
524 %************************************************************************
526 <sect2>Other options related to interface files
527 <label id="hi-options">
529 <nidx>interface files, options</nidx>
531 %************************************************************************
533 The interface output may be directed to another file
534 @bar2/Wurble.iface@ with the option @-ohi bar2/Wurble.iface@<nidx>-ohi
535 <file> option</nidx> (not recommended).
537 To avoid generating an interface file at all, use a @-nohi@
538 option.<nidx>-nohi option</nidx>
540 The compiler does not overwrite an existing @.hi@ interface file if
541 the new one is byte-for-byte the same as the old one; this is friendly
542 to @make@. When an interface does change, it is often enlightening to
543 be informed. The @-hi-diffs@<nidx>-hi-diffs option</nidx> option will
544 make @ghc@ run @diff@ on the old and new @.hi@ files. You can also
545 record the difference in the interface file itself, the
546 @-keep-hi-diffs@<nidx>-keep-hi-diffs</nidx> option takes care of that.
548 The @.hi@ files from GHC contain ``usage'' information which changes
549 often and uninterestingly. If you really want to see these changes
550 reported, you need to use the
551 @-hi-diffs-with-usages@<nidx>-hi-diffs-with-usages option</nidx>
554 Interface files are normally jammed full of compiler-produced
555 <em>pragmas</em>, which record arities, strictness info, etc. If you
556 think these pragmas are messing you up (or you are doing some kind of
557 weird experiment), you can tell GHC to ignore them with the
558 @-fignore-interface-pragmas@<nidx>-fignore-interface-pragmas
559 option</nidx> option.
561 When compiling without optimisations on, the compiler is extra-careful
562 about not slurping in data constructors and instance declarations that
563 it will not need. If you believe it is getting it wrong and not
564 importing stuff which you think it should, this optimisation can be
565 turned off with @-fno-prune-tydecls@ and @-fno-prune-instdecls@.
566 <nidx>-fno-prune-tydecls option</nidx><nidx>-fno-prune-instdecls
569 See also Section <ref name="Linking and consistency-checking"
570 id="options-linker">, which describes how the linker finds standard
573 %************************************************************************
575 <sect2>The recompilation checker
578 <nidx>recompilation checker</nidx>
580 %************************************************************************
582 In the olden days, GHC compared the newly-generated @.hi@ file with
583 the previous version; if they were identical, it left the old one
584 alone and didn't change its modification date. In consequence,
585 importers of a module with an unchanged output @.hi@ file were not
588 This doesn't work any more. In our earlier example, module @C@ does
589 not import module @A@ directly, yet changes to @A.hi@ should force a
590 recompilation of @C@. And some changes to @A@ (changing the
591 definition of a function that appears in an inlining of a function
592 exported by @B@, say) may conceivably not change @B.hi@ one jot. So
595 GHC keeps a version number on each interface file, and on each type
596 signature within the interface file. It also keeps in every interface
597 file a list of the version numbers of everything it used when it last
598 compiled the file. If the source file's modification date is earlier
599 than the @.o@ file's date (i.e. the source hasn't changed since the
600 file was last compiled), and you give GHC the @-recomp@<nidx>-recomp
601 option</nidx> flag, then GHC will be clever. It compares the version
602 numbers on the things it needs this time with the version numbers on
603 the things it needed last time (gleaned from the interface file of the
604 module being compiled); if they are all the same it stops compiling
605 rather early in the process saying ``Compilation IS NOT required''.
606 What a beautiful sight!
608 It's still an experimental feature (that's why @-recomp@ is off by
609 default), so tell us if you think it doesn't work.
611 Patrick Sansom has a workshop paper about how all this is done. Ask
612 him (email: <htmlurl name="sansom@@dcs.gla.ac.uk"
613 url="mailto:sansom@@dcs.gla.ac.uk">) if you want a copy.
615 %************************************************************************
618 <label id="using-make">
622 %************************************************************************
624 It is reasonably straightforward to set up a @Makefile@ to use with
625 GHC, assuming you name your source files the same as your modules.
630 HC_OPTS = -cpp $(EXTRA_HC_OPTS)
632 SRCS = Main.lhs Foo.lhs Bar.lhs
633 OBJS = Main.o Foo.o Bar.o
635 .SUFFIXES : .o .hi .lhs .hc .s
639 $(HC) -o $@ $(HC_OPTS) $(OBJS)
641 # Standard suffix rules
646 $(HC) -c $< $(HC_OPTS)
649 $(HC) -c $< $(HC_OPTS)
651 # Inter-module dependencies
652 Foo.o Foo.hc Foo.s : Baz.hi # Foo imports Baz
653 Main.o Main.hc Main.s : Foo.hi Baz.hi # Main imports Foo and Baz
656 (Sophisticated @make@ variants may achieve some of the above more
657 elegantly. Notably, @gmake@'s pattern rules let you write the more
662 $(HC) -c $< $(HC_OPTS)
665 What we've shown should work with any @make@.)
667 Note the cheesy @.o.hi@ rule: It records the dependency of the
668 interface (@.hi@) file on the source. The rule says a @.hi@ file can
669 be made from a @.o@ file by doing... nothing. Which is true.
671 Note the inter-module dependencies at the end of the Makefile, which
675 Foo.o Foo.hc Foo.s : Baz.hi # Foo imports Baz
678 They tell @make@ that if any of @Foo.o@, @Foo.hc@ or @Foo.s@ have an
679 earlier modification date than @Baz.hi@, then the out-of-date file
680 must be brought up to date. To bring it up to date, @make@ looks for
681 a rule to do so; one of the preceding suffix rules does the job
684 Putting inter-dependencies of the form @Foo.o : Bar.hi@ into your
685 @Makefile@ by hand is rather error-prone. @ghc@ offers you a helping
686 hand with it's @-M@ option. To automatically generate
687 inter-dependencies, add the following to your @Makefile@:
691 $(HC) -M $(HC_OPTS) $(SRCS)
694 Now, before you start compiling, and any time you change the @imports@
695 in your program, do @make depend@ before you do @make cool_pgm@.
696 @ghc -M@ will then append the needed dependencies to your @Makefile@.
698 The dependencies are actually generated by another utility,
699 @mkdependHS@, which @ghc -M@ just calls upon. @mkdependHS@ is
700 distributed with GHC and is documented in Section <ref name="Makefile
701 dependencies in Haskell: using mkdependHS" id="mkdependHS">.
703 A few caveats about this simple scheme:
707 <item> You may need to compile some modules explicitly to create their
708 interfaces in the first place (e.g., @make Bar.o@ to create @Bar.hi@).
710 <item> You may have to type @make@ more than once for the dependencies
711 to have full effect. However, a @make@ run that does nothing
712 <em>does</em> mean ``everything's up-to-date.''
714 <item> This scheme will work with mutually-recursive modules but,
715 again, it may take multiple iterations to ``settle.''
719 %************************************************************************
721 <sect2>How to compile mutually recursive modules
722 <label id="mutual-recursion">
724 <nidx>module system, recursion</nidx>
725 <nidx>recursion, between modules</nidx>
727 %************************************************************************
729 Currently, the compiler does not have proper support for dealing with
730 mutually recursive modules:
752 When compiling either module A and B, the compiler will try (in vain)
753 to look for the interface file of the other. So, to get mutually
754 recursive modules off the ground, you need to hand write an interface
755 file for A or B, so as to break the loop. These hand-written
756 interface files are called @hi-boot@ files, and are placed in a file
757 called @<module>.hi-boot@. To import from an @hi-boot@ file instead
758 of the standard @.hi@ file, use the following syntax in the importing module:
759 <nidx>hi-boot files</nidx>
760 <nidx>importing, hi-boot files</nidx>
763 import {-# SOURCE #-} A
766 The hand-written interface need only contain the bare minimum of
767 information needed to get the bootstrapping process started. For
768 example, it doesn't need to contain declarations for <em/everything/
769 that module @A@ exports, only the things required by the module that
770 imports @A@ recursively.
772 For the example at hand, the boot interface file for A would like the
780 1 newtype A = A PrelBase.Int ;
783 The syntax is essentially the same as a normal @.hi@ file
784 (unfortunately), but you can usually tailor an existing @.hi@ file to
785 make a @.hi-boot@ file.
787 Notice that we only put the declaration for the newtype @A@ in the
788 @hi-boot@ file, not the signature for @f@, since @f@ isn't used by
791 The number ``1'' at the beginning of a declaration is the <em>version
792 number</em> of that declaration: for the purposes of @.hi-boot@ files
793 these can all be set to 1. All names must be fully qualified with the
794 <em/original/ module that an object comes from: for example, the
795 reference to @Int@ in the interface for @A@ comes from @PrelBase@,
796 which is a module internal to GHC's prelude. It's a pain, but that's
799 <bf>Note:</bf> This is all a temporary solution, a version of the
800 compiler that handles mutually recursive properly without the manual
801 construction of interface file, is in the works.
803 %************************************************************************
805 <sect1>Optimisation (code improvement)
806 <label id="options-optimise">
808 <nidx>optimisation (GHC)</nidx>
809 <nidx>improvement, code (GHC)</nidx>
811 %************************************************************************
813 The @-O*@ options specify convenient ``packages'' of optimisation
814 flags; the @-f*@ options described later on specify
815 <em>individual</em> optimisations to be turned on/off; the @-m*@
816 options specify <em>machine-specific</em> optimisations to be turned
819 %----------------------------------------------------------------------
820 <sect2>@-O*@: convenient ``packages'' of optimisation flags.
821 <label id="optimise-pkgs">
823 <nidx>-O options</nidx>
825 There are <em>many</em> options that affect the quality of code
826 produced by GHC. Most people only have a general goal, something like
827 ``Compile quickly'' or ``Make my program run like greased lightning.''
828 The following ``packages'' of optimisations (or lack thereof) should
831 Once you choose a @-O*@ ``package,'' stick with it---don't chop and
832 change. Modules' interfaces <em>will</em> change with a shift to a new
833 @-O*@ option, and you may have to recompile a large chunk of all
834 importing modules before your program can again be run
835 safely (see Section <ref name="The recompilation checker" id="recomp">).
838 <tag>No @-O*@-type option specified:</tag>
839 <nidx>-O* not specified</nidx>
840 This is taken to mean: ``Please compile quickly; I'm not over-bothered
841 about compiled-code quality.'' So, for example: @ghc -c Foo.hs@
843 <tag>@-O@ or @-O1@:</tag>
844 <nidx>-O option</nidx>
845 <nidx>-O1 option</nidx>
846 <nidx>optimise normally</nidx>
847 Means: ``Generate good-quality code without taking too long about it.''
848 Thus, for example: @ghc -c -O Main.lhs@
851 <nidx>-O2 option</nidx>
852 <nidx>optimise aggressively</nidx>
853 Means: ``Apply every non-dangerous optimisation, even if it means
854 significantly longer compile times.''
856 The avoided ``dangerous'' optimisations are those that can make
857 runtime or space <em>worse</em> if you're unlucky. They are
858 normally turned on or off individually.
860 At the moment, @-O2@ is <em>unlikely</em> to produce
861 better code than @-O@.
863 <tag>@-O2-for-C@:</tag>
864 <nidx>-O2-for-C option</nidx>
865 <nidx>gcc, invoking with -O2</nidx>
867 Says to run GCC with @-O2@, which may be worth a few percent in
868 execution speed. Don't forget @-fvia-C@, lest you use the native-code
869 generator and bypass GCC altogether!
872 <nidx>-Onot option</nidx>
873 <nidx>optimising, reset</nidx>
875 This option will make GHC ``forget'' any -Oish options it has seen so
876 far. Sometimes useful; for example: @make all EXTRA_HC_OPTS=-Onot@.
878 <tag>@-Ofile <file>@:</tag>
879 <nidx>-Ofile <file> option</nidx>
880 <nidx>optimising, customised</nidx>
882 For those who need <em>absolute</em> control over <em>exactly</em>
883 what options are used (e.g., compiler writers, sometimes :-), a list
884 of options can be put in a file and then slurped in with @-Ofile@.
886 In that file, comments are of the @#@-to-end-of-line variety; blank
887 lines and most whitespace is ignored.
889 Please ask if you are baffled and would like an example of @-Ofile@!
892 At Glasgow, we don't use a @-O*@ flag for day-to-day work. We use
893 @-O@ to get respectable speed; e.g., when we want to measure
894 something. When we want to go for broke, we tend to use @-O -fvia-C
895 -O2-for-C@ (and we go for lots of coffee breaks).
897 The easiest way to see what @-O@ (etc) ``really mean'' is to run with
898 @-v@, then stand back in amazement. Alternatively, just look at the
899 @HsC_minus<blah>@ lists in the @ghc@ driver script.
901 %----------------------------------------------------------------------
902 <sect2>@-f*@: platform-independent flags
904 <nidx>-f* options (GHC)</nidx>
905 <nidx>-fno-* options (GHC)</nidx>
907 Flags can be turned <em>off</em> individually. (NB: I hope you have a
908 good reason for doing this....) To turn off the @-ffoo@ flag, just use
909 the @-fno-foo@ flag.<nidx>-fno-<opt> anti-option</nidx> So, for
910 example, you can say @-O2 -fno-strictness@, which will then drop out
911 any running of the strictness analyser.
913 The options you are most likely to want to turn off are:
914 @-fno-strictness@<nidx>-fno-strictness option</nidx> (strictness
915 analyser [because it is sometimes slow]),
916 @-fno-specialise@<nidx>-fno-specialise option</nidx> (automatic
917 specialisation of overloaded functions [because it makes your code
918 bigger]) [US spelling also accepted], and
919 @-fno-update-analyser@<nidx>-fno-update-analyser option</nidx>
920 (update analyser, because it sometimes takes a <em>long</em> time).
922 Should you wish to turn individual flags <em>on</em>, you are advised
923 to use the @-Ofile@ option, described above. Because the order in
924 which optimisation passes are run is sometimes crucial, it's quite
925 hard to do with command-line options.
927 Here are some ``dangerous'' optimisations you <em>might</em> want to try:
929 %------------------------------------------------------------------
930 <tag>@-fvia-C@:</tag>
931 <nidx>-fvia-C option</nidx>
932 <nidx>native code generator, turning off</nidx>
934 Compile via C, and don't use the native-code generator. (There are
935 many cases when GHC does this on its own.) You might pick up a little
936 bit of speed by compiling via C. If you use @_ccall_gc_@s or
937 @_casm_@s, you probably <em>have to</em> use @-fvia-C@.
939 The lower-case incantation, @-fvia-c@, is synonymous.
941 <tag>@-funfolding-creation-threshold<n>@:</tag>
942 <nidx>-funfolding-creation-threshold option</nidx>
943 <nidx>inlining, controlling</nidx>
944 <nidx>unfolding, controlling</nidx>
945 (Default: 30) By raising or lowering this number, you can raise or
946 lower the amount of pragmatic junk that gets spewed into interface
947 files. (An unfolding has a ``size'' that reflects the cost in terms
948 of ``code bloat'' of expanding that unfolding in another module. A
949 bigger Core expression would be assigned a bigger cost.)
951 <tag>@-funfolding-use-threshold<n>@:</tag>
952 <nidx>-funfolding-use-threshold option</nidx>
953 <nidx>inlining, controlling</nidx>
954 <nidx>unfolding, controlling</nidx>
955 (Default: 3) By raising or lowering this number, you can make the
956 compiler more or less keen to expand unfoldings.
958 OK, folks, these magic numbers `30' and `3' are mildly arbitrary; they
959 are of the ``seem to be OK'' variety. The `3' is the more critical
960 one; it's what determines how eager GHC is about expanding unfoldings.
962 % <tag>@-funfolding-override-threshold<n>@:</tag>
963 % (Default: 8) [Pretty obscure]
964 W hen deciding what unfoldings from a module should be made available
965 % to the rest of the world (via this module's interface), the compiler
966 % normally likes ``small'' expressions.
968 % For example, if it sees @foo = bar@, it will decide that the very
969 % small expression @bar@ is a great unfolding for @foo@. But if
970 % @bar@ turns out to be @(True,False,True)@, we would probably
971 % prefer <em>that</em> for the unfolding for @foo@.
973 % Should we ``override'' the initial small unfolding from @foo=bar@
974 % with the bigger-but-better one? Yes, if the bigger one's ``size'' is
975 % still under the ``override threshold.'' You can use this flag to
976 % adjust this threshold (why, I'm not sure).
978 % <tag>@-fliberated-case-threshold<n>@:</tag>
979 % (Default: 12) [Vastly obscure: NOT IMPLEMENTED YET]
980 % ``Case liberation'' lifts evaluation out of recursive functions; it
981 % does this by duplicating code. Done without constraint, you can get
982 % serious code bloat; so we only do it if the ``size'' of the duplicated
983 % code is smaller than some ``threshold.'' This flag can fiddle that
986 <tag>@-fsemi-tagging@:</tag>
987 This option (which <em>does not work</em> with the native-code generator)
988 tells the compiler to add extra code to test for already-evaluated
989 values. You win if you have lots of such values during a run of your
990 program, you lose otherwise. (And you pay in extra code space.)
992 We have not played with @-fsemi-tagging@ enough to recommend it.
993 (For all we know, it doesn't even work anymore... Sigh.)
996 %----------------------------------------------------------------------
997 <sect2>@-m*@: platform-specific flags
999 <nidx>-m* options (GHC)</nidx>
1000 <nidx>platform-specific options</nidx>
1001 <nidx>machine-specific options</nidx>
1003 Some flags only make sense for particular target platforms.
1007 (SPARC machines)<nidx>-mv8 option (SPARC only)</nidx>
1008 Means to pass the like-named option to GCC; it says to use the
1009 Version 8 SPARC instructions, notably integer multiply and divide.
1010 The similiar @-m*@ GCC options for SPARC also work, actually.
1012 <tag>@-mlong-calls@:</tag>
1013 (HPPA machines)<nidx>-mlong-calls option (HPPA only)</nidx>
1014 Means to pass the like-named option to GCC. Required for Very Big
1015 modules, maybe. (Probably means you're in trouble...)
1017 <tag>@-monly-[32]-regs@:</tag>
1018 (iX86 machines)<nidx>-monly-N-regs option (iX86 only)</nidx>
1019 GHC tries to ``steal'' four registers from GCC, for performance
1020 reasons; it almost always works. However, when GCC is compiling some
1021 modules with four stolen registers, it will crash, probably saying:
1023 Foo.hc:533: fixed or forbidden register was spilled.
1024 This may be due to a compiler bug or to impossible asm
1025 statements or clauses.
1027 Just give some registers back with @-monly-N-regs@. Try `3' first,
1028 then `2'. If `2' doesn't work, please report the bug to us.
1031 %----------------------------------------------------------------------
1032 <sect2>Code improvement by the C compiler.
1033 <label id="optimise-C-compiler">
1035 <nidx>optimisation by GCC</nidx>
1036 <nidx>GCC optimisation</nidx>
1038 The C~compiler (GCC) is run with @-O@ turned on. (It has
1041 If you want to run GCC with @-O2@---which may be worth a few
1042 percent in execution speed---you can give a
1043 @-O2-for-C@<nidx>-O2-for-C option</nidx> option.
1045 %************************************************************************
1047 <sect1>Options related to a particular phase
1048 <label id="options-phases">
1051 %************************************************************************
1053 <sect2> The C pre-processor
1054 <label id="c-pre-processor">
1056 <nidx>pre-processing: cpp</nidx>
1057 <nidx>C pre-processor options</nidx>
1058 <nidx>cpp, pre-processing with</nidx>
1060 The C pre-processor @cpp@ is run over your Haskell code only if the
1061 @-cpp@ option <nidx>-cpp option</nidx> is given. Unless you are
1062 building a large system with significant doses of conditional
1063 compilation, you really shouldn't need it.
1065 <tag>@-D<foo>@:</tag>
1066 <nidx>-D<name> option</nidx>
1067 Define macro @<foo>@ in the usual way. NB: does <em>not</em> affect
1068 @-D@ macros passed to the C~compiler when compiling via C! For
1069 those, use the @-optc-Dfoo@ hack...
1071 <tag>@-U<foo>@:</tag>
1072 <nidx>-U<name> option</nidx>
1073 Undefine macro @<foo>@ in the usual way.
1075 <tag>@-I<dir>@:</tag>
1076 <nidx>-I<dir> option</nidx>
1077 Specify a directory in which to look for @#include@ files, in
1081 The @ghc@ driver pre-defines several macros:
1083 <tag>@__HASKELL1__@:</tag>
1084 <nidx>__HASKELL1__ macro</nidx>
1085 If defined to $n$, that means GHC supports the
1086 Haskell language defined in the Haskell report version $1.n$.
1089 NB: This macro is set both when pre-processing Haskell source and
1090 when pre-processing generated C (@.hc@) files.
1092 <tag>@__GLASGOW_HASKELL__@:</tag>
1093 <nidx>__GLASGOW_HASKELL__ macro</nidx>
1094 For version $n$ of the GHC system, this will be @#define@d to
1095 $100 \times n$. So, for version~3.00, it is 300.
1097 This macro is <em>only</em> set when pre-processing Haskell source.
1098 (<em>Not</em> when pre-processing generated C.)
1100 With any luck, @__GLASGOW_HASKELL__@ will be undefined in all other
1101 implementations that support C-style pre-processing.
1103 (For reference: the comparable symbols for other systems are:
1104 @__HUGS__@ for Hugs and @__HBC__@ for Chalmers.)
1106 <tag>@__CONCURRENT_HASKELL__@:</tag>
1107 <nidx>__CONCURRENT_HASKELL__ macro</nidx>
1108 Only defined when @-concurrent@ is in use!
1109 This symbol is defined when pre-processing Haskell (input) and
1110 pre-processing C (GHC output).
1112 <tag>@__PARALLEL_HASKELL__@:</tag>
1113 <nidx>__PARALLEL_HASKELL__ macro</nidx>
1114 Only defined when @-parallel@ is in use! This symbol is defined when
1115 pre-processing Haskell (input) and pre-processing C (GHC output).
1118 Options other than the above can be forced through to the C
1119 pre-processor with the @-opt@ flags (see
1120 Section <ref name="Forcing options to a particular phase." id="forcing-options-through">).
1122 A small word of warning: @-cpp@ is not friendly to ``string
1123 gaps''.<nidx>-cpp vs string gaps</nidx><nidx>string gaps vs -cpp</nidx>. In
1124 other words, strings such as the following:
1132 don't work with @-cpp@; @/usr/bin/cpp@ elides the
1133 backslash-newline pairs.
1135 However, it appears that if you add a space at the end of the line,
1136 then @cpp@ (at least GNU @cpp@ and possibly other @cpp@s)
1137 leaves the backslash-space pairs alone and the string gap works as
1140 %************************************************************************
1142 <sect2>Options affecting the C compiler (if applicable)
1143 <label id="options-C-compiler">
1145 <nidx>include-file options</nidx>
1146 <nidx>C compiler options</nidx>
1147 <nidx>GCC options</nidx>
1149 %************************************************************************
1151 At the moment, quite a few common C-compiler options are passed on
1152 quietly to the C compilation of Haskell-compiler-generated C files.
1153 THIS MAY CHANGE. Meanwhile, options so sent are:
1156 @-ansi@ | do ANSI C (not K&R) @@
1157 @-pedantic@ | be so@@
1158 @-dgcc-lint@ | (hack) short for ``make GCC very paranoid''@@
1160 <nidx>-ansi option (for GCC)</nidx>
1161 <nidx>-pedantic option (for GCC)</nidx>
1162 <nidx>-dgcc-lint option (GCC paranoia)</nidx>
1164 If you are compiling with lots of @ccalls@, etc., you may need to
1165 tell the C~compiler about some @#include@ files. There is no real
1166 pretty way to do this, but you can use this hack from the
1170 % ghc -c '-#include <X/Xlib.h>' Xstuff.lhs
1174 %************************************************************************
1176 <sect2>Linking and consistency-checking
1177 <label id="options-linker">
1179 <nidx>linker options</nidx>
1180 <nidx>ld options</nidx>
1182 %************************************************************************
1184 GHC has to link your code with various libraries, possibly including:
1185 user-supplied, GHC-supplied, and system-supplied (@-lm@ math
1186 library, for example).
1189 <tag>@-l<FOO>@:</tag>
1190 <nidx>-l<lib> option</nidx>
1191 Link in a library named @lib<FOO>.a@ which resides somewhere on the
1192 library directories path.
1194 Because of the sad state of most UNIX linkers, the order of such
1195 options does matter. Thus: @ghc -lbar *.o@ is almost certainly
1196 wrong, because it will search @libbar.a@ <em>before</em> it has
1197 collected unresolved symbols from the @*.o@ files.
1198 @ghc *.o -lbar@ is probably better.
1200 The linker will of course be informed about some GHC-supplied
1201 libraries automatically; these are:
1204 <bf>-l equivalent</bf> | <bf>description</bf> @@
1206 @-lHSrts,-lHSclib@ | basic runtime libraries @@
1207 @-lHS@ | standard Prelude library @@
1208 @-lHS_cbits@ | C support code for standard Prelude library @@
1209 @-lgmp@ | GNU multi-precision library (for Integers)@@
1212 <nidx>-lHS library</nidx>
1213 <nidx>-lHS_cbits library</nidx>
1214 <nidx>-lHSrts library</nidx>
1215 <nidx>-lgmp library</nidx>
1217 <tag>@-syslib <name>@:</tag>
1218 <nidx>-syslib <name> option</nidx>
1220 If you are using a Haskell ``system library'' (e.g., the POSIX
1221 library), just use the @-syslib posix@ option, and the correct code
1222 should be linked in.
1224 <tag>@-L<dir>@:</tag>
1225 <nidx>-L<dir> option</nidx>
1226 Where to find user-supplied libraries... Prepend the directory
1227 @<dir>@ to the library directories path.
1229 <tag>@-static@:</tag>
1230 <nidx>-static option</nidx>
1231 Tell the linker to avoid shared libraries.
1233 <tag>@-no-link-chk@ and @-link-chk@:</tag>
1234 <nidx>-no-link-chk option</nidx>
1235 <nidx>-link-chk option</nidx>
1236 <nidx>consistency checking of executables</nidx>
1237 By default, immediately after linking an executable, GHC verifies that
1238 the pieces that went into it were compiled with compatible flags; a
1239 ``consistency check''.
1240 (This is to avoid mysterious failures caused by non-meshing of
1241 incompatibly-compiled programs; e.g., if one @.o@ file was compiled
1242 for a parallel machine and the others weren't.) You may turn off this
1243 check with @-no-link-chk@. You can turn it (back) on with
1244 @-link-chk@ (the default).
1247 %************************************************************************
1249 <sect1>Using Concurrent Haskell
1251 <nidx>Concurrent Haskell---use</nidx>
1253 %************************************************************************
1255 To compile a program as Concurrent Haskell, use the @-concurrent@
1256 option,<nidx>-concurrent option</nidx> both when compiling <em>and
1257 linking</em>. You will probably need the @-fglasgow-exts@ option, too.
1259 Three RTS options are provided for modifying the behaviour of the
1260 threaded runtime system. See the descriptions of @-C[<us>]@, @-q@,
1261 and @-t<num>@ in Section <ref name="RTS options for Concurrent/Parallel Haskell" id="parallel-rts-opts">.
1263 %************************************************************************
1265 <sect2>Potential problems with Concurrent Haskell
1266 <label id="concurrent-problems">
1268 <nidx>Concurrent Haskell problems</nidx>
1269 <nidx>problems, Concurrent Haskell</nidx>
1271 %************************************************************************
1273 The main thread in a Concurrent Haskell program is given its own
1274 private stack space, but all other threads are given stack space from
1275 the heap. Stack space for the main thread can be
1276 adjusted as usual with the @-K@ RTS
1277 option,<nidx>-K RTS option (concurrent, parallel)</nidx> but if this
1278 private stack space is exhausted, the main thread will switch to stack
1279 segments in the heap, just like any other thread. Thus, problems
1280 which would normally result in stack overflow in ``sequential Haskell''
1281 can be expected to result in heap overflow when using threads.
1283 The concurrent runtime system uses black holes as synchronisation
1284 points for subexpressions which are shared among multiple threads. In
1285 ``sequential Haskell'', a black hole indicates a cyclic data
1286 dependency, which is a fatal error. However, in concurrent execution, a
1287 black hole may simply indicate that the desired expression is being
1288 evaluated by another thread. Therefore, when a thread encounters a
1289 black hole, it simply blocks and waits for the black hole to be
1290 updated. Cyclic data dependencies will result in deadlock, and the
1291 program will fail to terminate.
1293 Because the concurrent runtime system uses black holes as
1294 synchronisation points, it is not possible to disable black-holing
1295 with the @-N@ RTS option.<nidx>-N RTS option</nidx> Therefore, the use
1296 of signal handlers (including timeouts) with the concurrent runtime
1297 system can lead to problems if a thread attempts to enter a black hole
1298 that was created by an abandoned computation. The use of signal
1299 handlers in conjunction with threads is strongly discouraged.
1302 %************************************************************************
1304 <sect1>Using Parallel Haskell
1306 <nidx>Parallel Haskell---use</nidx>
1308 %************************************************************************
1310 [You won't be able to execute parallel Haskell programs unless PVM3
1311 (Parallel Virtual Machine, version 3) is installed at your site.]
1313 To compile a Haskell program for parallel execution under PVM, use the
1314 @-parallel@ option,<nidx>-parallel option</nidx> both when compiling
1315 <em>and linking</em>. You will probably want to @import Parallel@
1316 into your Haskell modules.
1318 To run your parallel program, once PVM is going, just invoke it ``as
1319 normal''. The main extra RTS option is @-N<n>@, to say how many
1320 PVM ``processors'' your program to run on. (For more details of
1321 all relevant RTS options, please see Section <ref name="RTS options for Concurrent/Parallel Haskell" id="parallel-rts-opts">.)
1323 In truth, running Parallel Haskell programs and getting information
1324 out of them (e.g., parallelism profiles) is a battle with the vagaries of
1325 PVM, detailed in the following sections.
1327 %************************************************************************
1329 <sect2>Dummy's guide to using PVM
1331 <nidx>PVM, how to use</nidx>
1332 <nidx>Parallel Haskell---PVM use</nidx>
1334 %************************************************************************
1336 Before you can run a parallel program under PVM, you must set the
1337 required environment variables (PVM's idea, not ours); something like,
1338 probably in your @.cshrc@ or equivalent:
1340 setenv PVM_ROOT /wherever/you/put/it
1341 setenv PVM_ARCH `$PVM_ROOT/lib/pvmgetarch`
1342 setenv PVM_DPATH $PVM_ROOT/lib/pvmd
1345 Creating and/or controlling your ``parallel machine'' is a purely-PVM
1346 business; nothing specific to Parallel Haskell.
1348 You use the @pvm@<nidx>pvm command</nidx> command to start PVM on your
1349 machine. You can then do various things to control/monitor your
1350 ``parallel machine;'' the most useful being:
1353 @Control-D@ & exit @pvm@, leaving it running \\
1354 @halt@ & kill off this ``parallel machine'' \& exit \\
1355 @add <host>@ & add @<host>@ as a processor \\
1356 @delete <host>@ & delete @<host>@ \\
1357 @reset@ & kill what's going, but leave PVM up \\
1358 @conf@ & list the current configuration \\
1359 @ps@ & report processes' status \\
1360 @pstat <pid>@ & status of a particular process \\
1363 The PVM documentation can tell you much, much more about @pvm@!
1365 %************************************************************************
1367 <sect2>Parallelism profiles
1369 <nidx>parallelism profiles</nidx>
1370 <nidx>profiles, parallelism</nidx>
1371 <nidx>visualisation tools</nidx>
1373 %************************************************************************
1375 With Parallel Haskell programs, we usually don't care about the
1376 results---only with ``how parallel'' it was! We want pretty pictures.
1378 Parallelism profiles (\`a la @hbcpp@) can be generated with the
1379 @-q@<nidx>-q RTS option (concurrent, parallel)</nidx> RTS option. The
1380 per-processor profiling info is dumped into files named
1381 @<full-path><program>.gr@. These are then munged into a PostScript picture,
1382 which you can then display. For example, to run your program
1383 @a.out@ on 8 processors, then view the parallelism profile, do:
1386 % ./a.out +RTS -N8 -q
1387 % grs2gr *.???.gr > temp.gr # combine the 8 .gr files into one
1388 % gr2ps -O temp.gr # cvt to .ps; output in temp.ps
1389 % ghostview -seascape temp.ps # look at it!
1392 The scripts for processing the parallelism profiles are distributed
1393 in @ghc/utils/parallel/@.
1395 %************************************************************************
1397 <sect2>Other useful info about running parallel programs
1400 %************************************************************************
1402 The ``garbage-collection statistics'' RTS options can be useful for
1403 seeing what parallel programs are doing. If you do either
1404 @+RTS -Sstderr@<nidx>-Sstderr RTS option</nidx> or @+RTS -sstderr@, then
1405 you'll get mutator, garbage-collection, etc., times on standard
1406 error. The standard error of all PE's other than the `main thread'
1407 appears in @/tmp/pvml.nnn@, courtesy of PVM.
1409 Whether doing @+RTS -Sstderr@ or not, a handy way to watch
1410 what's happening overall is: @tail -f /tmp/pvml.nnn@.
1412 %************************************************************************
1414 <sect2>RTS options for Concurrent/Parallel Haskell
1415 <label id="parallel-rts-opts">
1417 <nidx>RTS options, concurrent</nidx>
1418 <nidx>RTS options, parallel</nidx>
1419 <nidx>Concurrent Haskell---RTS options</nidx>
1420 <nidx>Parallel Haskell---RTS options</nidx>
1422 %************************************************************************
1424 Besides the usual runtime system (RTS) options
1425 (Section <ref name="Running a compiled program" id="runtime-control">), there are a few options particularly
1426 for concurrent/parallel execution.
1430 <nidx>-N<N> RTS option (parallel)</nidx>
1431 (PARALLEL ONLY) Use @<N>@ PVM processors to run this program;
1434 <tag>@-C[<us>]@:</tag>
1435 <nidx>-C<us> RTS option</nidx>
1436 Sets the context switch interval to @<us>@ microseconds. A context
1437 switch will occur at the next heap allocation after the timer expires.
1438 With @-C0@ or @-C@, context switches will occur as often as
1439 possible (at every heap allocation). By default, context switches
1440 occur every 10 milliseconds. Note that many interval timers are only
1441 capable of 10 millisecond granularity, so the default setting may be
1442 the finest granularity possible, short of a context switch at every
1446 <nidx>-q RTS option</nidx>
1447 Produce a quasi-parallel profile of thread activity, in the file
1448 @<program>.qp@. In the style of @hbcpp@, this profile records
1449 the movement of threads between the green (runnable) and red (blocked)
1450 queues. If you specify the verbose suboption (@-qv@), the green
1451 queue is split into green (for the currently running thread only) and
1452 amber (for other runnable threads). We do not recommend that you use
1453 the verbose suboption if you are planning to use the @hbcpp@
1454 profiling tools or if you are context switching at every heap check
1457 <tag>@-t<num>@:</tag>
1458 <nidx>-t<num> RTS option</nidx>
1459 Limit the number of concurrent threads per processor to @<num>@.
1460 The default is 32. Each thread requires slightly over 1K <em>words</em>
1461 in the heap for thread state and stack objects. (For 32-bit machines,
1462 this translates to 4K bytes, and for 64-bit machines, 8K bytes.)
1465 <nidx>-d RTS option (parallel)</nidx>
1466 (PARALLEL ONLY) Turn on debugging. It pops up one xterm (or GDB, or
1467 something...) per PVM processor. We use the standard @debugger@
1468 script that comes with PVM3, but we sometimes meddle with the
1469 @debugger2@ script. We include ours in the GHC distribution,
1470 in @ghc/utils/pvm/@.
1472 <tag>@-e<num>@:</tag>
1473 <nidx>-e<num> RTS option (parallel)</nidx>
1474 (PARALLEL ONLY) Limit the number of pending sparks per processor to
1475 @<num>@. The default is 100. A larger number may be appropriate if
1476 your program generates large amounts of parallelism initially.
1478 <tag>@-Q<num>@:</tag>
1479 <nidx>-Q<num> RTS option (parallel)</nidx>
1480 (PARALLEL ONLY) Set the size of packets transmitted between processors
1481 to @<num>@. The default is 1024 words. A larger number may be
1482 appropriate if your machine has a high communication cost relative to
1486 %************************************************************************
1488 <sect2>Potential problems with Parallel Haskell
1489 <label id="parallel-problems">
1491 <nidx>Parallel Haskell---problems</nidx>
1492 <nidx>problems, Parallel Haskell</nidx>
1494 %************************************************************************
1496 The ``Potential problems'' for Concurrent Haskell also apply for
1497 Parallel Haskell. Please see Section <ref name="Potential problems with Concurrent Haskell" id="concurrent-problems">.