1 \section[invoking-GHC]{Invoking GHC: Command-line options}
2 \index{command-line options}
3 \index{options, GHC command-line}
5 Command-line arguments are either options or file names.
7 Command-line options begin with \tr{-}. They may {\em not} be
8 grouped: \tr{-vO} is different from \tr{-v -O}. Options need not
9 precede filenames: e.g., \tr{ghc *.o -o foo}. All options are
10 processed and then applied to all files; you cannot, for example, invoke
11 \tr{ghc -c -O1 Foo.hs -O2 Bar.hs} to apply different optimisation
12 levels to the files \tr{Foo.hs} and \tr{Bar.hs}. For conflicting
13 options, e.g., \tr{-c -S}, we reserve the right to do anything we
14 want. (Usually, the last one applies.)
16 Options related to profiling, Glasgow extensions to Haskell (e.g.,
17 unboxed values), Concurrent and Parallel Haskell are described in
18 \sectionref{profiling}, \sectionref{glasgow-exts}, and
19 \sectionref{concurrent-and-parallel}, respectively.
21 %************************************************************************
23 \subsection[file-suffixes]{Meaningful file suffixes}
24 \index{suffixes, file}
25 \index{file suffixes for GHC}
27 %************************************************************************
29 File names with ``meaningful'' suffixes (e.g., \tr{.lhs} or \tr{.o})
30 cause the ``right thing'' to happen to those files.
34 \index{lhs suffix@.lhs suffix}
35 A ``literate Haskell'' module.
38 A not-so-literate Haskell module.
41 A Haskell interface file, probably compiler-generated.
44 Intermediate C file produced by the Haskell compiler.
47 A C~file not produced by the Haskell compiler.
50 % C code after it has be preprocessed by the C compiler (using the
54 An assembly-language source file, usually
55 produced by the compiler.
58 An object file, produced by an assembler.
61 Files with other suffixes (or without suffixes) are passed straight
64 %************************************************************************
66 \subsection[options-help]{Help and verbosity options}
67 \index{help options (GHC)}
68 \index{verbose option (GHC)}
70 %************************************************************************
72 A good option to start with is the \tr{-help} (or \tr{-?}) option.
75 GHC spews a long message to standard output and then exits.
77 The \tr{-v}\index{-v option} option makes GHC {\em verbose}: it
78 reports its version number and shows (on stderr) exactly how it invokes each
79 phase of the compilation system. Moreover, it passes
80 the \tr{-v} flag to most phases; each reports
81 its version number (and possibly some other information).
83 Please, oh please, use the \tr{-v} option when reporting bugs!
84 Knowing that you ran the right bits in the right order is always the
85 first thing we want to verify.
87 If you're just interested in the compiler version number, the
88 \tr{--version}\index{--version option} option prints out a
89 one-line string containing the requested info.
91 %************************************************************************
93 \subsection[options-order]{Running the right phases in the right order}
94 \index{order of passes in GHC}
95 \index{pass ordering in GHC}
97 %************************************************************************
99 The basic task of the \tr{ghc} driver is to run each input file
100 through the right phases (parsing, linking, etc.).
102 The first phase to run is determined by the input-file suffix, and the
103 last phase is determined by a flag. If no relevant flag is present,
104 then go all the way through linking. This table summarises:
106 \begin{tabular}{llll}
107 phase of the & suffix saying & flag saying & (suffix of) \\
108 compilation system & ``start here''& ``stop after''& output file \\ \hline
110 literate pre-processor & .lhs & - & - \\
111 C pre-processor (opt.) & - & - & - \\
112 Haskell compiler & .hs & -C, -S & .hc, .s \\
113 C compiler (opt.) & .hc or .c & -S & .s \\
114 assembler & .s & -c & .o \\
115 linker & other & - & a.out \\
121 Thus, a common invocation would be: \tr{ghc -c Foo.hs}
123 Note: What the Haskell compiler proper produces depends on whether a
124 native-code generator is used (producing assembly language) or not
127 %The suffix information may be overridden with a \tr{-lang <suf>}
128 %\index{-lang <suf> option} option. This says: process all inputs
129 %files as if they had suffix \pl{<suf>}. [NOT IMPLEMENTED YET]
131 The option \tr{-cpp}\index{-cpp option} must be given for the C
132 pre-processor phase to be run, that is, the pre-processor will be run
133 over your Haskell source file before continuing.
135 The option \tr{-E}\index{-E option} runs just the pre-processing
136 passes of the compiler, outputting the result on stdout before
137 stopping. If used in conjunction with -cpp, the output is the
138 code blocks of the original (literal) source after having put it
139 through the grinder that is the C pre-processor. Sans \tr{-cpp}, the
140 output is the de-litted version of the original source.
142 The option \tr{-optcpp-E}\index{-optcpp-E option} runs just the
143 pre-processing stage of the C-compiling phase, sending the result to
144 stdout. (For debugging or obfuscation contests, usually.)
146 %************************************************************************
148 \subsection[options-optimise]{Optimisation (code improvement)}
149 \index{optimisation (GHC)}
150 \index{improvement, code (GHC)}
152 %************************************************************************
154 The \tr{-O*} options specify convenient ``packages'' of optimisation
155 flags; the \tr{-f*} options described later on specify {\em individual}
156 optimisations to be turned on/off; the \tr{-m*} options specify {\em
157 machine-specific} optimisations to be turned on/off.
159 %----------------------------------------------------------------------
160 \subsubsection[optimise-pkgs]{\tr{-O*}: convenient ``packages'' of optimisation flags.}
161 \index{-O options (GHC)}
163 There are {\em many} options that affect the quality of code produced by
164 GHC. Most people only have a general goal, something like ``Compile
165 quickly'' or ``Make my program run like greased lightning.'' The
166 following ``packages'' of optimisations (or lack thereof) should suffice.
168 Once you choose a \tr{-O*} ``package,'' stick with it---don't chop and
169 change. Modules' interfaces {\em will} change with a shift to a new
170 \tr{-O*} option, and you may have to recompile a large chunk of all
171 importing modules before your program can again be run
172 safely\sectionref{recomp}.
175 \item[No \tr{-O*}-type option specified:]
176 \index{-O* not specified}
177 This is taken to mean: ``Please compile quickly; I'm not over-bothered
178 about compiled-code quality.'' So, for example: \tr{ghc -c Foo.hs}
180 \item[\tr{-O} or \tr{-O1}:]
183 \index{optimise normally}
184 Means: ``Generate good-quality code without taking too long about it.''
185 Thus, for example: \tr{ghc -c -O Main.lhs}
189 \index{optimise aggressively}
190 Means: ``Apply every non-dangerous optimisation, even if it means
191 significantly longer compile times.''
193 The avoided ``dangerous'' optimisations are those that can make
194 runtime or space {\em worse} if you're unlucky. They are
195 normally turned on or off individually.
197 At the moment, \tr{-O2} is {\em unlikely} to produce
198 better code than \tr{-O}.
202 % \index{optimise minimally}
203 % [``Oh zero''] Means: ``Turn {\em off} as many optimisations (e.g.,
204 % simplifications) as possible.'' This is the only optimisation level
205 % at which the GCC-register-trickery is turned off. {\em You can't use
206 % it unless you have a suitably-built Prelude to match.} Intended for
207 % hard-core debugging.
210 \index{-fvia-C option}
211 \index{-fvia-c option}
212 Compile via C, and don't use the native-code generator.
213 (There are many cases when GHC does this on its own.) You might
214 pick up a little bit of speed by compiling via C. If you use
215 \tr{_ccall_}s or \tr{_casm_}s, you probably {\em have to} use
218 The lower-case incantation, \tr{-fvia-c}, is synonymous.
220 \item[\tr{-O2-for-C}:]
221 \index{-O2-for-C option}
222 Says to run GCC with \tr{-O2}, which may be worth a few percent in
223 execution speed. Don't forget \tr{-fvia-C}, lest you use the
224 native-code generator and bypass GCC altogether!
228 \index{optimising, reset}
229 This option will make GHC ``forget'' any -Oish options it has seen
230 so far. Sometimes useful; for example: \tr{make all EXTRA_HC_OPTS=-Onot}.
232 \item[\tr{-Ofile <file>}:]
233 \index{-Ofile <file> option}
234 \index{optimising, customised}
235 For those who need {\em absolute} control over {\em exactly} what
236 options are used (e.g., compiler writers, sometimes :-), a list of
237 options can be put in a file and then slurped in with \tr{-Ofile}.
239 In that file, comments are of the \tr{#}-to-end-of-line variety; blank
240 lines and most whitespace is ignored.
242 Please ask if you are baffled and would like an example of \tr{-Ofile}!
245 At Glasgow, we don't use a \tr{-O*} flag for day-to-day work. We use
246 \tr{-O} to get respectable speed; e.g., when we want to measure
247 something. When we want to go for broke, we tend to use
248 \tr{-O -fvia-C -O2-for-C} (and we go for lots of coffee breaks).
250 %Here is a table to summarise whether pragmatic interface information
251 %is used or not, whether the native-code generator is used (if
252 %available), and whether we use GCC register tricks (for speed!) on the
255 %\begin{tabular}{lccl}
256 %\tr{-O*} & Interface & Native code & `Registerised' C \\
257 % & pragmas? & (if avail.) & (if avail.) \\ \hline
259 %\pl{<none>} & no & yes & yes, only if \tr{-fvia-C} \\
260 %\tr{-O,-O1} & yes & yes & yes, only if \tr{-fvia-C} \\
261 %\tr{-O2} & yes & no & yes \\
262 %\tr{-Ofile} & yes & yes & yes, only if \tr{-fvia-C} \\
265 The easiest way to see what \tr{-O} (etc) ``really mean'' is to run
266 with \tr{-v}, then stand back in amazement.
267 Alternatively, just look at the
268 \tr{@HsC_minus<blah>} lists in the \tr{ghc} driver script.
270 %----------------------------------------------------------------------
271 \subsubsection{\tr{-f*}: platform-independent flags}
272 \index{-f* options (GHC)}
273 \index{-fno-* options (GHC)}
275 Flags can be turned {\em off} individually. (NB: I hope
276 you have a good reason for doing this....) To turn off the \tr{-ffoo}
277 flag, just use the \tr{-fno-foo} flag.\index{-fno-<opt> anti-option}
278 So, for example, you can say
279 \tr{-O2 -fno-strictness}, which will then drop out any running of the
282 The options you are most likely to want to turn off are:
283 \tr{-fno-strictness}\index{-fno-strictness option} (strictness
284 analyser [because it is sometimes slow]),
285 \tr{-fno-specialise}\index{-fno-specialise option} (automatic
286 specialisation of overloaded functions [because it makes your code
287 bigger]) [US spelling also accepted],
289 \tr{-fno-foldr-build}\index{-fno-foldr-build option}.
291 Should you wish to turn individual flags {\em on}, you are advised to
292 use the \tr{-Ofile} option, described above. Because the order in
293 which optimisation passes are run is sometimes crucial, it's quite
294 hard to do with command-line options.
296 Here are some ``dangerous'' optimisations you {\em might} want to try:
298 %------------------------------------------------------------------
299 \item[\tr{-funfolding-creation-threshold<n>}:]
300 (Default: 30) By raising or lowering this number, you can raise or
301 lower the amount of pragmatic junk that gets spewed into interface
302 files. (An unfolding has a ``size'' that reflects the cost in terms
303 of ``code bloat'' of expanding that unfolding in another module. A
304 bigger Core expression would be assigned a bigger cost.)
306 \item[\tr{-funfolding-use-threshold<n>}:]
307 (Default: 3) By raising or lowering this number, you can make the
308 compiler more or less keen to expand unfoldings.
310 OK, folks, these magic numbers `30' and `3' are mildly arbitrary; they
311 are of the ``seem to be OK'' variety. The `3' is the more critical
312 one; it's what determines how eager GHC is about expanding unfoldings.
314 \item[\tr{-funfolding-override-threshold<n>}:]
315 (Default: 8) [Pretty obscure]
316 When deciding what unfoldings from a module should be made available
317 to the rest of the world (via this module's interface), the compiler
318 normally likes ``small'' expressions.
320 For example, if it sees \tr{foo = bar}, it will decide that the very
321 small expression \tr{bar} is a great unfolding for \tr{foo}. But if
322 \tr{bar} turns out to be \tr{(True,False,True)}, we would probably
323 prefer {\em that} for the unfolding for \tr{foo}.
325 Should we ``override'' the initial small unfolding from \tr{foo=bar}
326 with the bigger-but-better one? Yes, if the bigger one's ``size'' is
327 still under the ``override threshold.'' You can use this flag to
328 adjust this threshold (why, I'm not sure).
330 % \item[\tr{-fliberated-case-threshold<n>}:]
331 % (Default: 12) [Vastly obscure: NOT IMPLEMENTED YET]
332 % ``Case liberation'' lifts evaluation out of recursive functions; it
333 % does this by duplicating code. Done without constraint, you can get
334 % serious code bloat; so we only do it if the ``size'' of the duplicated
335 % code is smaller than some ``threshold.'' This flag can fiddle that
338 \item[\tr{-fsemi-tagging}:]
339 This option (which {\em does not work} with the native-code generator)
340 tells the compiler to add extra code to test for already-evaluated
341 values. You win if you have lots of such values during a run of your
342 program, you lose otherwise. (And you pay in extra code space.)
344 We have not played with \tr{-fsemi-tagging} enough to recommend it.
345 (For all we know, it doesn't even work anymore... Sigh.)
348 %----------------------------------------------------------------------
349 % \subsubsection[optimise-simplifier]{Controlling ``simplification'' in the Haskell compiler.}
351 %Almost everyone turns program transformation
352 % (a.k.a. ``simplification'') on/off via one of the ``packages'' above,
353 %but you can exert absolute control if you want to. Do a \tr{ghc -v -O ...},
354 %and you'll see there are plenty of knobs to turn!
356 %The Core-to-Core and STG-to-STG passes can be run multiple times, and
357 %in varying orders (though you may live to regret it). The on-or-off
358 %global flags, however, are simply, well, on or off.
360 %The best way to give an exact list of options is the \tr{-Ofile}
361 %option, described elsewhere.
363 % [Check out \tr{ghc/compiler/simplCore/SimplCore.lhs} and
364 %\tr{simplStg/SimplStg.lhs} if you {\em really} want to see every
365 %possible Core-to-Core and STG-to-STG pass, respectively. The
366 %on-or-off global flags that effect what happens {\em within} one of
367 %these passes are defined by the \tr{GlobalSwitch} datatype in
368 %\tr{compiler/main/CmdLineOpts.lhs}.]
370 %----------------------------------------------------------------------
371 \subsubsection{\tr{-m*}: platform-specific flags}
372 \index{-m* options (GHC)}
373 \index{platform-specific options}
374 \index{machine-specific options}
376 Some flags only make sense for particular target platforms.
380 (SPARC machines)\index{-mv8 option (SPARC only)}
381 Means to pass the like-named option to GCC; it says to use the
382 Version 8 SPARC instructions, notably integer multiply and divide.
383 The similiar \tr{-m*} GCC options for SPARC also work, actually.
385 \item[\tr{-mlong-calls}:]
386 (HPPA machines)\index{-mlong-calls option (HPPA only)}
387 Means to pass the like-named option to GCC. Required for Very Big
388 modules, maybe. (Probably means you're in trouble...)
390 \item[\tr{-monly-[32]-regs}:]
391 (iX86 machines)\index{-monly-N-regs option (iX86 only)}
392 GHC tries to ``steal'' four registers from GCC, for performance
393 reasons; it almost always works. However, when GCC is compiling some
394 modules with four stolen registers, it will crash, probably saying:
396 Foo.hc:533: fixed or forbidden register was spilled.
397 This may be due to a compiler bug or to impossible asm
398 statements or clauses.
400 Just give some registers back with \tr{-monly-N-regs}. Try `3' first,
401 then `2'. If `2' doesn't work, please report the bug to us.
404 %----------------------------------------------------------------------
405 \subsubsection[optimise-C-compiler]{Code improvement by the C compiler.}
406 \index{optimisation by GCC}
407 \index{GCC optimisation}
409 The C~compiler (GCC) is run with \tr{-O} turned on. (It has
412 If you want to run GCC with \tr{-O2}---which may be worth a few
413 percent in execution speed---you can give a
414 \tr{-O2-for-C}\index{-O2-for-C option} option.
416 %If you are brave or foolish, you might want to omit some checking code
417 % (e.g., for stack-overflow checks), as sketched in
418 %\sectionref{omit-checking}.
420 %************************************************************************
422 \subsection[options-sanity]{Warnings and sanity-checking}
423 \index{sanity-checking options}
425 %************************************************************************
427 If you would like GHC to check that every top-level value has a type
428 signature, use the \tr{-fsignatures-required}
429 option.\index{-fsignatures-required option}
431 If you would like to disallow ``name shadowing,'' i.e., an inner-scope
432 value has the same name as an outer-scope value, then use the
433 \tr{-fwarn-name-shadowing}
434 option.\index{-fwarn-name-shadowing option}
435 This option catches typographical errors that turn into hard-to-find
436 bugs, e.g., in the inadvertent cyclic definition \tr{let x = ... x ... in}.
438 Consequently, this option does {\em not} allow cyclic recursive
441 By default, the compiler will warn you if a set of patterns are either
442 incomplete (i.e., you're only matching on a subset of an algebraic
443 data type's constructors), or overlapping, i.e.,
454 where the last pattern match in \tr{f} won't ever be reached, as the
455 second pattern overlaps it. More often than not, redundant patterns
456 is a programmer mistake/error, but if you don't want the compiler to
457 ``baby-sit'', use the \tr{-fno-warn-overlapping-patterns} option to
458 turn these warnings off.\index{-fno-warn-overlapping-patterns option}
460 Similarly for incomplete patterns, the function \tr{g} will fail when
461 applied to non-empty lists, so the compiler will by default emit a
462 warning about this. The option \tr{-fno-warn-incomplete-patterns}
463 turns rhis off.\index{-fno-warn-incomplete-pattern option}
466 If you're feeling really paranoid, the \tr{-dcore-lint}
467 option\index{-dcore-lint option} is a good choice. It turns on
468 heavyweight intra-pass sanity-checking within GHC. (It checks GHC's
471 %************************************************************************
473 \subsection[options-output]{Re-directing the compilation output(s)}
474 \index{output-directing options}
476 %************************************************************************
478 When compiling a Haskell module, GHC may produce several files of
479 output (usually two).
481 One file is usually an {\em interface file}. If compiling
482 \tr{bar/Foo.hs}, the interface file would normally be \tr{bar/Foo.hi}.
483 The interface output may be directed to another file
484 \tr{bar2/Wurble.iface} with the option
485 \tr{-ohi bar2/Wurble.iface}\index{-ohi <file> option} (not recommended).
487 To avoid generating an interface file at all, use a \tr{-nohi}
488 option.\index{-nohi option}
490 The compiler does not overwrite an existing \tr{.hi} interface file if
491 the new one is byte-for-byte the same as the old one; this is friendly to
492 \tr{make}. When an interface does change, it is often enlightening to
493 be informed. The \tr{-hi-diffs}\index{-hi-diffs option} option will
494 make \tr{ghc} run \tr{diff} on the old and new \tr{.hi} files. You can
495 also record the difference in the interface file itself, the
496 \tr{-keep-hi-diffs}\index{-keep-hi-diffs} option takes care of that.
498 The \tr{.hi} files from GHC 2.xx contain ``usage'' information which
499 changes often and uninterestingly. If you really want to see these
500 changes reported, you need to use the
501 \tr{-hi-diffs-with-usages}\index{-hi-diffs-with-usages option} option.
503 GHC's non-interface output normally goes into a \tr{.hc}, \tr{.o},
504 etc., file, depending on the last-run compilation phase. The option
505 \tr{-o foo}\index{-o option} re-directs the output of that last-run
506 phase to file \tr{foo}.
508 Note: this ``feature'' can be counterintuitive:
509 \tr{ghc -C -o foo.o foo.hs} will put the intermediate C code in the
510 file \tr{foo.o}, name notwithstanding!
512 EXOTICA: But the \tr{-o} option isn't of much use if you have {\em
513 several} input files... Non-interface output files are normally put
514 in the same directory as their corresponding input file came from.
515 You may specify that they be put in another directory using the
516 \tr{-odir <dir>}\index{-odir <dir> option} (the ``Oh, dear'' option).
520 % ghc -c parse/Foo.hs parse/Bar.hs gurgle/Bumble.hs -odir `arch`
523 The output files, \tr{Foo.o}, \tr{Bar.o}, and \tr{Bumble.o} would be
524 put into a subdirectory named after the architecture of the executing
525 machine (\tr{sun4}, \tr{mips}, etc). The directory must already
526 exist; it won't be created.
528 Note that the \tr{-odir} option does {\em not} affect where the
529 interface files are put. In the above example, they would still be
530 put in \tr{parse/Foo.hi}, \tr{parse/Bar.hi}, and
531 \tr{gurgle/Bumble.hi}.
533 MORE EXOTICA: The \tr{-osuf <suffix>}\index{-osuf <suffix> option}
534 will change the \tr{.o} file suffix for object files to whatever
535 you specify. (We use this in compiling the prelude.)
537 Similarly, the \tr{-hisuf <suffix>}\index{-hisuf <suffix> option} will
538 change the \tr{.hi} file suffix for non-system interface files. This
539 can be useful when you are trying to compile a program several ways,
540 all in the same directory. The suffix given is used for {\em all}
541 interfaces files written, {\em and} for all non-system interface files
544 The \tr{-hisuf}/\tr{-osuf} game is useful if you want to compile a
545 program with both GHC and HBC (say) in the same directory. Let HBC
546 use the standard \tr{.hi}/\tr{.o} suffixes; add
547 \tr{-hisuf g_hi -osuf g_o} to your \tr{make} rule for GHC compiling...
549 NB: {\em A change from 0.26 and before:} Before, you might have said
550 \tr{-hisuf _g.hi -osuf _g.o}; now, the \tr{.} is assumed and you
551 specify what comes {\em after} it. (This is a more portable solution
554 % THIS SHOULD HAPPEN AUTOMAGICALLY:
555 % If you want to change the suffix looked for on system-supplied
556 % interface files (notably the \tr{Prelude.hi} file), use the
557 % \tr{-hisuf-prelude <suffix>}\index{-hisuf-prelude <suffix> option}
558 % option. (This may be useful if you've built GHC in various funny
559 % ways, and you are running tests in even more funny ways. It happens.)
561 FURTHER EXOTICA: If you are doing a normal \tr{.hs}-to-\tr{.o} compilation
562 but would like to hang onto the intermediate \tr{.hc} C file, just
563 throw in a \tr{-keep-hc-file-too} option\index{-keep-hc-file-too option}.
564 If you would like to look at the assembler output, toss in a
565 \tr{-keep-s-file-too},\index{-keep-hc-file-too option} too.
567 SAVING GHC STDERR OUTPUT: Sometimes, you may cause GHC to be rather
568 chatty on standard error; with \tr{-fshow-import-specs}, for example.
569 You can instruct GHC to {\em append} this output to a particular log
570 file with a \tr{-odump <blah>}\index{-odump <blah> option} option.
572 TEMPORARY FILES: If you have trouble because of running out of space
573 in \tr{/tmp/} (or wherever your installation thinks temporary files
574 should go), you may use the \tr{-tmpdir <dir>}\index{-tmpdir <dir> option}
575 option to specify an alternate directory. For example, \tr{-tmpdir .}
576 says to put temporary files in the current working directory.
578 BETTER IDEA FOR TEMPORARY FILES: Use your \tr{TMPDIR} environment
579 variable.\index{TMPDIR environment variable} Set it to the name of
580 the directory where temporary files should be put. GCC and other
581 programs will honour the \tr{TMPDIR} variable as well.
583 EVEN BETTER IDEA: Set the \tr{TMPDIR} variable when building
584 GHC, and never worry about \tr{TMPDIR} again. (see the build
587 %************************************************************************
589 \subsection[options-finding-imports-etc]{For finding interface files, etc.}
590 \index{interface files, finding them}
591 \index{finding interface files}
593 %************************************************************************
595 In your program, you import a module \tr{Foo} by saying
596 \tr{import Foo}. GHC goes looking for an interface file, \tr{Foo.hi}.
597 It has a builtin list of directories (notably including \tr{.}) where
600 The \tr{-i<dirs>} option\index{-i<dirs> option} prepends a
601 colon-separated list of \tr{dirs} to the ``import directories'' list.
603 A plain \tr{-i} resets the ``import directories'' list back to nothing.
605 GHC normally imports \tr{Prelude.hi} files for you. If you'd rather
606 it didn't, then give it a \tr{-fno-implicit-prelude}
607 option\index{-fno-implicit-prelude option}. You are unlikely to get
608 very far without a Prelude, but, hey, it's a free country.
610 If you are using a system-supplied non-Prelude library (e.g., the HBC
611 library), just use a \tr{-syslib hbc}\index{-syslib <lib> option}
612 option (for example). The right interface files should then be
615 Once a Haskell module has been compiled to C (\tr{.hc} file), you may
616 wish to specify where GHC tells the C compiler to look for \tr{.h}
617 files. (Or, if you are using the \tr{-cpp} option\index{-cpp option},
618 where it tells the C pre-processor to look...) For this purpose, use
619 a \tr{-I<dir>}\index{-I<dir> option} in the usual C-ish way.
621 Pragmas: Interface files are normally jammed full of
622 compiler-produced {\em pragmas}, which record arities, strictness
623 info, etc. If you think these pragmas are messing you up (or you are
624 doing some kind of weird experiment), you can tell GHC to ignore them
625 with the \tr{-fignore-interface-pragmas}\index{-fignore-interface-pragmas option}
628 When compiling without optimisations on, the compiler is extra-careful
629 about not slurping in data constructors and instance declarations that
630 it will not need. If you believe it is getting it wrong and not
631 importing stuff which you think it should, this optimisation can be
632 turned off with \tr{-fno-prune-tydecls} and \tr{-fno-prune-instdecls}.
633 \index{-fno-prune-tydecls option}\index{-fno-prune-instdecls}
635 See also \sectionref{options-linker}, which describes how the linker
636 finds standard Haskell libraries.
638 %************************************************************************
640 %\subsection[options-names]{Fiddling with namespaces}
642 %************************************************************************
644 %-split-objs and -fglobalise-toplev-names. You don't need them and you
645 %don't want to know; used for the prelude (ToDo).
647 %************************************************************************
649 \subsection[options-CPP]{Related to the C pre-processor}
650 \index{C pre-processor options}
651 \index{pre-processor (cpp) options}
653 %************************************************************************
655 The C pre-processor \tr{cpp} is run over your Haskell code only if the
656 \tr{-cpp} option \index{-cpp option} is given. Unless you are
657 building a large system with significant doses of conditional
658 compilation, you really shouldn't need it.
661 \index{-D<name> option}
662 Define macro \tr{<foo>} in the usual way. NB: does {\em not} affect
663 \tr{-D} macros passed to the C~compiler when compiling via C! For
664 those, use the \tr{-optc-Dfoo} hack...
667 \index{-U<name> option}
668 Undefine macro \tr{<foo>} in the usual way.
671 \index{-I<dir> option}
672 Specify a directory in which to look for \tr{#include} files, in
676 The \tr{ghc} driver pre-defines several macros:
678 \item[\tr{__HASKELL1__}:]
679 \index{__HASKELL1__ macro}
680 If defined to $n$, that means GHC supports the
681 Haskell language defined in the Haskell report version $1.n$.
684 NB: This macro is set both when pre-processing Haskell source and
685 when pre-processing generated C (\tr{.hc}) files.
687 % If you give the \tr{-fhaskell-1.3} flag\index{-fhaskell-1.3 option},
688 % then \tr{__HASKELL1__} is set to 3. Obviously.
690 \item[\tr{__GLASGOW_HASKELL__}:]
691 \index{__GLASGOW_HASKELL__ macro}
692 For version $n$ of the GHC system, this will be \tr{#define}d to
693 $100 \times n$. So, for version~2.02, it is 202.
695 This macro is {\em only} set when pre-processing Haskell source.
696 ({\em Not} when pre-processing generated C.)
698 With any luck, \tr{__GLASGOW_HASKELL__} will be undefined in all other
699 implementations that support C-style pre-processing.
701 (For reference: the comparable symbols for other systems are:
702 \tr{__HUGS__} for Hugs and \tr{__HBC__} for Chalmers.)
704 \item[\tr{__CONCURRENT_HASKELL__}:]
705 \index{__CONCURRENT_HASKELL__ macro}
706 Only defined when \tr{-concurrent} is in use!
707 This symbol is defined when pre-processing Haskell (input) and
708 pre-processing C (GHC output).
710 \item[\tr{__PARALLEL_HASKELL__}:]
711 \index{__PARALLEL_HASKELL__ macro}
712 Only defined when \tr{-parallel} is in use! This symbol is defined when
713 pre-processing Haskell (input) and pre-processing C (GHC output).
716 Options other than the above can be forced through to the C
717 pre-processor with the \tr{-opt} flags (see
718 \sectionref{forcing-options-through}).
720 A small word of warning: \tr{-cpp} is not friendly to ``string
721 gaps''.\index{-cpp vs string gaps}\index{string gaps vs -cpp}. In
722 other words, strings such as the following:
730 don't work with \tr{-cpp}; \tr{/usr/bin/cpp} elides the
731 backslash-newline pairs.
733 However, it appears that if you add a space at the end of the line,
734 then \tr{cpp} (at least GNU \tr{cpp} and possibly other \tr{cpp}s)
735 leaves the backslash-space pairs alone and the string gap works as
738 %************************************************************************
740 \subsection[options-C-compiler]{Options affecting the C compiler (if applicable)}
741 \index{include-file-option}
742 \index{C compiler options}
745 %************************************************************************
747 At the moment, quite a few common C-compiler options are passed on
748 quietly to the C compilation of Haskell-compiler-generated C files.
749 THIS MAY CHANGE. Meanwhile, options so sent are:
752 \tr{-Wall} & get all warnings from GCC \\
753 \tr{-ansi} & do ANSI C (not K\&R) \\
754 \tr{-pedantic} & be so\\
755 \tr{-dgcc-lint} & (hack) short for ``make GCC very paranoid''\\
757 \index{-Wall option (for GCC)}
758 \index{-ansi option (for GCC)}
759 \index{-pedantic option (for GCC)}
760 \index{-dgcc-lint option (GCC paranoia)}
762 If you are compiling with lots of \tr{ccalls}, etc., you may need to
763 tell the C~compiler about some \tr{#include} files. There is no real
764 pretty way to do this, but you can use this hack from the
767 % ghc -c '-#include <X/Xlib.h>' Xstuff.lhs
771 %************************************************************************
773 %\subsection[options-native-code]{Options affecting the native-code generator(s)}
775 %************************************************************************
777 %The only option is to select the target architecture. Right now,
778 %you have only at most one choice: \tr{-fasm-sparc}.\index{-fasm-<target> option}
780 %EXPECT this native-code stuff to change in the future.
782 %************************************************************************
784 \subsection[options-linker]{Linking and consistency-checking}
785 \index{linker options}
788 %************************************************************************
790 GHC has to link your code with various libraries, possibly including:
791 user-supplied, GHC-supplied, and system-supplied (\tr{-lm} math
792 library, for example).
796 \index{-l<lib> option}
797 Link in a library named \tr{lib<FOO>.a} which resides somewhere on the
798 library directories path.
800 Because of the sad state of most UNIX linkers, the order of such
801 options does matter. Thus: \tr{ghc -lbar *.o} is almost certainly
802 wrong, because it will search \tr{libbar.a} {\em before} it has
803 collected unresolved symbols from the \tr{*.o} files.
804 \tr{ghc *.o -lbar} is probably better.
806 The linker will of course be informed about some GHC-supplied
807 libraries automatically; these are:
810 -l equivalent & description \\ \hline
812 -lHSrts,-lHSclib & basic runtime libraries \\
813 -lHS & standard Prelude library \\
814 -lHS\_cbits & C support code for standard Prelude library \\
815 -lgmp & GNU multi-precision library (for Integers)\\
818 \index{-lHS_cbits library}
819 \index{-lHSrts library}
820 \index{-lgmp library}
822 \item[\tr{-syslib <name>}:]
823 \index{-syslib <name> option}
825 If you are using a Haskell ``system library'' (e.g., the HBC
826 library), just use the \tr{-syslib hbc} option, and the correct code
829 %Please see \sectionref{syslibs} for information about
830 %``system libraries.''
833 \index{-L<dir> option}
834 Where to find user-supplied libraries... Prepend the directory
835 \tr{<dir>} to the library directories path.
838 \index{-static option}
839 Tell the linker to avoid shared libraries.
841 \item[\tr{-no-link-chk} and \tr{-link-chk}:]
842 \index{-no-link-chk option}
843 \index{-link-chk option}
844 \index{consistency checking of executables}
845 By default, immediately after linking an executable, GHC verifies that
846 the pieces that went into it were compiled with compatible flags; a
847 ``consistency check''.
848 (This is to avoid mysterious failures caused by non-meshing of
849 incompatibly-compiled programs; e.g., if one \tr{.o} file was compiled
850 for a parallel machine and the others weren't.) You may turn off this
851 check with \tr{-no-link-chk}. You can turn it (back) on with
852 \tr{-link-chk} (the default).
855 %************************************************************************
857 \subsection[options-compiler-RTS]{For the compiler's RTS: heap, stack sizes, etc.}
858 \index{heap-size options (for GHC)}
859 \index{stack-size options (for GHC)}
861 %************************************************************************
863 The compiler is itself a Haskell program, so it has a tweakable
864 runtime-system (RTS), just like any other Haskell program.
867 \item[\tr{-H<size>} or \tr{-Rmax-heapsize <size>}:]
868 \index{-H<size> option}
869 \index{-Rmax-heapsize <size> option}
870 Don't use more than \tr{<size>} {\em bytes} for heap space. If more
871 than one of these arguments is given, the largest will be taken.
873 A size of zero can be used to reset the heap size downwards. For
874 example, to run GHC with a heap of 250KB (the default is 6MB), do
877 \item[\tr{-K<size>} or \tr{-Rmax-stksize <size>}:]
878 \index{-K<size> option}
879 \index{-Rmax-stksize <size> option}
880 Set the stack space to \tr{<size>} bytes. If you have to set it very
881 high [a megabyte or two, say], the compiler is probably looping, which
882 is a BUG (please report).
884 A size of zero can be used to rest the stack size downwards, as above.
886 \item[\tr{-Rscale-sizes<factor>}:]
887 \index{-Rscale-sizes<factor> option}
888 Multiply the given (or default) heap and stack sizes by \tr{<factor>}.
889 For example, on a DEC Alpha (a 64-bit machine), you might want to
890 double those space sizes; just use \tr{-Rscale-sizes2}.
892 A non-integral factor is OK, too: \tr{-Rscale-sizes1.2}.
894 \item[\tr{-Rghc-timing}:]
895 \index{-Rghc-timing option}
896 Reports a one-line useful collection of time- and space- statistics
897 for a module's compilation.
899 \item[\tr{-Rgc-stats}:]
900 \index{-Rgc-stats option}
901 Report garbage-collection statistics. It will create a
902 \tr{<foo>.stat} file, in some obvious place (I hope).
904 Alternatively, if you'd rather the GC stats went straight to standard
905 error, you can ``cheat'' by using, instead: \tr{-optCrts-Sstderr}.
908 %\index{-Rhbc option}
909 %Tell the compiler it has an HBC-style RTS; i.e., it was compiled with
910 %HBC. Not used in Real Life.
913 %\index{-Rghc option}
914 %Tell the compiler it has a GHC-style RTS; i.e., it was compiled with
915 %GHC. Not used in Real Life.
918 For all \tr{<size>}s: If the last character of \tr{size} is a K,
919 multiply by 1000; if an M, by 1,000,000; if a G, by 1,000,000,000.
920 Sizes are always in {\em bytes}, not words. Good luck on the G's (I
921 think the counter is still only 32-bits [WDP])!
923 %************************************************************************
925 %\subsection[options-cross-compiling]{For cross-compiling to another architecture}
927 %************************************************************************
929 % (We do this for GRIP at Glasgow; it's hacked in---not proper
930 %cross-compiling support. But you could do the same, if required...)
932 %The \tr{-target <arch>} option\index{-target <arch> option} says to
933 %generate code for the \tr{<arch>} architecture.
935 %************************************************************************
937 \subsection[options-parallel]{For Concurrent and Parallel Haskell}
939 %************************************************************************
941 For the full story on using GHC for concurrent \& parallel Haskell
942 programming, please see \Sectionref{concurrent-and-parallel}.
944 %The \tr{-fparallel} option\index{-fparallel option} tells the compiler
945 %to generate code for parallel execution. The \tr{-mgrip}
946 %option\index{-mgrip option} says that the code should be explicitly
947 %suitable for the GRIP multiprocessor (the one in our Glasgow basement).
949 %************************************************************************
951 %\subsection[options-experimental]{For experimental purposes}
952 %\index{experimental options}
954 %************************************************************************
956 %From time to time, we provide GHC options for ``experimenting.'' Easy
957 %come, easy go. In version~0.26, the ``experimental'' options are:
959 %\item[\tr{-firrefutable-tuples} option:]
960 %\index{-firrefutable-tuples option (experimental)}
961 %Pretend that every tuple pattern is irrefutable; i.e., has a
962 %``twiddle'' (\tr{~}) in front of it.
964 %Some parts of the GHC system {\em depend} on strictness properties which
965 %\tr{-firrefutable-tuples} may undo, notably the low-level state-transformer
966 %stuff, which includes I/O (!). You're on your own...
968 %\item[\tr{-fall-strict} option:]
969 %\index{-fall-strict option (experimental)}
970 % (DOESN'T REALLY WORK, I THINK) Changes the strictness analyser so
971 %that, when it asks the question ``Is this function argument certain to
972 %be evaluated?'', the answer is always ``yes''.
974 %Compilation is changed in no other way.
977 % -firrefutable-everything
980 %************************************************************************
982 \subsection[options-debugging]{For debugging the compiler}
983 \index{debugging options (for GHC)}
985 %************************************************************************
987 HACKER TERRITORY. HACKER TERRITORY.
990 %----------------------------------------------------------------------
991 \subsubsection[replacing-phases]{Replacing the program for one or more phases.}
992 \index{GHC phases, changing}
993 \index{phases, changing GHC}
995 You may specify that a different program
996 be used for one of the phases of the compilation system, in place of
997 whatever the driver \tr{ghc} has wired into it. For example, you
998 might want to try a different assembler. The
999 \tr{-pgm<phase-code><program-name>}\index{-pgm<phase><stuff> option} option to
1000 \tr{ghc} will cause it to use \pl{<program-name>} for phase
1001 \pl{<phase-code>}, where the codes to indicate the phases are:
1004 code & phase \\ \hline
1005 L & literate pre-processor \\
1006 P & C pre-processor (if -cpp only) \\
1007 C & Haskell compiler \\
1013 %----------------------------------------------------------------------
1014 \subsubsection[forcing-options-through]{Forcing options to a particular phase.}
1015 \index{forcing GHC-phase options}
1017 The preceding sections describe driver options that are mostly
1018 applicable to one particular phase. You may also {\em force} a
1019 specific option \tr{<option>} to be passed to a particular phase
1020 \tr{<phase-code>} by feeding the driver the option
1021 \tr{-opt<phase-code><option>}.\index{-opt<phase><stuff> option} The
1022 codes to indicate the phases are the same as in the previous section.
1024 So, for example, to force an \tr{-Ewurble} option to the assembler, you
1025 would tell the driver \tr{-opta-Ewurble} (the dash before the E is
1028 Besides getting options to the Haskell compiler with \tr{-optC<blah>},
1029 you can get options through to its runtime system with
1030 \tr{-optCrts<blah>}\index{-optCrts<blah> option}.
1032 So, for example: when I want to use my normal driver but with my
1033 profiled compiler binary, I use this script:
1036 exec /local/grasp_tmp3/simonpj/ghc-BUILDS/working-alpha/ghc/driver/ghc \
1037 -pgmC/local/grasp_tmp3/simonpj/ghc-BUILDS/working-hsc-prof/hsc \
1043 %----------------------------------------------------------------------
1044 \subsubsection[dumping-output]{Dumping out compiler intermediate structures}
1045 \index{dumping GHC intermediates}
1046 \index{intermediate passes, output}
1051 Don't bother generating C output {\em or} an interface file. Usually
1052 used in conjunction with one or more of the \tr{-ddump-*} options; for
1053 example: \tr{ghc -noC -ddump-simpl Foo.hs}
1057 {\em Do} generate an interface file. This would normally be used in
1058 conjunction with \tr{-noC}, which turns off interface generation;
1059 thus: \tr{-noC -hi}.
1061 \item[\tr{-dshow-passes}:]
1062 \index{-dshow-passes option}
1063 Prints a message to stderr as each pass starts. Gives a warm but
1064 undoubtedly misleading feeling that GHC is telling you what's
1067 \item[\tr{-ddump-<pass>}:]
1068 \index{-ddump-<pass> options}
1069 Make a debugging dump after pass \tr{<pass>} (may be common enough to
1070 need a short form...). Some of the most useful ones are:
1073 \tr{-ddump-rdr} & reader output (earliest stuff in the compiler) \\
1074 \tr{-ddump-rn} & renamer output \\
1075 \tr{-ddump-tc} & typechecker output \\
1076 \tr{-ddump-deriv} & derived instances \\
1077 \tr{-ddump-ds} & desugarer output \\
1078 \tr{-ddump-simpl} & simplifer output (Core-to-Core passes) \\
1079 \tr{-ddump-stranal} & strictness analyser output \\
1080 \tr{-ddump-occur-anal} & `occurrence analysis' output \\
1081 \tr{-ddump-spec} & dump specialisation info \\
1082 \tr{-ddump-stg} & output of STG-to-STG passes \\
1083 \tr{-ddump-absC} & {\em un}flattened Abstract~C \\
1084 \tr{-ddump-flatC} & {\em flattened} Abstract~C \\
1085 \tr{-ddump-realC} & same as what goes to the C compiler \\
1086 \tr{-ddump-asm} & assembly language from the native-code generator \\
1088 \index{-ddump-rdr option}%
1089 \index{-ddump-rn option}%
1090 \index{-ddump-tc option}%
1091 \index{-ddump-deriv option}%
1092 \index{-ddump-ds option}%
1093 \index{-ddump-simpl option}%
1094 \index{-ddump-stranal option}%
1095 \index{-ddump-occur-anal option}%
1096 \index{-ddump-spec option}%
1097 \index{-ddump-stg option}%
1098 \index{-ddump-absC option}%
1099 \index{-ddump-flatC option}%
1100 \index{-ddump-realC option}%
1101 \index{-ddump-asm option}
1103 %For any other \tr{-ddump-*} options: consult the source, notably
1104 %\tr{ghc/compiler/main/CmdLineOpts.lhs}.
1106 \item[\tr{-dverbose-simpl} and \tr{-dverbose-stg}:]
1107 \index{-dverbose-simpl option}
1108 \index{-dverbose-stg option}
1109 Show the output of the intermediate Core-to-Core and STG-to-STG
1110 passes, respectively. ({\em Lots} of output!) So: when we're
1113 % ghc -noC -O -ddump-simpl -dverbose-simpl -dcore-lint Foo.hs
1116 \item[\tr{-dppr-{user,debug,all}}:]
1117 \index{-dppr-user option}
1118 \index{-dppr-debug option}
1119 \index{-dppr-all option}
1120 Debugging output is in one of several ``styles.'' Take the printing
1121 of types, for example. In the ``user'' style, the compiler's internal
1122 ideas about types are presented in Haskell source-level syntax,
1123 insofar as possible. In the ``debug'' style (which is the default for
1124 debugging output), the types are printed in the most-often-desired
1125 form, with explicit foralls, etc. In the ``show all'' style, very
1126 verbose information about the types (e.g., the Uniques on the
1127 individual type variables) is displayed.
1129 \item[\tr{-ddump-raw-asm}:]
1130 \index{-ddump-raw-asm option}
1131 Dump out the assembly-language stuff, before the ``mangler'' gets it.
1133 \item[\tr{-ddump-rn-trace}:]
1134 \index{-ddump-rn-trace}
1135 Make the renamer be *real* chatty about what it is upto.
1137 \item[\tr{-dshow-rn-stats}:]
1138 \index{-dshow-rn-stats}
1139 Print out summary of what kind of information the renamer had to bring
1141 \item[\tr{-dshow-unused-imports}:]
1142 \index{-dshow-unused-imports}
1143 Have the renamer report what imports does not contribute.
1145 \item[\tr{-fwarn-unused-names}:]
1146 \index{-fwarn-unused-names}
1147 Have the renamer report which locally defined names are not used/exported.
1150 %\item[\tr{-dgc-debug}:]
1151 %\index{-dgc-debug option}
1152 %Enables some debugging code related to the garbage-collector.
1155 %ToDo: -ddump-asm-insn-counts
1156 %-ddump-asm-globals-info
1158 %----------------------------------------------------------------------
1159 \subsubsection{How to read Core syntax (from some \tr{-ddump-*} flags)}
1160 \index{reading Core syntax}
1161 \index{Core syntax, how to read}
1163 Let's do this by commenting an example. It's from doing
1164 \tr{-ddump-ds} on this code:
1166 skip2 m = m : skip2 (m+2)
1169 Before we jump in, a word about names of things. Within GHC,
1170 variables, type constructors, etc., are identified by their
1171 ``Uniques.'' These are of the form `letter' plus `number' (both
1172 loosely interpreted). The `letter' gives some idea of where the
1173 Unique came from; e.g., \tr{_} means ``built-in type variable'';
1174 \tr{t} means ``from the typechecker''; \tr{s} means ``from the
1175 simplifier''; and so on. The `number' is printed fairly compactly in
1176 a `base-62' format, which everyone hates except me (WDP).
1178 Remember, everything has a ``Unique'' and it is usually printed out
1179 when debugging, in some form or another. So here we go...
1183 Main.skip2{-r1L6-} :: _forall_ a$_4 =>{{Num a$_4}} -> a$_4 -> [a$_4]
1185 --# `r1L6' is the Unique for Main.skip2;
1186 --# `_4' is the Unique for the type-variable (template) `a'
1187 --# `{{Num a$_4}}' is a dictionary argument
1191 --# `_NI_' means "no (pragmatic) information" yet; it will later
1192 --# evolve into the GHC_PRAGMA info that goes into interface files.
1194 Main.skip2{-r1L6-} =
1195 /\ _4 -> \ d.Num.t4Gt ->
1198 +.t4Hg :: _4 -> _4 -> _4
1200 +.t4Hg = (+{-r3JH-} _4) d.Num.t4Gt
1202 fromInt.t4GS :: Int{-2i-} -> _4
1204 fromInt.t4GS = (fromInt{-r3JX-} _4) d.Num.t4Gt
1206 --# The `+' class method (Unique: r3JH) selects the addition code
1207 --# from a `Num' dictionary (now an explicit lamba'd argument).
1208 --# Because Core is 2nd-order lambda-calculus, type applications
1209 --# and lambdas (/\) are explicit. So `+' is first applied to a
1210 --# type (`_4'), then to a dictionary, yielding the actual addition
1211 --# function that we will use subsequently...
1213 --# We play the exact same game with the (non-standard) class method
1214 --# `fromInt'. Unsurprisingly, the type `Int' is wired into the
1221 ds.d4Qz :: Int{-2i-}
1224 } in fromInt.t4GS ds.d4Qz
1226 --# `I# 2#' is just the literal Int `2'; it reflects the fact that
1227 --# GHC defines `data Int = I# Int#', where Int# is the primitive
1228 --# unboxed type. (see relevant info about unboxed types elsewhere...)
1230 --# The `!' after `I#' indicates that this is a *saturated*
1231 --# application of the `I#' data constructor (i.e., not partially
1234 skip2.t3Ja :: _4 -> [_4]
1238 let { ds.d4QQ :: [_4]
1244 ds.d4QY = +.t4Hg m.r1H4 lit.t4Hb
1245 } in skip2.t3Ja ds.d4QY
1247 :! _4 m.r1H4 ds.d4QQ
1253 (``It's just a simple functional language'' is an unregisterised
1254 trademark of Peyton Jones Enterprises, plc.)
1256 %----------------------------------------------------------------------
1257 \subsubsection[source-file-options]{Command line options in source files}
1258 \index{source-file options}
1260 Sometimes it is useful to make the connection between a source file
1261 and the command-line options it requires quite tight. For instance,
1262 if a (Glasgow) Haskell source file uses \tr{casm}s, the C back-end
1263 often needs to be told about which header files to include. Rather than
1264 maintaining the list of files the source depends on in a
1265 \tr{Makefile} (using the \tr{-#include} command-line option), it is
1266 possible to do this directly in the source file using the \tr{OPTIONS}
1267 pragma \index{OPTIONS pragma}:
1270 {-# OPTIONS -#include "foo.h" #-}
1276 \tr{OPTIONS} pragmas are only looked for at the top of your source
1277 files, upto the first (non-literate,non-empty) line not containing
1278 \tr{OPTIONS}. Multiple \tr{OPTIONS} pragmas are recognised. Note
1279 that your command shell does not get to the source file options, they
1280 are just included literally in the array of command-line arguments
1281 the compiler driver maintains internally, so you'll be desperately
1282 disappointed if you try to glob etc. inside \tr{OPTIONS}.
1284 NOTE: the contents of OPTIONS are prepended to the command-line
1285 options, so you *do* have the ability to override OPTIONS settings
1286 via the command line.
1288 It is not recommended to move all the contents of your Makefiles into
1289 your source files, but in some circumstances, the \tr{OPTIONS} pragma
1290 is the Right Thing. (If you use \tr{-keep-hc-file-too} and have OPTION
1291 flags in your module, the OPTIONS will get put into the generated .hc
1294 %----------------------------------------------------------------------
1295 \subsubsection{How to compile mutually recursive modules}
1296 \index{module system, recursion}
1298 Currently, the compiler does not have proper support for dealing with
1299 mutually recursive modules:
1321 When compiling either module A and B, the compiler will try (in vain)
1322 to look for the interface file of the other. So, to get mutually
1323 recursive modules off the ground, you need to hand write an interface
1324 file for A or B, so as to break the loop. For the example at hand, the
1325 boot interface file for A would like the following:
1332 1 newtype A = A PrelBase.Int ;
1333 1 f _:_ B.B -> A.A ;;
1336 To make sure you get the syntax right, tailoring an existing interface
1337 file is a Good Idea.
1339 {\bf Note:} This is all a temporary solution, a version of the compiler
1340 that handles mutually recursive properly without the manual
1341 construction of interface file, is in the works.
1343 %----------------------------------------------------------------------
1344 %\subsubsection[arity-checking]{Options to insert arity-checking code}
1345 %\index{arity checking}
1347 %The \tr{-darity-checks}\index{-darity-checks option} option inserts
1348 %code to check for arity violations. Unfortunately, it's not that
1349 %simple: you have to link with a prelude that was also built with arity
1350 %checks. If you have one, then great; otherwise...
1352 %The \tr{-darity-checks-C-only}\index{-darity-checks-C-only option}
1353 %option inserts the self-same arity checking code into \tr{.hc} files,
1354 %but doesn't compile it into the \tr{.o} files. We use this flag with
1355 %the \tr{-keep-hc-file-too}\index{-keep-hc-file-too option}, where we
1356 %are keeping \tr{.hc} files around for debugging purposes.
1358 %----------------------------------------------------------------------
1359 %\subsubsection[omit-checking]{Options to omit checking code}
1360 %\index{omitting runtime checks}
1362 %By default, the GHC system emits all possible not-too-expensive
1363 %runtime checking code. If you are brave or experimenting, you might
1364 %want to turn off some of this (not recommended):
1366 %\begin{tabular}{ll}
1367 %-dno-black-holing & won't buy you much (even if it works) \\
1368 %-dno-updates & you're crazy if you do this \\
1369 %-dno-stk-stubbing & omit stack stubbing (NOT DONE YET) \\
1371 %\index{-dno-black-holing option}%
1372 %\index{-dno-updates option}%
1373 %\index{-dno-stk-stubbing option}
1375 %Warning: all very lightly tested, if at all...
1377 %% %************************************************************************
1379 %% \subsection[options-GC]{Choosing a garbage collector}
1381 %% %************************************************************************
1383 %% (Note: you need a Good Reason before launching into this territory.)
1385 %% There are up to four garbage collectors to choose from (it depends how
1386 %% your local system was built); the Appel-style generational collector
1389 %% If you choose a non-default collector, you must specify it both when
1390 %% compiling the modules and when linking them together into an
1391 %% executable. Also, the native-code generator only works with the
1392 %% default collector (a small point to bear in mind).
1394 %% \begin{description}
1395 %% \item[\tr{-gc-ap} option:]
1396 %% \index{-gc-ap option}
1397 %% Appel-like generational collector (the default).
1399 %% \item[\tr{-gc-2s} option:]
1400 %% \index{-gc-2s option}
1401 %% Two-space copying collector.
1403 %% \item[\tr{-gc-1s} option:]
1404 %% \index{-gc-1s option}
1405 %% One-space compacting collector.
1407 %% \item[\tr{-gc-du} option:]
1408 %% \index{-gc-du option}
1409 %% Dual-mode collector (swaps between copying and compacting).
1410 %% \end{description}