1 <!DOCTYPE Article PUBLIC "-//OASIS//DTD DocBook V3.1//EN">
3 <Article id="building-guide">
7 <Title>Building the Glasgow Functional Programming Tools Suite</Title>
8 <Author><OtherName>The GHC Team</OtherName></Author>
9 <Address><Email>glasgow-haskell-{users,bugs}@haskell.org</Email></Address>
10 <PubDate>November 2001</PubDate>
13 <para>The Glasgow fptools suite is a collection of Functional
14 Programming related tools, including the Glasgow Haskell
15 Compiler (GHC). The source code for the whole suite is kept in
16 a single CVS repository and shares a common build and
17 installation system.</para>
19 <para>This guide is intended for people who want to build or
20 modify programs from the Glasgow <Literal>fptools</Literal>
21 suite (as distinct from those who merely want to
22 <Emphasis>run</Emphasis> them). Installation instructions are
23 now provided in the user guide.</para>
25 <para>The bulk of this guide applies to building on Unix
26 systems; see <XRef LinkEnd="winbuild"> for Windows notes.</para>
32 <sect1 id="sec-getting">
33 <title>Getting the sources</title>
35 <para>You can get your hands on the <literal>fptools</literal>
41 <term><indexterm><primary>Source
42 distributions</primary></indexterm>Source distributions</term>
44 <para>You have a supported platform, but (a) you like
45 the warm fuzzy feeling of compiling things yourself;
46 (b) you want to build something ``extra”—e.g., a
47 set of libraries with strictness-analysis turned off; or
48 (c) you want to hack on GHC yourself.</para>
50 <para>A source distribution contains complete sources for
51 one or more projects in the <literal>fptools</literal>
52 suite. Not only that, but the more awkward
53 machine-independent steps are done for you. For example, if
55 <command>happy</command><indexterm><primary>happy</primary></indexterm>
56 you'll find it convenient that the source distribution
57 contains the result of running <command>happy</command> on
58 the parser specifications. If you don't want to alter the
59 parser then this saves you having to find and install
60 <command>happy</command>. You will still need a working
61 version of GHC (preferably version 4.08+) on your machine in
62 order to compile (most of) the sources, however.</para>
67 <term>The CVS repository.</term>
68 <indexterm><primary>CVS repository</primary>
71 <para>We make releases infrequently. If you want more
72 up-to-the minute (but less tested) source code then you need
73 to get access to our CVS repository.</para>
75 <para>All the <literal>fptools</literal> source code is held
76 in a CVS repository. CVS is a pretty good source-code
77 control system, and best of all it works over the
80 <para>The repository holds source code only. It holds no
81 mechanically generated files at all. So if you check out a
82 source tree from CVS you will need to install every utility
83 so that you can build all the derived files from
86 <para>More information about our CVS repository can be found
87 in <xref linkend="sec-cvs">.</para>
92 <para>If you are going to do any building from sources (either
93 from a source distribution or the CVS repository) then you need to
94 read all of this manual in detail.</para>
98 <title>Using the CVS repository</title>
100 <para>We use <ulink url="http://www.cvshome.org/">CVS</ulink> (Concurrent Version System) to keep track of our
101 sources for various software projects. CVS lets several people
102 work on the same software at the same time, allowing changes to be
103 checked in incrementally. </para>
105 <para>This section is a set of guidelines for how to use our CVS
106 repository, and will probably evolve in time. The main thing to
107 remember is that most mistakes can be undone, but if there's
108 anything you're not sure about feel free to bug the local CVS
109 meister (namely Jeff Lewis
110 <email>jlewis@galconn.com</email>). </para>
112 <sect2 id="cvs-access">
113 <title>Getting access to the CVS Repository</title>
115 <para>You can access the repository in one of two ways:
116 read-only (<xref linkend="cvs-read-only">), or read-write (<xref
117 linkend="cvs-read-write">).</para>
119 <sect3 id="cvs-read-only">
120 <title>Remote Read-only CVS Access</title>
122 <para>Read-only access is available to anyone - there's no
123 need to ask us first. With read-only CVS access you can do
124 anything except commit changes to the repository. You can
125 make changes to your local tree, and still use CVS's merge
126 facility to keep your tree up to date, and you can generate
127 patches using 'cvs diff' in order to send to us for
130 <para>To get read-only access to the repository:</para>
134 <para>Make sure that <application>cvs</application> is
135 installed on your machine.</para>
138 <para>Set your <literal>$CVSROOT</literal> environment variable to
139 <literal>:pserver:anoncvs@glass.cse.ogi.edu:/cvs</literal></para>
142 <para>Run the command</para>
146 <para>The password is simply <literal>cvs</literal>. This
147 sets up a file in your home directory called
148 <literal>.cvspass</literal>, which squirrels away the
149 dummy password, so you only need to do this step once.</para>
153 <para>Now go to <xref linkend="cvs-first">.</para>
158 <sect3 id="cvs-read-write">
159 <title>Remote Read-Write CVS Access</title>
161 <para>We generally supply read-write access to folk doing
162 serious development on some part of the source tree, when
163 going through us would be a pain. If you're developing some
164 feature, or think you have the time and inclination to fix
165 bugs in our sources, feel free to ask for read-write
166 access. There is a certain amount of responsibility that goes
167 with commit privileges; we are more likely to grant you access
168 if you've demonstrated your competence by sending us patches
169 via mail in the past.</para>
171 <para>To get remote read-write CVS access, you need to do the
172 following steps.</para>
176 <para>Make sure that <literal>cvs</literal> and
177 <literal>ssh</literal> are both installed on your
182 <para>Generate a DSA private-key/public-key pair, thus:</para>
186 <para>(<literal>ssh-keygen</literal> comes with
187 <literal>ssh</literal>.) Running <literal>ssh-keygen
188 -d</literal> creates the private and public keys in
189 <literal>$HOME/.ssh/id_dsa</literal> and
190 <literal>$HOME/.ssh/id_dsa.pub</literal> respectively
191 (assuming you accept the standard defaults).</para>
193 <para><literal>ssh-keygen -d</literal> will only work if
194 you have Version 2 <literal>ssh</literal> installed; it
195 will fail harmlessly otherwise. If you only have Version
196 1 you can instead generate an RSA key pair using plain</para>
201 <para>Doing so creates the private and public RSA keys in
202 <literal>$HOME/.ssh/identity</literal> and
203 <literal>$HOME/.ssh/identity.pub</literal>
206 <para>[Deprecated.] Incidentally, you can force a Version
207 2 <literal>ssh</literal> to use the Version 1 protocol by
208 creating <literal>$HOME/config</literal> with the
209 following in it:</para>
217 <para>In both cases, <literal>ssh-keygen</literal> will
218 ask for a <firstterm>passphrase</firstterm>. The
219 passphrase is a password that protects your private key.
220 In response to the 'Enter passphrase' question, you can
224 <para>[Recommended.] Enter a passphrase, which you
225 will quote each time you use CVS.
226 <literal>ssh-agent</literal> makes this entirely
230 <para>[Deprecated.] Just hit return (i.e. use an empty
231 passphrase); then you won't need to quote the
232 passphrase when using CVS. The downside is that
233 anyone who can see into your <literal>.ssh</literal>
234 directory, and thereby get your private key, can mess
235 up the repository. So you must keep the
236 <literal>.ssh</literal> directory with draconian
237 no-access permissions.</para>
243 <emphasis>Windows users: see the notes in <xref linkend="configure-ssh"> about <command>ssh</command> wrinkles!</emphasis>
250 <para>Send a message to to the CVS repository
251 administrator (currently Jeff Lewis
252 <email>jeff@galconn.com</email>), containing:</para>
255 <para>Your desired user-name.</para>
258 <para>Your <literal>.ssh/id_dsa.pub</literal> (or
259 <literal>.ssh/identity.pub</literal>).</para>
262 <para>He will set up your account.</para>
266 <para>Set the following environment variables:</para>
270 <constant>$HOME</constant>: points to your home directory. This is where CVS
271 will look for its <filename>.cvsrc</filename> file.
277 <constant>$CVS_RSH</constant> to <filename>ssh</filename>
279 <para>[Windows users.] Setting your <literal>CVS_RSH</literal> to
280 <literal>ssh</literal> assumes that your CVS client
281 understands how to execute shell script
282 ("#!"s,really), which is what
283 <literal>ssh</literal> is. This may not be the case on
284 Win32 platforms, so in that case set <literal>CVS_RSH</literal> to
285 <literal>ssh1</literal>.</para>
289 <para><literal>$CVSROOT</literal> to
290 <literal>:ext:</literal><replaceable>your-username</replaceable>
291 <literal>@cvs.haskell.org:/home/cvs/root</literal>
292 where <replaceable>your-username</replaceable> is your user name on
293 <literal>cvs.haskell.org</literal>.
295 <para>The <literal>CVSROOT</literal> environment variable will
296 be recorded in the checked-out tree, so you don't need to set
297 this every time. </para>
303 <constant>$CVSEDITOR</constant>: <filename>bin/gnuclient.exe</filename>
304 if you want to use an Emacs buffer for typing in those long commit messages.
310 <constant>$SHELL</constant>: To use bash as the shell in Emacs, you need to
311 set this to point to <filename>bash.exe</filename>.
322 Put the following in <filename>$HOME/.cvsrc</filename>:
333 These are the default options for the specified CVS commands,
334 and represent better defaults than the usual ones. (Feel
335 free to change them.)
339 [Windows users.] Filenames starting with <filename>.</filename> were illegal in
340 the 8.3 DOS filesystem, but that restriction should have
341 been lifted by now (i.e., you're using VFAT or later filesystems.) If
342 you're still having problems creating it, don't worry; <filename>.cvsrc</filename> is entirely
350 <para>[Experts.] Once your account is set up, you can get
351 access from other machines without bothering Jeff, thus:</para>
354 <para>Generate a public/private key pair on the new
358 <para>Use ssh to log in to
359 <literal>cvs.haskell.org</literal>, from your old
363 <para>Add the public key for the new machine to the file
364 <literal>$HOME/ssh/authorized_keys</literal> on
365 <literal>cvs.haskell.org</literal>.
366 (<literal>authorized_keys2</literal>, I think, for Version
370 <para>Make sure that the new version of
371 <literal>authorized_keys</literal> still has 600 file
380 <sect2 id="cvs-first">
381 <title>Checking Out a Source Tree</title>
385 <para>Make sure you set your <literal>CVSROOT</literal>
386 environment variable according to either of the remote
387 methods above. The Approved Way to check out a source tree
388 is as follows:</para>
391 $ cvs checkout fpconfig
394 <para>At this point you have a new directory called
395 <literal>fptools</literal> which contains the basic stuff
396 for the fptools suite, including the configuration files and
397 some other junk. </para>
399 <para>[Windows users.] The following messages appear to be harmless:
401 setsockopt IPTOS_LOWDELAY: Invalid argument
402 setsockopt IPTOS_THROUGHPUT: Invalid argument
407 <para>You can call the fptools directory whatever you like,
408 CVS won't mind: </para>
411 $ mv fptools <replaceable>directory</replaceable>
414 <para> NB: after you've read the CVS manual you might be
415 tempted to try</para>
417 $ cvs checkout -d <replaceable>directory</replaceable> fpconfig
420 <para>instead of checking out <literal>fpconfig</literal>
421 and then renaming it. But this doesn't work, and will
422 result in checking out the entire repository instead of just
423 the <literal>fpconfig</literal> bit.</para>
425 $ cd <replaceable>directory</replaceable>
426 $ cvs checkout ghc hslibs libraries
429 <para>The second command here checks out the relevant
430 modules you want to work on. For a GHC build, for instance,
431 you need at least the <literal>ghc</literal>,
432 <literal>hslibs</literal> and <literal>libraries</literal>
433 modules (for a full list of the projects available, see
434 <xref linkend="projects">).</para>
436 <para>Remember that if you do not have
437 <literal>happy</literal> and/or <literal>Alex</literal>
438 installed, you need to check them out as well.</para>
443 <sect2 id="cvs-committing">
444 <title>Committing Changes</title>
446 <para>This is only if you have read-write access to the
447 repository. For anoncvs users, CVS will issue a "read-only
448 repository" error if you try to commit changes.</para>
452 <para>Build the software, if necessary. Unless you're just
453 working on documentation, you'll probably want to build the
454 software in order to test any changes you make.</para>
458 <para>Make changes. Preferably small ones first.</para>
462 <para>Test them. You can see exactly what changes you've
463 made by using the <literal>cvs diff</literal> command:</para>
467 <para>lists all the changes (using the
468 <literal>diff</literal> command) in and below the current
469 directory. In emacs, <literal>C-c C-v =</literal> runs
470 <literal>cvs diff</literal> on the current buffer and shows
471 you the results.</para>
475 <para>If you changed something in the
476 <literal>fptools/libraries</literal> subdirectories, also run
477 <literal>make html</literal> to check if the documentation can
478 be generated successfully, too.</para>
482 <para>Before checking in a change, you need to update your
489 <para>This pulls in any changes that other people have made,
490 and merges them with yours. If there are any conflicts, CVS
491 will tell you, and you'll have to resolve them before you
492 can check your changes in. The documentation describes what
493 to do in the event of a conflict.</para>
495 <para>It's not always necessary to do a full cvs update
496 before checking in a change, since CVS will always tell you
497 if you try to check in a file that someone else has changed.
498 However, you should still update at regular intervals to
499 avoid making changes that don't work in conjuction with
500 changes that someone else made. Keeping an eye on what goes
501 by on the mailing list can help here.</para>
505 <para>When you're happy that your change isn't going to
506 break anything, check it in. For a one-file change:</para>
509 $ cvs commit <replaceable>filename</replaceable>
512 <para>CVS will then pop up an editor for you to enter a
513 "commit message", this is just a short description
514 of what your change does, and will be kept in the history of
517 <para>If you're using emacs, simply load up the file into a
518 buffer and type <literal>C-x C-q</literal>, and emacs will
519 prompt for a commit message and then check in the file for
522 <para>For a multiple-file change, things are a bit
523 trickier. There are several ways to do this, but this is the
524 way I find easiest. First type the commit message into a
525 temporary file. Then either</para>
528 $ cvs commit -F <replaceable>commit-message</replaceable> <replaceable>file_1</replaceable> .... <replaceable>file_n</replaceable>
531 <para>or, if nothing else has changed in this part of the
535 $ cvs commit -F <replaceable>commit-message</replaceable> <replaceable>directory</replaceable>
538 <para>where <replaceable>directory</replaceable> is a common
539 parent directory for all your changes, and
540 <replaceable>commit-message</replaceable> is the name of the
541 file containing the commit message.</para>
543 <para>Shortly afterwards, you'll get some mail from the
544 relevant mailing list saying which files changed, and giving
545 the commit message. For a multiple-file change, you should
546 still get only <emphasis>one</emphasis> message.</para>
551 <sect2 id="cvs-update">
552 <title>Updating Your Source Tree</title>
554 <para>It can be tempting to cvs update just part of a source
555 tree to bring in some changes that someone else has made, or
556 before committing your own changes. This is NOT RECOMMENDED!
557 Quite often changes in one part of the tree are dependent on
558 changes in another part of the tree (the
559 <literal>mk/*.mk</literal> files are a good example where
560 problems crop up quite often). Having an inconsistent tree is a
561 major cause of headaches. </para>
563 <para>So, to avoid a lot of hassle, follow this recipe for
564 updating your tree:</para>
568 $ cvs update -P 2>&1 | tee log</screen>
570 <para>Look at the log file, and fix any conflicts (denoted by a
571 <quote>C</quote> in the first column). New directories may have
572 appeared in the repository; CVS doesn't check these out by
573 default, so to get new directories you have to explicitly do
575 $ cvs update -d</screen>
576 in each project subdirectory. Don't do this at the top level,
577 because then <emphasis>all</emphasis> the projects will be
580 <para>If you're using multiple build trees, then for every build
581 tree you have pointing at this source tree, you need to update
582 the links in case any new files have appeared: </para>
585 $ cd <replaceable>build-tree</replaceable>
586 $ lndir <replaceable>source-tree</replaceable>
589 <para>Some files might have been removed, so you need to remove
590 the links pointing to these non-existent files:</para>
593 $ find . -xtype l -exec rm '{}' \;
596 <para>To be <emphasis>really</emphasis> safe, you should do
599 <screen>$ gmake all</screen>
601 <para>from the top-level, to update the dependencies and build
602 any changed files. </para>
605 <sect2 id="cvs-tags">
606 <title>GHC Tag Policy</title>
608 <para>If you want to check out a particular version of GHC,
609 you'll need to know how we tag versions in the repository. The
610 policy (as of 4.04) is:</para>
614 <para>The tree is branched before every major release. The
615 branch tag is <literal>ghc-x-xx-branch</literal>, where
616 <literal>x-xx</literal> is the version number of the release
617 with the <literal>'.'</literal> replaced by a
618 <literal>'-'</literal>. For example, the 4.04 release lives
619 on <literal>ghc-4-04-branch</literal>.</para>
623 <para>The release itself is tagged with
624 <literal>ghc-x-xx</literal> (on the branch). eg. 4.06 is
625 called <literal>ghc-4-06</literal>.</para>
629 <para>We didn't always follow these guidelines, so to see
630 what tags there are for previous versions, do <literal>cvs
631 log</literal> on a file that's been around for a while (like
632 <literal>fptools/ghc/README</literal>).</para>
636 <para>So, to check out a fresh GHC 4.06 tree you would
640 $ cvs co -r ghc-4-06 fpconfig
642 $ cvs co -r ghc-4-06 ghc hslibs
646 <sect2 id="cvs-hints">
647 <title>General Hints</title>
651 <para>As a general rule: commit changes in small units,
652 preferably addressing one issue or implementing a single
653 feature. Provide a descriptive log message so that the
654 repository records exactly which changes were required to
655 implement a given feature/fix a bug. I've found this
656 <emphasis>very</emphasis> useful in the past for finding out
657 when a particular bug was introduced: you can just wind back
658 the CVS tree until the bug disappears.</para>
662 <para>Keep the sources at least *buildable* at any given
663 time. No doubt bugs will creep in, but it's quite easy to
664 ensure that any change made at least leaves the tree in a
665 buildable state. We do nightly builds of GHC to keep an eye
666 on what things work/don't work each day and how we're doing
667 in relation to previous verions. This idea is truely wrecked
668 if the compiler won't build in the first place!</para>
672 <para>To check out extra bits into an already-checked-out
673 tree, use the following procedure. Suppose you have a
674 checked-out fptools tree containing just ghc, and you want
675 to add nofib to it:</para>
686 $ cvs update -d nofib
689 <para>(the -d flag tells update to create a new
690 directory). If you just want part of the nofib suite, you
695 $ cvs checkout nofib/spectral
698 <para>This works because <literal>nofib</literal> is a
699 module in its own right, and spectral is a subdirectory of
700 the nofib module. The path argument to checkout must always
701 start with a module name. There's no equivalent form of this
702 command using <literal>update</literal>.</para>
708 <sect1 id="projects">
709 <title>What projects are there?</title>
711 <para>The <literal>fptools</literal> suite consists of several
712 <firstterm>projects</firstterm>, most of which can be downloaded,
713 built and installed individually. Each project corresponds to a
714 subdirectory in the source tree, and if checking out from CVS then
715 each project can be checked out individually by sitting in the top
716 level of your source tree and typing <command>cvs checkout
717 <replaceable>project</replaceable></command>.</para>
719 <para>Here is a list of the projects currently available:</para>
723 <term><literal>alex</literal></term>
724 <indexterm><primary><literal>alex</literal></primary>
725 <secondary>project</secondary></indexterm>
728 url="http://www.haskell.org/alex/">Alex</ulink> lexical
729 analyser generator for Haskell.</para>
734 <term><literal>ghc</literal></term>
735 <indexterm><primary><literal>ghc</literal></primary>
736 <secondary>project</secondary></indexterm>
738 <para>The <ulink url="http://www.haskell.org/ghc/">Glasgow
739 Haskell Compiler</ulink> (minus libraries). Absolutely
740 required for building GHC.</para>
745 <term><literal>glafp-utils</literal></term>
746 <indexterm><primary><literal>glafp-utils</literal></primary><secondary>project</secondary></indexterm>
748 <para>Utility programs, some of which are used by the
749 build/installation system. Required for pretty much
755 <term><literal>greencard</literal></term>
756 <indexterm><primary><literal>greencard</literal></primary><secondary>project</secondary></indexterm>
759 url="http://www.haskell.org/greencard/">GreenCard</ulink>
760 system for generating Haskell foreign function
766 <term><literal>haggis</literal></term>
767 <indexterm><primary><literal>haggis</literal></primary><secondary>project</secondary></indexterm>
770 url="http://www.dcs.gla.ac.uk/fp/software/haggis/">Haggis</ulink>
771 Haskell GUI framework.</para>
776 <term><literal>haddock</literal></term>
777 <indexterm><primary><literal>haddock</literal></primary><secondary>project</secondary></indexterm>
780 url="http://www.haskell.org/haddock/">Haddock</ulink>
781 documentation tool.</para>
786 <term><literal>happy</literal></term>
787 <indexterm><primary><literal>happy</literal></primary><secondary>project</secondary></indexterm>
790 url="http://www.haskell.org/happy/">Happy</ulink> Parser
796 <term><literal>hdirect</literal></term>
797 <indexterm><primary><literal>hdirect</literal></primary><secondary>project</secondary></indexterm>
800 url="http://www.haskell.org/hdirect/">H/Direct</ulink>
801 Haskell interoperability tool.</para>
806 <term><literal>hood</literal></term>
807 <indexterm><primary><literal>hood</literal></primary><secondary>project</secondary></indexterm>
809 <para>The <ulink url="http://www.haskell.org/hood/">Haskell
810 Object Observation Debugger</ulink>.</para>
815 <term><literal>hslibs</literal></term>
816 <indexterm><primary><literal>hslibs</literal></primary><secondary>project</secondary></indexterm>
818 <para>Supplemental libraries for GHC
819 (<emphasis>required</emphasis> for building GHC).</para>
824 <term><literal>libraries</literal></term>
825 <indexterm><primary><literal></literal></primary><secondary>project</secondary></indexterm>
827 <para>Hierarchical Haskell library suite
828 (<emphasis>required</emphasis> for building GHC).</para>
833 <term><literal>mhms</literal></term>
834 <indexterm><primary><literal></literal></primary><secondary>project</secondary></indexterm>
836 <para>The Modular Haskell Metric System.</para>
841 <term><literal>nofib</literal></term>
842 <indexterm><primary><literal>nofib</literal></primary><secondary>project</secondary></indexterm>
844 <para>The NoFib suite: A collection of Haskell programs used
845 primarily for benchmarking.</para>
850 <term><literal>testsuite</literal></term>
851 <indexterm><primary><literal>testsuite</literal></primary><secondary>project</secondary></indexterm>
853 <para>A testing framework, including GHC's regression test
859 <para>So, to build GHC you need at least the
860 <literal>ghc</literal>, <literal>libraries</literal> and
861 <literal>hslibs</literal> projects (a GHC source distribution will
862 already include the bits you need).</para>
865 <sect1 id="sec-build-checks">
866 <title>Things to check before you start</title>
868 <para>Here's a list of things to check before you get
874 <indexterm><primary>Disk space needed</primary></indexterm>
875 <para>Disk space needed: from about 100Mb for a basic GHC
876 build, up to probably 500Mb for a GHC build with everything
877 included (libraries built several different ways,
882 <para>Use an appropriate machine / operating system. <xref
883 linkend="sec-port-info"> lists the supported platforms; if
884 yours isn't amongst these then you can try porting GHC (see
885 <xref linkend="sec-porting-ghc">).</para>
889 <para>Be sure that the “pre-supposed” utilities are
890 installed. <Xref LinkEnd="sec-pre-supposed">
895 <para>If you have any problem when building or installing the
896 Glasgow tools, please check the “known pitfalls” (<Xref
897 LinkEnd="sec-build-pitfalls">). Also check the FAQ for the
898 version you're building, which is part of the User's Guide and
899 available on the <ulink URL="http://www.haskell.org/ghc/" >GHC web
902 <indexterm><primary>bugs</primary><secondary>known</secondary></indexterm>
904 <para>If you feel there is still some shortcoming in our
905 procedure or instructions, please report it.</para>
907 <para>For GHC, please see the <ulink
908 url="http://www.haskell.org/ghc/docs/latest/set/bug-reporting.html">bug-reporting
909 section of the GHC Users' Guide</ulink>, to maximise the
910 usefulness of your report.</para>
912 <indexterm><primary>bugs</primary><secondary>seporting</secondary></indexterm>
913 <para>If in doubt, please send a message to
914 <email>glasgow-haskell-bugs@haskell.org</email>.
915 <indexterm><primary>bugs</primary><secondary>mailing
916 list</secondary></indexterm></para>
921 <sect1 id="sec-port-info">
922 <title>What machines the Glasgow tools run on</title>
924 <indexterm><primary>ports</primary><secondary>GHC</secondary></indexterm>
925 <indexterm><primary>GHC</primary><secondary>ports</secondary></indexterm>
926 <indexterm><primary>platforms</primary><secondary>supported</secondary></indexterm>
928 <para>The main question is whether or not the Haskell compiler
929 (GHC) runs on your platform.</para>
931 <para>A “platform” is a
932 architecture/manufacturer/operating-system combination, such as
933 <literal>sparc-sun-solaris2</literal>. Other common ones are
934 <literal>alpha-dec-osf2</literal>,
935 <literal>hppa1.1-hp-hpux9</literal>,
936 <literal>i386-unknown-linux</literal>,
937 <literal>i386-unknown-solaris2</literal>,
938 <literal>i386-unknown-freebsd</literal>,
939 <literal>i386-unknown-cygwin32</literal>,
940 <literal>m68k-sun-sunos4</literal>,
941 <literal>mips-sgi-irix5</literal>,
942 <literal>sparc-sun-sunos4</literal>,
943 <literal>sparc-sun-solaris2</literal>,
944 <literal>powerpc-ibm-aix</literal>.</para>
946 <para>Some libraries may only work on a limited number of
947 platforms; for example, a sockets library is of no use unless the
948 operating system supports the underlying BSDisms.</para>
951 <title>What platforms the Haskell compiler (GHC) runs on</title>
953 <indexterm><primary>fully-supported platforms</primary></indexterm>
954 <indexterm><primary>native-code generator</primary></indexterm>
955 <indexterm><primary>registerised ports</primary></indexterm>
956 <indexterm><primary>unregisterised ports</primary></indexterm>
958 <para>The GHC hierarchy of Porting Goodness: (a) Best is a
959 native-code generator; (b) next best is a
960 “registerised” port; (c) the bare minimum is an
961 “unregisterised” port.
962 (“Unregisterised” is so terrible that we won't say
963 more about it).</para>
965 <para>We use Sparcs running Solaris 2.7 and x86 boxes running
966 FreeBSD and Linux, so those are the best supported platforms,
967 unsurprisingly.</para>
969 <para>Here's everything that's known about GHC ports. We
970 identify platforms by their “canonical”
971 CPU/Manufacturer/OS triple.</para>
975 <term>alpha-dec-{osf,linux,freebsd,openbsd,netbsd}:</term>
976 <indexterm><primary>alpha-dec-osf</primary></indexterm>
977 <indexterm><primary>alpha-dec-linux</primary></indexterm>
978 <indexterm><primary>alpha-dec-freebsd</primary></indexterm>
979 <indexterm><primary>alpha-dec-openbsd</primary></indexterm>
980 <indexterm><primary>alpha-dec-netbsd</primary></indexterm>
983 <para>The OSF port is currently working (as of GHC version
984 5.02.1) and well supported. The native code generator is
985 currently non-working. Other operating systems will
986 require some minor porting.</para>
991 <term>sparc-sun-sunos4</term>
992 <indexterm><primary>sparc-sun-sunos4</primary></indexterm>
994 <para>Probably works with minor tweaks, hasn't been tested
1000 <term>sparc-sun-solaris2</term>
1001 <indexterm><primary>sparc-sun-solaris2</primary></indexterm>
1003 <para>Fully supported (at least for Solaris 2.7 and 2.6),
1004 including native-code generator.</para>
1009 <term>sparc-unknown-openbsd</term>
1010 <indexterm><primary>sparc-unknown-openbsd</primary></indexterm>
1012 <para>Supported, including native-code generator. The
1013 same should also be true of NetBSD</para>
1018 <term>hppa1.1-hp-hpux (HP-PA boxes running HPUX 9.x)</term>
1019 <indexterm><primary>hppa1.1-hp-hpux</primary></indexterm>
1021 <para>A registerised port is available for version 4.08,
1022 but GHC hasn't been built on that platform since (as far
1023 as we know). No native-code generator.</para>
1028 <term>i386-unknown-linux (PCs running Linux, ELF binary format)</term>
1029 <indexterm><primary>i386-*-linux</primary></indexterm>
1031 <para>GHC works registerised and has a native code
1032 generator. You <Emphasis>must</Emphasis> have GCC 2.7.x
1033 or later. NOTE about <literal>glibc</literal> versions:
1034 GHC binaries built on a system running <literal>glibc
1035 2.0</literal> won't work on a system running
1036 <literal>glibc 2.1</literal>, and vice versa. In general,
1037 don't expect compatibility between
1038 <literal>glibc</literal> versions, even if the shared
1039 library version hasn't changed.</para>
1044 <term>i386-unknown-freebsd (PCs running FreeBSD 2.2 or
1046 <indexterm><primary>i386-unknown-freebsd</primary></indexterm>
1048 <para>GHC works registerised. Pre-built packages are
1049 available in the native package format, so if you just
1050 need binaries you're better off just installing the
1051 package (it might even be on your installation
1057 <term>i386-unknown-openbsd (PCs running OpenBSD)</term>
1058 <indexterm><primary>i386-unknown-openbsd</primary></indexterm>
1060 <para>Supported, with native code generator. Packages are
1061 available through the ports system in the native package
1067 <term>i386-unknown-netbsd (PCs running NetBSD and
1069 <indexterm><primary>i386-unknown-netbsd</primary></indexterm>
1071 <para>Will require some minor porting effort, but should
1072 work registerised.</para>
1077 <term>i386-unknown-mingw32 (PCs running Windows)</term>
1078 <indexterm><primary>i386-unknown-mingw32</primary></indexterm>
1080 <para>Fully supported under Win9x, WinNT, Win2k, and
1081 WinXP. Includes a native code generator. Building from
1082 source requires a recent <ulink
1083 url="http://www.cygwin.com/">Cygwin</ulink> distribution
1084 to be installed.</para>
1089 <term>ia64-unknown-linux</term>
1090 <indexterm><primary>ia64-unknown-linux</primary></indexterm>
1092 <para>Supported, except there is no native code
1098 <term>x86_64-unknown-linux</term>
1099 <indexterm><primary>x86_64-unknown-linux</primary></indexterm>
1101 <para>GHC currently works unregisterised. A registerised
1102 port is in progress.</para>
1107 <term>mips-sgi-irix5</term>
1108 <indexterm><primary>mips-sgi-irix[5-6]</primary></indexterm>
1110 <para>Port has worked in the past, but hasn't been tested
1111 for some time (and will certainly have rotted in various
1112 ways). As usual, we don't have access to machines and
1113 there hasn't been an overwhelming demand for this port,
1114 but feel free to get in touch.</para>
1119 <term>powerpc-ibm-aix</term>
1120 <indexterm><primary>powerpc-ibm-aix</primary></indexterm>
1122 <para>Port currently doesn't work, needs some minimal
1123 porting effort. As usual, we don't have access to
1124 machines and there hasn't been an overwhelming demand for
1125 this port, but feel free to get in touch.</para>
1130 <term>powerpc-apple-darwin</term>
1131 <indexterm><primary>powerpc-apple-darwin</primary></indexterm>
1133 <para>Supported registerised. Native code generator is
1134 almost working.</para>
1139 <term>powerpc-apple-linux</term>
1140 <indexterm><primary>powerpc-apple-linux</primary></indexterm>
1142 <para>Not supported (yet).</para>
1147 <para>Various other systems have had GHC ported to them in the
1148 distant past, including various Motorola 68k boxes. The 68k
1149 support still remains, but porting to one of these systems will
1150 certainly be a non-trivial task.</para>
1154 <title>What machines the other tools run on</title>
1156 <para>Unless you hear otherwise, the other tools work if GHC
1162 <sect1 id="sec-pre-supposed">
1163 <title>Installing pre-supposed utilities</title>
1165 <indexterm><primary>pre-supposed utilities</primary></indexterm>
1166 <indexterm><primary>utilities, pre-supposed</primary></indexterm>
1168 <para>Here are the gory details about some utility programs you
1169 may need; <command>perl</command>, <command>gcc</command> and
1170 <command>happy</command> are the only important
1171 ones. (PVM<indexterm><primary>PVM</primary></indexterm> is
1172 important if you're going for Parallel Haskell.) The
1173 <command>configure</command><indexterm><primary>configure</primary></indexterm>
1174 script will tell you if you are missing something.</para>
1180 <indexterm><primary>pre-supposed: GHC</primary></indexterm>
1181 <indexterm><primary>GHC, pre-supposed</primary></indexterm>
1183 <para>GHC is required to build many of the tools, including
1184 GHC itself. If you need to port GHC to your platform
1185 because there isn't a binary distribution of GHC available,
1186 then see <xref linkend="sec-porting-ghc">.</para>
1188 <para>Which version of GHC you need will depend on the
1189 packages you intend to build. GHC itself will normally
1190 build using one of several older versions of itself - check
1191 the announcement or release notes for details.</para>
1197 <indexterm><primary>pre-supposed: Perl</primary></indexterm>
1198 <indexterm><primary>Perl, pre-supposed</primary></indexterm>
1200 <para><emphasis>You have to have Perl to proceed!</emphasis>
1201 Perl version 5 at least is required. GHC has been known to
1202 tickle bugs in Perl, so if you find that Perl crashes when
1203 running GHC try updating (or downgrading) your Perl
1204 installation. Versions of Perl that we use and are known to
1205 be fairly stable are 5.005 and 5.6.1.</para>
1207 <para>For Win32 platforms, you should use the binary
1208 supplied in the InstallShield (copy it to
1209 <filename>/bin</filename>). The Cygwin-supplied Perl seems
1212 <para>Perl should be put somewhere so that it can be invoked
1213 by the <literal>#!</literal> script-invoking
1214 mechanism. The full pathname may need to be less than 32
1215 characters long on some systems.</para>
1220 <term>GNU C (<command>gcc</command>)</term>
1221 <indexterm><primary>pre-supposed: GCC (GNU C
1222 compiler)</primary></indexterm> <indexterm><primary>GCC (GNU C
1223 compiler), pre-supposed</primary></indexterm>
1225 <para>We recommend using GCC version 2.95.2 on all
1226 platforms. Failing that, version 2.7.2 is stable on most
1227 platforms. Earlier versions of GCC can be assumed not to
1228 work, and versions in between 2.7.2 and 2.95.2 (including
1229 <command>egcs</command>) have varying degrees of stability
1230 depending on the platform.</para>
1232 <para>GCC 3.2 is currently known to have problems building
1233 GHC on Sparc, but is stable on x86.</para>
1235 <para>If your GCC dies with “internal error” on
1236 some GHC source file, please let us know, so we can report
1237 it and get things improved. (Exception: on x86
1238 boxes—you may need to fiddle with GHC's
1239 <option>-monly-N-regs</option> option; see the User's
1245 <term>GNU Make</term>
1246 <indexterm><primary>make</primary><secondary>GNU</secondary>
1249 <para>The fptools build system makes heavy use of features
1250 specific to GNU <command>make</command>, so you must have
1251 this installed in order to build any of the fptools
1258 <indexterm><primary>Happy</primary></indexterm>
1260 <para>Happy is a parser generator tool for Haskell, and is
1261 used to generate GHC's parsers. Happy is written in
1262 Haskell, and is a project in the CVS repository
1263 (<literal>fptools/happy</literal>). It can be built from
1264 source, but bear in mind that you'll need GHC installed in
1265 order to build it. To avoid the chicken/egg problem,
1266 install a binary distribution of either Happy or GHC to get
1267 started. Happy distributions are available from <ulink
1268 url="http://www.haskell.org/happy/">Happy's Web
1269 Page</ulink>.</para>
1275 <indexterm><primary>Alex</primary></indexterm>
1277 <para>Alex is a lexical-analyser generator for Haskell,
1278 which GHC uses to generate its lexer. Like Happy, Alex is
1279 written in Haskell and is a project in the CVS repository.
1280 Alex distributions are available from <ulink
1281 url="http://www.haskell.org/alex/">Alex's Web
1282 Page</ulink>.</para>
1287 <term>Autoconf</term>
1288 <indexterm><primary>pre-supposed: Autoconf</primary></indexterm>
1289 <indexterm><primary>Autoconf, pre-supposed</primary></indexterm>
1291 <para>GNU Autoconf is needed if you intend to build from the
1292 CVS sources, it is <emphasis>not</emphasis> needed if you
1293 just intend to build a standard source distribution.</para>
1295 <para>Version 2.52 or later of autoconf is required.
1296 NB. vesrion 2.13 will no longer work, as of GHC version
1299 <para>Autoconf builds the <command>configure</command>
1300 script from <filename>configure.ac</filename> and
1301 <filename>aclocal.m4</filename>. If you modify either of
1302 these files, you'll need <command>autoconf</command> to
1303 rebuild <filename>configure</filename>.</para>
1308 <term><command>sed</command></term>
1309 <indexterm><primary>pre-supposed: sed</primary></indexterm>
1310 <indexterm><primary>sed, pre-supposed</primary></indexterm>
1312 <para>You need a working <command>sed</command> if you are
1313 going to build from sources. The build-configuration stuff
1314 needs it. GNU sed version 2.0.4 is no good! It has a bug
1315 in it that is tickled by the build-configuration. 2.0.5 is
1316 OK. Others are probably OK too (assuming we don't create too
1317 elaborate configure scripts.)</para>
1322 <para>One <literal>fptools</literal> project is worth a quick note
1323 at this point, because it is useful for all the others:
1324 <literal>glafp-utils</literal> contains several utilities which
1325 aren't particularly Glasgow-ish, but Occasionally Indispensable.
1326 Like <command>lndir</command> for creating symbolic link
1329 <sect2 id="pre-supposed-gph-tools">
1330 <title>Tools for building parallel GHC (GPH)</title>
1334 <term>PVM version 3:</term>
1335 <indexterm><primary>pre-supposed: PVM3 (Parallel Virtual Machine)</primary></indexterm>
1336 <indexterm><primary>PVM3 (Parallel Virtual Machine), pre-supposed</primary></indexterm>
1338 <para>PVM is the Parallel Virtual Machine on which
1339 Parallel Haskell programs run. (You only need this if you
1340 plan to run Parallel Haskell. Concurrent Haskell, which
1341 runs concurrent threads on a uniprocessor doesn't need
1342 it.) Underneath PVM, you can have (for example) a network
1343 of workstations (slow) or a multiprocessor box
1346 <para>The current version of PVM is 3.3.11; we use 3.3.7.
1347 It is readily available on the net; I think I got it from
1348 <literal>research.att.com</literal>, in
1349 <filename>netlib</filename>.</para>
1351 <para>A PVM installation is slightly quirky, but easy to
1352 do. Just follow the <filename>Readme</filename>
1353 instructions.</para>
1358 <term><command>bash</command>:</term>
1359 <indexterm><primary>bash, presupposed (Parallel Haskell only)</primary></indexterm>
1361 <para>Sadly, the <command>gr2ps</command> script, used to
1362 convert “parallelism profiles” to PostScript,
1363 is written in Bash (GNU's Bourne Again shell). This bug
1364 will be fixed (someday).</para>
1370 <sect2 id="pre-supposed-other-tools">
1371 <title>Other useful tools</title>
1376 <indexterm><primary>pre-supposed: flex</primary></indexterm>
1377 <indexterm><primary>flex, pre-supposed</primary></indexterm>
1379 <para>This is a quite-a-bit-better-than-Lex lexer. Used
1380 to build a couple of utilities in
1381 <literal>glafp-utils</literal>. Depending on your
1382 operating system, the supplied <command>lex</command> may
1383 or may not work; you should get the GNU version.</para>
1388 <para>More tools are required if you want to format the documentation
1389 that comes with GHC and other fptools projects. See <xref
1390 linkend="building-docs">.</para>
1394 <sect1 id="sec-building-from-source">
1395 <title>Building from source</title>
1397 <indexterm><primary>Building from source</primary></indexterm>
1398 <indexterm><primary>Source, building from</primary></indexterm>
1400 <para>You've been rash enough to want to build some of the Glasgow
1401 Functional Programming tools (GHC, Happy, nofib, etc.) from
1402 source. You've slurped the source, from the CVS repository or
1403 from a source distribution, and now you're sitting looking at a
1404 huge mound of bits, wondering what to do next.</para>
1406 <para>Gingerly, you type <command>make</command>. Wrong
1409 <para>This rest of this guide is intended for duffers like me, who
1410 aren't really interested in Makefiles and systems configurations,
1411 but who need a mental model of the interlocking pieces so that
1412 they can make them work, extend them consistently when adding new
1413 software, and lay hands on them gently when they don't
1416 <sect2 id="quick-start">
1417 <title>Quick Start</title>
1419 <para>If you are starting from a source distribution, and just
1420 want a completely standard build, then the following should
1423 <screen>$ ./configure
1428 <para>For GHC, this will do a 2-stage bootstrap build of the
1429 compiler, with profiling libraries, and install the
1432 <para>If you want to do anything at all non-standard, or you
1433 want to do some development, read on...</para>
1436 <sect2 id="sec-source-tree">
1437 <title>Your source tree</title>
1439 <para>The source code is held in your <emphasis>source
1440 tree</emphasis>. The root directory of your source tree
1441 <emphasis>must</emphasis> contain the following directories and
1446 <para><filename>Makefile</filename>: the root
1451 <para><filename>mk/</filename>: the directory that contains
1452 the main Makefile code, shared by all the
1453 <literal>fptools</literal> software.</para>
1457 <para><filename>configure.ac</filename>,
1458 <filename>config.sub</filename>,
1459 <filename>config.guess</filename>: these files support the
1460 configuration process.</para>
1464 <para><filename>install-sh</filename>.</para>
1468 <para>All the other directories are individual
1469 <emphasis>projects</emphasis> of the <literal>fptools</literal>
1470 system—for example, the Glasgow Haskell Compiler
1471 (<literal>ghc</literal>), the Happy parser generator
1472 (<literal>happy</literal>), the <literal>nofib</literal>
1473 benchmark suite, and so on. You can have zero or more of these.
1474 Needless to say, some of them are needed to build others.</para>
1476 <para>The important thing to remember is that even if you want
1477 only one project (<literal>happy</literal>, say), you must have
1478 a source tree whose root directory contains
1479 <filename>Makefile</filename>, <filename>mk/</filename>,
1480 <filename>configure.ac</filename>, and the project(s) you want
1481 (<filename>happy/</filename> in this case). You cannot get by
1482 with just the <filename>happy/</filename> directory.</para>
1486 <title>Build trees</title>
1487 <indexterm><primary>build trees</primary></indexterm>
1488 <indexterm><primary>link trees, for building</primary></indexterm>
1490 <para>If you just want to build the software once on a single
1491 platform, then your source tree can also be your build tree, and
1492 you can skip the rest of this section.</para>
1494 <para>We often want to build multiple versions of our software
1495 for different architectures, or with different options
1496 (e.g. profiling). It's very desirable to share a single copy of
1497 the source code among all these builds.</para>
1499 <para>So for every source tree we have zero or more
1500 <emphasis>build trees</emphasis>. Each build tree is initially
1501 an exact copy of the source tree, except that each file is a
1502 symbolic link to the source file, rather than being a copy of
1503 the source file. There are “standard” Unix
1504 utilities that make such copies, so standard that they go by
1506 <command>lndir</command><indexterm><primary>lndir</primary></indexterm>,
1507 <command>mkshadowdir</command><indexterm><primary>mkshadowdir</primary></indexterm>
1508 are two (If you don't have either, the source distribution
1509 includes sources for the X11
1510 <command>lndir</command>—check out
1511 <filename>fptools/glafp-utils/lndir</filename>). See <Xref
1512 LinkEnd="sec-storysofar"> for a typical invocation.</para>
1514 <para>The build tree does not need to be anywhere near the
1515 source tree in the file system. Indeed, one advantage of
1516 separating the build tree from the source is that the build tree
1517 can be placed in a non-backed-up partition, saving your systems
1518 support people from backing up untold megabytes of
1519 easily-regenerated, and rapidly-changing, gubbins. The golden
1520 rule is that (with a single exception—<XRef
1521 LinkEnd="sec-build-config">) <emphasis>absolutely everything in
1522 the build tree is either a symbolic link to the source tree, or
1523 else is mechanically generated</emphasis>. It should be
1524 perfectly OK for your build tree to vanish overnight; an hour or
1525 two compiling and you're on the road again.</para>
1527 <para>You need to be a bit careful, though, that any new files
1528 you create (if you do any development work) are in the source
1529 tree, not a build tree!</para>
1531 <para>Remember, that the source files in the build tree are
1532 <emphasis>symbolic links</emphasis> to the files in the source
1533 tree. (The build tree soon accumulates lots of built files like
1534 <filename>Foo.o</filename>, as well.) You can
1535 <emphasis>delete</emphasis> a source file from the build tree
1536 without affecting the source tree (though it's an odd thing to
1537 do). On the other hand, if you <emphasis>edit</emphasis> a
1538 source file from the build tree, you'll edit the source-tree
1539 file directly. (You can set up Emacs so that if you edit a
1540 source file from the build tree, Emacs will silently create an
1541 edited copy of the source file in the build tree, leaving the
1542 source file unchanged; but the danger is that you think you've
1543 edited the source file whereas actually all you've done is edit
1544 the build-tree copy. More commonly you do want to edit the
1545 source file.)</para>
1547 <para>Like the source tree, the top level of your build tree
1548 must be (a linked copy of) the root directory of the
1549 <literal>fptools</literal> suite. Inside Makefiles, the root of
1550 your build tree is called
1551 <constant>$(FPTOOLS_TOP)</constant><indexterm><primary>FPTOOLS_TOP</primary></indexterm>.
1552 In the rest of this document path names are relative to
1553 <constant>$(FPTOOLS_TOP)</constant> unless
1554 otherwise stated. For example, the file
1555 <filename>ghc/mk/target.mk</filename> is actually
1556 <filename><constant>$(FPTOOLS_TOP)</constant>/ghc/mk/target.mk</filename>.</para>
1559 <sect2 id="sec-build-config">
1560 <title>Getting the build you want</title>
1562 <para>When you build <literal>fptools</literal> you will be
1563 compiling code on a particular <emphasis>host
1564 platform</emphasis>, to run on a particular <emphasis>target
1565 platform</emphasis> (usually the same as the host
1566 platform)<indexterm><primary>platform</primary></indexterm>.
1567 The difficulty is that there are minor differences between
1568 different platforms; minor, but enough that the code needs to be
1569 a bit different for each. There are some big differences too:
1570 for a different architecture we need to build GHC with a
1571 different native-code generator.</para>
1573 <para>There are also knobs you can turn to control how the
1574 <literal>fptools</literal> software is built. For example, you
1575 might want to build GHC optimised (so that it runs fast) or
1576 unoptimised (so that you can compile it fast after you've
1577 modified it. Or, you might want to compile it with debugging on
1578 (so that extra consistency-checking code gets included) or off.
1581 <para>All of this stuff is called the
1582 <emphasis>configuration</emphasis> of your build. You set the
1583 configuration using a three-step process.</para>
1587 <term>Step 1: get ready for configuration.</term>
1589 <para>NOTE: if you're starting from a source distribution,
1590 rather than CVS sources, you can skip this step.</para>
1592 <para>Change directory to
1593 <constant>$(FPTOOLS_TOP)</constant> and
1595 <command>autoconf</command><indexterm><primary>autoconf</primary></indexterm>
1596 (with no arguments). This GNU program converts
1597 <filename><constant>$(FPTOOLS_TOP)</constant>/configure.ac</filename>
1598 to a shell script called
1599 <filename><constant>$(FPTOOLS_TOP)</constant>/configure</filename>.
1602 <para>Some projects, including GHC, have their own
1603 configure script. If there's an
1604 <constant>$(FPTOOLS_TOP)/<project>/configure.ac</constant>,
1605 then you need to run <command>autoconf</command> in that
1606 directory too.</para>
1608 <para>Both these steps are completely
1609 platform-independent; they just mean that the
1610 human-written file (<filename>configure.ac</filename>) can
1611 be short, although the resulting shell script,
1612 <command>configure</command>, and
1613 <filename>mk/config.h.in</filename>, are long.</para>
1618 <term>Step 2: system configuration.</term>
1620 <para>Runs the newly-created <command>configure</command>
1621 script, thus:</para>
1624 ./configure <optional><parameter>args</parameter></optional>
1627 <para><command>configure</command>'s mission is to scurry
1628 round your computer working out what architecture it has,
1629 what operating system, whether it has the
1630 <Function>vfork</Function> system call, where
1631 <command>tar</command> is kept, whether
1632 <command>gcc</command> is available, where various obscure
1633 <literal>#include</literal> files are, whether it's a
1634 leap year, and what the systems manager had for lunch. It
1635 communicates these snippets of information in two
1642 <filename>mk/config.mk.in</filename><indexterm><primary>config.mk.in</primary></indexterm>
1644 <filename>mk/config.mk</filename><indexterm><primary>config.mk</primary></indexterm>,
1645 substituting for things between
1646 “<literal>@</literal>” brackets. So,
1647 “<literal>@HaveGcc@</literal>” will be
1648 replaced by “<literal>YES</literal>” or
1649 “<literal>NO</literal>” depending on what
1650 <command>configure</command> finds.
1651 <filename>mk/config.mk</filename> is included by every
1652 Makefile (directly or indirectly), so the
1653 configuration information is thereby communicated to
1654 all Makefiles.</para>
1658 <para> It translates
1659 <filename>mk/config.h.in</filename><indexterm><primary>config.h.in</primary></indexterm>
1661 <filename>mk/config.h</filename><indexterm><primary>config.h</primary></indexterm>.
1662 The latter is <literal>#include</literal>d by
1663 various C programs, which can thereby make use of
1664 configuration information.</para>
1668 <para><command>configure</command> takes some optional
1669 arguments. Use <literal>./configure --help</literal> to
1670 get a list of the available arguments. Here are some of
1671 the ones you might need:</para>
1675 <term><literal>--with-ghc=<parameter>path</parameter></literal></term>
1676 <indexterm><primary><literal>--with-ghc</literal></primary>
1679 <para>Specifies the path to an installed GHC which
1680 you would like to use. This compiler will be used
1681 for compiling GHC-specific code (eg. GHC itself).
1682 This option <emphasis>cannot</emphasis> be specified
1683 using <filename>build.mk</filename> (see later),
1684 because <command>configure</command> needs to
1685 auto-detect the version of GHC you're using. The
1686 default is to look for a compiler named
1687 <literal>ghc</literal> in your path.</para>
1692 <term><literal>--with-hc=<parameter>path</parameter></literal></term>
1693 <indexterm><primary><literal>--with-hc</literal></primary>
1696 <para>Specifies the path to any installed Haskell
1697 compiler. This compiler will be used for compiling
1698 generic Haskell code. The default is to use
1699 <literal>ghc</literal>.</para>
1704 <term><literal>--with-gcc=<parameter>path</parameter></literal></term>
1705 <indexterm><primary><literal>--with-gcc</literal></primary>
1708 <para>Specifies the path to the installed GCC. This
1709 compiler will be used to compile all C files,
1710 <emphasis>except</emphasis> any generated by the
1711 installed Haskell compiler, which will have its own
1712 idea of which C compiler (if any) to use. The
1713 default is to use <literal>gcc</literal>.</para>
1718 <para><command>configure</command> caches the results of
1719 its run in <filename>config.cache</filename>. Quite often
1720 you don't want that; you're running
1721 <command>configure</command> a second time because
1722 something has changed. In that case, simply delete
1723 <filename>config.cache</filename>.</para>
1728 <term>Step 3: build configuration.</term>
1730 <para>Next, you say how this build of
1731 <literal>fptools</literal> is to differ from the standard
1732 defaults by creating a new file
1733 <filename>mk/build.mk</filename><indexterm><primary>build.mk</primary></indexterm>
1734 <emphasis>in the build tree</emphasis>. This file is the
1735 one and only file you edit in the build tree, precisely
1736 because it says how this build differs from the source.
1737 (Just in case your build tree does die, you might want to
1738 keep a private directory of <filename>build.mk</filename>
1739 files, and use a symbolic link in each build tree to point
1740 to the appropriate one.) So
1741 <filename>mk/build.mk</filename> never exists in the
1742 source tree—you create one in each build tree from
1743 the template. We'll discuss what to put in it
1749 <para>And that's it for configuration. Simple, eh?</para>
1751 <para>What do you put in your build-specific configuration file
1752 <filename>mk/build.mk</filename>? <emphasis>For almost all
1753 purposes all you will do is put make variable definitions that
1754 override those in</emphasis>
1755 <filename>mk/config.mk.in</filename>. The whole point of
1756 <filename>mk/config.mk.in</filename>—and its derived
1757 counterpart <filename>mk/config.mk</filename>—is to define
1758 the build configuration. It is heavily commented, as you will
1759 see if you look at it. So generally, what you do is look at
1760 <filename>mk/config.mk.in</filename>, and add definitions in
1761 <filename>mk/build.mk</filename> that override any of the
1762 <filename>config.mk</filename> definitions that you want to
1763 change. (The override occurs because the main boilerplate file,
1764 <filename>mk/boilerplate.mk</filename><indexterm><primary>boilerplate.mk</primary></indexterm>,
1765 includes <filename>build.mk</filename> after
1766 <filename>config.mk</filename>.)</para>
1768 <para>For your convenience, there's a file called <filename>build.mk.sample</filename>
1769 that can serve as a starting point for your <filename>build.mk</filename>.</para>
1771 <para>For example, <filename>config.mk.in</filename> contains
1772 the definition:</para>
1775 GhcHcOpts=-O -Rghc-timing
1778 <para>The accompanying comment explains that this is the list of
1779 flags passed to GHC when building GHC itself. For doing
1780 development, it is wise to add <literal>-DDEBUG</literal>, to
1781 enable debugging code. So you would add the following to
1782 <filename>build.mk</filename>:</para>
1784 <para>or, if you prefer,</para>
1787 GhcHcOpts += -DDEBUG
1790 <para>GNU <command>make</command> allows existing definitions to
1791 have new text appended using the “<literal>+=</literal>”
1792 operator, which is quite a convenient feature.)</para>
1794 <para>If you want to remove the <literal>-O</literal> as well (a
1795 good idea when developing, because the turn-around cycle gets a
1796 lot quicker), you can just override
1797 <literal>GhcLibHcOpts</literal> altogether:</para>
1800 GhcHcOpts=-DDEBUG -Rghc-timing
1803 <para>When reading <filename>config.mk.in</filename>, remember
1804 that anything between “@...@” signs is going to be substituted
1805 by <command>configure</command> later. You
1806 <emphasis>can</emphasis> override the resulting definition if
1807 you want, but you need to be a bit surer what you are doing.
1808 For example, there's a line that says:</para>
1814 <para>This defines the Make variables <constant>TAR</constant>
1815 to the pathname for a <command>tar</command> that
1816 <command>configure</command> finds somewhere. If you have your
1817 own pet <command>tar</command> you want to use instead, that's
1818 fine. Just add this line to <filename>mk/build.mk</filename>:</para>
1824 <para>You do not <emphasis>have</emphasis> to have a
1825 <filename>mk/build.mk</filename> file at all; if you don't,
1826 you'll get all the default settings from
1827 <filename>mk/config.mk.in</filename>.</para>
1829 <para>You can also use <filename>build.mk</filename> to override
1830 anything that <command>configure</command> got wrong. One place
1831 where this happens often is with the definition of
1832 <constant>FPTOOLS_TOP_ABS</constant>: this
1833 variable is supposed to be the canonical path to the top of your
1834 source tree, but if your system uses an automounter then the
1835 correct directory is hard to find automatically. If you find
1836 that <command>configure</command> has got it wrong, just put the
1837 correct definition in <filename>build.mk</filename>.</para>
1841 <sect2 id="sec-storysofar">
1842 <title>The story so far</title>
1844 <para>Let's summarise the steps you need to carry to get
1845 yourself a fully-configured build tree from scratch.</para>
1849 <para> Get your source tree from somewhere (CVS repository
1850 or source distribution). Say you call the root directory
1851 <filename>myfptools</filename> (it does not have to be
1852 called <filename>fptools</filename>). Make sure that you
1853 have the essential files (see <XRef
1854 LinkEnd="sec-source-tree">).</para>
1859 <para>(Optional) Use <command>lndir</command> or
1860 <command>mkshadowdir</command> to create a build tree.</para>
1864 $ mkshadowdir . /scratch/joe-bloggs/myfptools-sun4
1867 <para>(N.B. <command>mkshadowdir</command>'s first argument
1868 is taken relative to its second.) You probably want to give
1869 the build tree a name that suggests its main defining
1870 characteristic (in your mind at least), in case you later
1875 <para>Change directory to the build tree. Everything is
1876 going to happen there now.</para>
1879 $ cd /scratch/joe-bloggs/myfptools-sun4
1885 <para>Prepare for system configuration:</para>
1891 <para>(You can skip this step if you are starting from a
1892 source distribution, and you already have
1893 <filename>configure</filename> and
1894 <filename>mk/config.h.in</filename>.)</para>
1896 <para>Some projects, including GHC itself, have their own
1897 configure scripts, so it is necessary to run autoconf again
1898 in the appropriate subdirectories. eg:</para>
1901 $ (cd ghc; autoconf)
1906 <para>Do system configuration:</para>
1912 <para>Don't forget to check whether you need to add any
1913 arguments to <literal>configure</literal>; for example, a
1914 common requirement is to specify which GHC to use with
1915 <option>--with-ghc=<replaceable>ghc</replaceable></option>.</para>
1919 <para>Create the file <filename>mk/build.mk</filename>,
1920 adding definitions for your desired configuration
1929 <para>You can make subsequent changes to
1930 <filename>mk/build.mk</filename> as often as you like. You do
1931 not have to run any further configuration programs to make these
1932 changes take effect. In theory you should, however, say
1933 <command>gmake clean</command>, <command>gmake all</command>,
1934 because configuration option changes could affect
1935 anything—but in practice you are likely to know what's
1940 <title>Making things</title>
1942 <para>At this point you have made yourself a fully-configured
1943 build tree, so you are ready to start building real
1946 <para>The first thing you need to know is that <emphasis>you
1947 must use GNU <command>make</command>, usually called
1948 <command>gmake</command>, not standard Unix
1949 <command>make</command></emphasis>. If you use standard Unix
1950 <command>make</command> you will get all sorts of error messages
1951 (but no damage) because the <literal>fptools</literal>
1952 <command>Makefiles</command> use GNU <command>make</command>'s
1953 facilities extensively.</para>
1955 <para>To just build the whole thing, <command>cd</command> to
1956 the top of your <literal>fptools</literal> tree and type
1957 <command>gmake</command>. This will prepare the tree and build
1958 the various projects in the correct order.</para>
1961 <sect2 id="sec-bootstrapping">
1962 <title>Bootstrapping GHC</title>
1964 <para>GHC requires a 2-stage bootstrap in order to provide
1965 full functionality, including GHCi. By a 2-stage bootstrap, we
1966 mean that the compiler is built once using the installed GHC,
1967 and then again using the compiler built in the first stage. You
1968 can also build a stage 3 compiler, but this normally isn't
1969 necessary except to verify that the stage 2 compiler is working
1972 <para>Note that when doing a bootstrap, the stage 1 compiler
1973 must be built, followed by the runtime system and libraries, and
1974 then the stage 2 compiler. The correct ordering is implemented
1975 by the top-level fptools <filename>Makefile</filename>, so if
1976 you want everything to work automatically it's best to start
1977 <command>make</command> from the top of the tree. When building
1978 GHC, the top-level fptools <filename>Makefile</filename> is set
1979 up to do a 2-stage bootstrap by default (when you say
1980 <command>make</command>). Some other targets it supports
1987 <para>Build everything as normal, including the stage 1
1995 <para>Build the stage 2 compiler only.</para>
2002 <para>Build the stage 3 compiler only.</para>
2007 <term>bootstrap</term> <term>bootstrap2</term>
2009 <para>Build stage 1 followed by stage 2.</para>
2014 <term>bootstrap3</term>
2016 <para>Build stages 1, 2 and 3.</para>
2021 <term>install</term>
2023 <para>Install everything, including the compiler built in
2024 stage 2. To override the stage, say <literal>make install
2025 stage=<replaceable>n</replaceable></literal> where
2026 <replaceable>n</replaceable> is the stage to install.</para>
2031 <para>The top-level <filename>Makefile</filename> also arranges
2032 to do the appropriate <literal>make boot</literal> steps (see
2033 below) before actually building anything.</para>
2035 <para>The <literal>stage1</literal>, <literal>stage2</literal>
2036 and <literal>stage3</literal> targets also work in the
2037 <literal>ghc/compiler</literal> directory, but don't forget that
2038 each stage requires its own <literal>make boot</literal> step:
2039 for example, you must do</para>
2041 <screen>$ make boot stage=2</screen>
2043 <para>before <literal>make stage2</literal> in
2044 <literal>ghc/compiler</literal>.</para>
2047 <sect2 id="sec-standard-targets">
2048 <title>Standard Targets</title>
2049 <indexterm><primary>targets, standard makefile</primary></indexterm>
2050 <indexterm><primary>makefile targets</primary></indexterm>
2052 <para>In any directory you should be able to make the following:</para>
2056 <term><literal>boot</literal></term>
2058 <para>does the one-off preparation required to get ready
2059 for the real work. Notably, it does <command>gmake
2060 depend</command> in all directories that contain programs.
2061 It also builds the necessary tools for compilation to
2064 <para>Invoking the <literal>boot</literal> target
2065 explicitly is not normally necessary. From the top-level
2066 <literal>fptools</literal> directory, invoking
2067 <literal>gmake</literal> causes <literal>gmake boot
2068 all</literal> to be invoked in each of the project
2069 subdirectories, in the order specified by
2070 <literal>$(AllTargets)</literal> in
2071 <literal>config.mk</literal>.</para>
2073 <para>If you're working in a subdirectory somewhere and
2074 need to update the dependencies, <literal>gmake
2075 boot</literal> is a good way to do it.</para>
2080 <term><literal>all</literal></term>
2082 <para>makes all the final target(s) for this Makefile.
2083 Depending on which directory you are in a “final
2084 target” may be an executable program, a library
2085 archive, a shell script, or a Postscript file. Typing
2086 <command>gmake</command> alone is generally the same as
2087 typing <command>gmake all</command>.</para>
2092 <term><literal>install</literal></term>
2094 <para>installs the things built by <literal>all</literal>
2095 (except for the documentation). Where does it install
2096 them? That is specified by
2097 <filename>mk/config.mk.in</filename>; you can override it
2098 in <filename>mk/build.mk</filename>, or by running
2099 <command>configure</command> with command-line arguments
2100 like <literal>--bindir=/home/simonpj/bin</literal>; see
2101 <literal>./configure --help</literal> for the full
2107 <term><literal>install-docs</literal></term>
2109 <para>installs the documentation. Otherwise behaves just
2110 like <literal>install</literal>.</para>
2115 <term><literal>uninstall</literal></term>
2117 <para>reverses the effect of
2118 <literal>install</literal>.</para>
2123 <term><literal>clean</literal></term>
2125 <para>Delete all files from the current directory that are
2126 normally created by building the program. Don't delete
2127 the files that record the configuration, or files
2128 generated by <command>gmake boot</command>. Also preserve
2129 files that could be made by building, but normally aren't
2130 because the distribution comes with them.</para>
2135 <term><literal>distclean</literal></term>
2137 <para>Delete all files from the current directory that are
2138 created by configuring or building the program. If you
2139 have unpacked the source and built the program without
2140 creating any other files, <literal>make
2141 distclean</literal> should leave only the files that were
2142 in the distribution.</para>
2147 <term><literal>mostlyclean</literal></term>
2149 <para>Like <literal>clean</literal>, but may refrain from
2150 deleting a few files that people normally don't want to
2156 <term><literal>maintainer-clean</literal></term>
2158 <para>Delete everything from the current directory that
2159 can be reconstructed with this Makefile. This typically
2160 includes everything deleted by
2161 <literal>distclean</literal>, plus more: C source files
2162 produced by Bison, tags tables, Info files, and so
2165 <para>One exception, however: <literal>make
2166 maintainer-clean</literal> should not delete
2167 <filename>configure</filename> even if
2168 <filename>configure</filename> can be remade using a rule
2169 in the <filename>Makefile</filename>. More generally,
2170 <literal>make maintainer-clean</literal> should not delete
2171 anything that needs to exist in order to run
2172 <filename>configure</filename> and then begin to build the
2178 <term><literal>check</literal></term>
2180 <para>run the test suite.</para>
2185 <para>All of these standard targets automatically recurse into
2186 sub-directories. Certain other standard targets do not:</para>
2190 <term><literal>configure</literal></term>
2192 <para>is only available in the root directory
2193 <constant>$(FPTOOLS_TOP)</constant>; it has
2194 been discussed in <XRef
2195 LinkEnd="sec-build-config">.</para>
2200 <term><literal>depend</literal></term>
2202 <para>make a <filename>.depend</filename> file in each
2203 directory that needs it. This <filename>.depend</filename>
2204 file contains mechanically-generated dependency
2205 information; for example, suppose a directory contains a
2206 Haskell source module <filename>Foo.lhs</filename> which
2207 imports another module <literal>Baz</literal>. Then the
2208 generated <filename>.depend</filename> file will contain
2209 the dependency:</para>
2215 <para>which says that the object file
2216 <filename>Foo.o</filename> depends on the interface file
2217 <filename>Baz.hi</filename> generated by compiling module
2218 <literal>Baz</literal>. The <filename>.depend</filename>
2219 file is automatically included by every Makefile.</para>
2224 <term><literal>binary-dist</literal></term>
2226 <para>make a binary distribution. This is the target we
2227 use to build the binary distributions of GHC and
2233 <term><literal>dist</literal></term>
2235 <para>make a source distribution. Note that this target
2236 does “make distclean” as part of its work;
2237 don't use it if you want to keep what you've built.</para>
2242 <para>Most <filename>Makefile</filename>s have targets other
2243 than these. You can discover them by looking in the
2244 <filename>Makefile</filename> itself.</para>
2248 <title>Using a project from the build tree</title>
2250 <para>If you want to build GHC (say) and just use it direct from
2251 the build tree without doing <literal>make install</literal>
2252 first, you can run the in-place driver script:
2253 <filename>ghc/compiler/ghc-inplace</filename>.</para>
2255 <para> Do <emphasis>NOT</emphasis> use
2256 <filename>ghc/compiler/ghc</filename>, or
2257 <filename>ghc/compiler/ghc-6.xx</filename>, as these are the
2258 scripts intended for installation, and contain hard-wired paths
2259 to the installed libraries, rather than the libraries in the
2262 <para>Happy can similarly be run from the build tree, using
2263 <filename>happy/src/happy-inplace</filename>, and similarly for
2264 Alex and Haddock.</para>
2268 <title>Fast Making</title>
2270 <indexterm><primary>fastmake</primary></indexterm>
2271 <indexterm><primary>dependencies, omitting</primary></indexterm>
2272 <indexterm><primary>FAST, makefile variable</primary></indexterm>
2274 <para>Sometimes the dependencies get in the way: if you've made
2275 a small change to one file, and you're absolutely sure that it
2276 won't affect anything else, but you know that
2277 <command>make</command> is going to rebuild everything anyway,
2278 the following hack may be useful:</para>
2284 <para>This tells the make system to ignore dependencies and just
2285 build what you tell it to. In other words, it's equivalent to
2286 temporarily removing the <filename>.depend</filename> file in
2287 the current directory (where <command>mkdependHS</command> and
2288 friends store their dependency information).</para>
2290 <para>A bit of history: GHC used to come with a
2291 <command>fastmake</command> script that did the above job, but
2292 GNU make provides the features we need to do it without
2293 resorting to a script. Also, we've found that fastmaking is
2294 less useful since the advent of GHC's recompilation checker (see
2295 the User's Guide section on "Separate Compilation").</para>
2299 <sect1 id="sec-makefile-arch">
2300 <title>The <filename>Makefile</filename> architecture</title>
2301 <indexterm><primary>makefile architecture</primary></indexterm>
2303 <para><command>make</command> is great if everything
2304 works—you type <command>gmake install</command> and lo! the
2305 right things get compiled and installed in the right places. Our
2306 goal is to make this happen often, but somehow it often doesn't;
2307 instead some weird error message eventually emerges from the
2308 bowels of a directory you didn't know existed.</para>
2310 <para>The purpose of this section is to give you a road-map to
2311 help you figure out what is going right and what is going
2315 <title>Debugging</title>
2317 <para>Debugging <filename>Makefile</filename>s is something of a
2318 black art, but here's a couple of tricks that we find
2319 particularly useful. The following command allows you to see
2320 the contents of any make variable in the context of the current
2321 <filename>Makefile</filename>:</para>
2323 <screen>$ make show VALUE=HS_SRCS</screen>
2325 <para>where you can replace <literal>HS_SRCS</literal> with the
2326 name of any variable you wish to see the value of.</para>
2328 <para>GNU make has a <option>-d</option> option which generates
2329 a dump of the decision procedure used to arrive at a conclusion
2330 about which files should be recompiled. Sometimes useful for
2331 tracking down problems with superfluous or missing
2332 recompilations.</para>
2336 <title>A small project</title>
2338 <para>To get started, let us look at the
2339 <filename>Makefile</filename> for an imaginary small
2340 <literal>fptools</literal> project, <literal>small</literal>.
2341 Each project in <literal>fptools</literal> has its own directory
2342 in <constant>FPTOOLS_TOP</constant>, so the
2343 <literal>small</literal> project will have its own directory
2344 <constant>FPOOLS_TOP/small/</constant>. Inside the
2345 <filename>small/</filename> directory there will be a
2346 <filename>Makefile</filename>, looking something like
2349 <indexterm><primary>Makefile, minimal</primary></indexterm>
2352 # Makefile for fptools project "small"
2355 include $(TOP)/mk/boilerplate.mk
2357 SRCS = $(wildcard *.lhs) $(wildcard *.c)
2360 include $(TOP)/target.mk
2363 <para>this <filename>Makefile</filename> has three
2368 <para>The first section includes
2371 One of the most important
2372 features of GNU <command>make</command> that we use is the ability for a <filename>Makefile</filename> to
2373 include another named file, very like <command>cpp</command>'s <literal>#include</literal>
2378 a file of “boilerplate” code from the level
2379 above (which in this case will be
2380 <filename><constant>FPTOOLS_TOP</constant>/mk/boilerplate.mk</filename><indexterm><primary>boilerplate.mk</primary></indexterm>).
2381 As its name suggests, <filename>boilerplate.mk</filename>
2382 consists of a large quantity of standard
2383 <filename>Makefile</filename> code. We discuss this
2384 boilerplate in more detail in <XRef LinkEnd="sec-boiler">.
2385 <indexterm><primary>include, directive in
2386 Makefiles</primary></indexterm> <indexterm><primary>Makefile
2387 inclusion</primary></indexterm></para>
2389 <para>Before the <literal>include</literal> statement, you
2390 must define the <command>make</command> variable
2391 <constant>TOP</constant><indexterm><primary>TOP</primary></indexterm>
2392 to be the directory containing the <filename>mk</filename>
2393 directory in which the <filename>boilerplate.mk</filename>
2394 file is. It is <emphasis>not</emphasis> OK to simply say</para>
2397 include ../mk/boilerplate.mk # NO NO NO
2401 <para>Why? Because the <filename>boilerplate.mk</filename>
2402 file needs to know where it is, so that it can, in turn,
2403 <literal>include</literal> other files. (Unfortunately,
2404 when an <literal>include</literal>d file does an
2405 <literal>include</literal>, the filename is treated relative
2406 to the directory in which <command>gmake</command> is being
2407 run, not the directory in which the
2408 <literal>include</literal>d sits.) In general,
2409 <emphasis>every file <filename>foo.mk</filename> assumes
2411 <filename><constant>$(TOP)</constant>/mk/foo.mk</filename>
2412 refers to itself.</emphasis> It is up to the
2413 <filename>Makefile</filename> doing the
2414 <literal>include</literal> to ensure this is the case.</para>
2416 <para>Files intended for inclusion in other
2417 <filename>Makefile</filename>s are written to have the
2418 following property: <emphasis>after
2419 <filename>foo.mk</filename> is <literal>include</literal>d,
2420 it leaves <constant>TOP</constant> containing the same value
2421 as it had just before the <literal>include</literal>
2422 statement</emphasis>. In our example, this invariant
2423 guarantees that the <literal>include</literal> for
2424 <filename>target.mk</filename> will look in the same
2425 directory as that for <filename>boilerplate.mk</filename>.</para>
2429 <para> The second section defines the following standard
2430 <command>make</command> variables:
2431 <constant>SRCS</constant><indexterm><primary>SRCS</primary></indexterm>
2432 (the source files from which is to be built), and
2433 <constant>HS_PROG</constant><indexterm><primary>HS_PROG</primary></indexterm>
2434 (the executable binary to be built). We will discuss in
2435 more detail what the “standard variables” are,
2436 and how they affect what happens, in <XRef
2437 LinkEnd="sec-targets">.</para>
2439 <para>The definition for <constant>SRCS</constant> uses the
2440 useful GNU <command>make</command> construct
2441 <literal>$(wildcard $pat$)</literal><indexterm><primary>wildcard</primary></indexterm>,
2442 which expands to a list of all the files matching the
2443 pattern <literal>pat</literal> in the current directory. In
2444 this example, <constant>SRCS</constant> is set to the list
2445 of all the <filename>.lhs</filename> and
2446 <filename>.c</filename> files in the directory. (Let's
2447 suppose there is one of each, <filename>Foo.lhs</filename>
2448 and <filename>Baz.c</filename>.)</para>
2452 <para>The last section includes a second file of standard
2454 <filename>target.mk</filename><indexterm><primary>target.mk</primary></indexterm>.
2455 It contains the rules that tell <command>gmake</command> how
2456 to make the standard targets (<Xref
2457 LinkEnd="sec-standard-targets">). Why, you ask, can't this
2458 standard code be part of
2459 <filename>boilerplate.mk</filename>? Good question. We
2460 discuss the reason later, in <Xref
2461 LinkEnd="sec-boiler-arch">.</para>
2463 <para>You do not <emphasis>have</emphasis> to
2464 <literal>include</literal> the
2465 <filename>target.mk</filename> file. Instead, you can write
2466 rules of your own for all the standard targets. Usually,
2467 though, you will find quite a big payoff from using the
2468 canned rules in <filename>target.mk</filename>; the price
2469 tag is that you have to understand what canned rules get
2470 enabled, and what they do (<Xref
2471 LinkEnd="sec-targets">).</para>
2475 <para>In our example <filename>Makefile</filename>, most of the
2476 work is done by the two <literal>include</literal>d files. When
2477 you say <command>gmake all</command>, the following things
2482 <para><command>gmake</command> figures out that the object
2483 files are <filename>Foo.o</filename> and
2484 <filename>Baz.o</filename>.</para>
2488 <para>It uses a boilerplate pattern rule to compile
2489 <filename>Foo.lhs</filename> to <filename>Foo.o</filename>
2490 using a Haskell compiler. (Which one? That is set in the
2491 build configuration.)</para>
2495 <para>It uses another standard pattern rule to compile
2496 <filename>Baz.c</filename> to <filename>Baz.o</filename>,
2497 using a C compiler. (Ditto.)</para>
2501 <para>It links the resulting <filename>.o</filename> files
2502 together to make <literal>small</literal>, using the Haskell
2503 compiler to do the link step. (Why not use
2504 <command>ld</command>? Because the Haskell compiler knows
2505 what standard libraries to link in. How did
2506 <command>gmake</command> know to use the Haskell compiler to
2507 do the link, rather than the C compiler? Because we set the
2508 variable <constant>HS_PROG</constant> rather than
2509 <constant>C_PROG</constant>.)</para>
2513 <para>All <filename>Makefile</filename>s should follow the above
2514 three-section format.</para>
2518 <title>A larger project</title>
2520 <para>Larger projects are usually structured into a number of
2521 sub-directories, each of which has its own
2522 <filename>Makefile</filename>. (In very large projects, this
2523 sub-structure might be iterated recursively, though that is
2524 rare.) To give you the idea, here's part of the directory
2525 structure for the (rather large) GHC project:</para>
2535 ...source files for documentation...
2538 ...source files for driver...
2541 parser/...source files for parser...
2542 renamer/...source files for renamer...
2546 <para>The sub-directories <filename>docs</filename>,
2547 <filename>driver</filename>, <filename>compiler</filename>, and
2548 so on, each contains a sub-component of GHC, and each has its
2549 own <filename>Makefile</filename>. There must also be a
2550 <filename>Makefile</filename> in
2551 <filename><constant>$(FPTOOLS_TOP)</constant>/ghc</filename>.
2552 It does most of its work by recursively invoking
2553 <command>gmake</command> on the <filename>Makefile</filename>s
2554 in the sub-directories. We say that
2555 <filename>ghc/Makefile</filename> is a <emphasis>non-leaf
2556 <filename>Makefile</filename></emphasis>, because it does little
2557 except organise its children, while the
2558 <filename>Makefile</filename>s in the sub-directories are all
2559 <emphasis>leaf <filename>Makefile</filename>s</emphasis>. (In
2560 principle the sub-directories might themselves contain a
2561 non-leaf <filename>Makefile</filename> and several
2562 sub-sub-directories, but that does not happen in GHC.)</para>
2564 <para>The <filename>Makefile</filename> in
2565 <filename>ghc/compiler</filename> is considered a leaf
2566 <filename>Makefile</filename> even though the
2567 <filename>ghc/compiler</filename> has sub-directories, because
2568 these sub-directories do not themselves have
2569 <filename>Makefile</filename>s in them. They are just used to
2570 structure the collection of modules that make up GHC, but all
2571 are managed by the single <filename>Makefile</filename> in
2572 <filename>ghc/compiler</filename>.</para>
2574 <para>You will notice that <filename>ghc/</filename> also
2575 contains a directory <filename>ghc/mk/</filename>. It contains
2576 GHC-specific <filename>Makefile</filename> boilerplate code.
2577 More precisely:</para>
2581 <para><filename>ghc/mk/boilerplate.mk</filename> is included
2582 at the top of <filename>ghc/Makefile</filename>, and of all
2583 the leaf <filename>Makefile</filename>s in the
2584 sub-directories. It in turn <literal>include</literal>s the
2585 main boilerplate file
2586 <filename>mk/boilerplate.mk</filename>.</para>
2590 <para><filename>ghc/mk/target.mk</filename> is
2591 <literal>include</literal>d at the bottom of
2592 <filename>ghc/Makefile</filename>, and of all the leaf
2593 <filename>Makefile</filename>s in the sub-directories. It
2594 in turn <literal>include</literal>s the file
2595 <filename>mk/target.mk</filename>.</para>
2599 <para>So these two files are the place to look for GHC-wide
2600 customisation of the standard boilerplate.</para>
2603 <sect2 id="sec-boiler-arch">
2604 <title>Boilerplate architecture</title>
2605 <indexterm><primary>boilerplate architecture</primary></indexterm>
2607 <para>Every <filename>Makefile</filename> includes a
2608 <filename>boilerplate.mk</filename><indexterm><primary>boilerplate.mk</primary></indexterm>
2609 file at the top, and
2610 <filename>target.mk</filename><indexterm><primary>target.mk</primary></indexterm>
2611 file at the bottom. In this section we discuss what is in these
2612 files, and why there have to be two of them. In general:</para>
2616 <para><filename>boilerplate.mk</filename> consists of:</para>
2620 <para><emphasis>Definitions of millions of
2621 <command>make</command> variables</emphasis> that
2622 collectively specify the build configuration. Examples:
2623 <constant>HC_OPTS</constant><indexterm><primary>HC_OPTS</primary></indexterm>,
2624 the options to feed to the Haskell compiler;
2625 <constant>NoFibSubDirs</constant><indexterm><primary>NoFibSubDirs</primary></indexterm>,
2626 the sub-directories to enable within the
2627 <literal>nofib</literal> project;
2628 <constant>GhcWithHc</constant><indexterm><primary>GhcWithHc</primary></indexterm>,
2629 the name of the Haskell compiler to use when compiling
2630 GHC in the <literal>ghc</literal> project.</para>
2634 <para><emphasis>Standard pattern rules</emphasis> that
2635 tell <command>gmake</command> how to construct one file
2636 from another.</para>
2640 <para><filename>boilerplate.mk</filename> needs to be
2641 <literal>include</literal>d at the <emphasis>top</emphasis>
2642 of each <filename>Makefile</filename>, so that the user can
2643 replace the boilerplate definitions or pattern rules by
2644 simply giving a new definition or pattern rule in the
2645 <filename>Makefile</filename>. <command>gmake</command>
2646 simply takes the last definition as the definitive one.</para>
2648 <para>Instead of <emphasis>replacing</emphasis> boilerplate
2649 definitions, it is also quite common to
2650 <emphasis>augment</emphasis> them. For example, a
2651 <filename>Makefile</filename> might say:</para>
2657 <para>thereby adding “<option>-O</option>” to
2659 <constant>SRC_HC_OPTS</constant><indexterm><primary>SRC_HC_OPTS</primary></indexterm>.</para>
2663 <para><filename>target.mk</filename> contains
2664 <command>make</command> rules for the standard targets
2665 described in <Xref LinkEnd="sec-standard-targets">. These
2666 rules are selectively included, depending on the setting of
2667 certain <command>make</command> variables. These variables
2668 are usually set in the middle section of the
2669 <filename>Makefile</filename> between the two
2670 <literal>include</literal>s.</para>
2672 <para><filename>target.mk</filename> must be included at the
2673 end (rather than being part of
2674 <filename>boilerplate.mk</filename>) for several tiresome
2680 <para><command>gmake</command> commits target and
2681 dependency lists earlier than it should. For example,
2682 <FIlename>target.mk</FIlename> has a rule that looks
2686 $(HS_PROG) : $(OBJS)
2687 $(HC) $(LD_OPTS) $< -o $@
2690 <para>If this rule was in
2691 <filename>boilerplate.mk</filename> then
2692 <constant>$(HS_PROG)</constant><indexterm><primary>HS_PROG</primary></indexterm>
2694 <constant>$(OBJS)</constant><indexterm><primary>OBJS</primary></indexterm>
2695 would not have their final values at the moment
2696 <command>gmake</command> encountered the rule. Alas,
2697 <command>gmake</command> takes a snapshot of their
2698 current values, and wires that snapshot into the rule.
2699 (In contrast, the commands executed when the rule
2700 “fires” are only substituted at the moment
2701 of firing.) So, the rule must follow the definitions
2702 given in the <filename>Makefile</filename> itself.</para>
2706 <para>Unlike pattern rules, ordinary rules cannot be
2707 overriden or replaced by subsequent rules for the same
2708 target (at least, not without an error message).
2709 Including ordinary rules in
2710 <filename>boilerplate.mk</filename> would prevent the
2711 user from writing rules for specific targets in specific
2716 <para>There are a couple of other reasons I've
2717 forgotten, but it doesn't matter too much.</para>
2724 <sect2 id="sec-boiler">
2725 <title>The main <filename>mk/boilerplate.mk</filename> file</title>
2726 <indexterm><primary>boilerplate.mk</primary></indexterm>
2728 <para>If you look at
2729 <filename><constant>$(FPTOOLS_TOP)</constant>/mk/boilerplate.mk</filename>
2730 you will find that it consists of the following sections, each
2731 held in a separate file:</para>
2735 <term><filename>config.mk</filename></term>
2736 <indexterm><primary>config.mk</primary></indexterm>
2738 <para>is the build configuration file we discussed at
2739 length in <Xref LinkEnd="sec-build-config">.</para>
2744 <term><filename>paths.mk</filename></term>
2745 <indexterm><primary>paths.mk</primary></indexterm>
2747 <para>defines <command>make</command> variables for
2748 pathnames and file lists. This file contains code for
2749 automatically compiling lists of source files and deriving
2750 lists of object files from those. The results can be
2751 overriden in the <filename>Makefile</filename>, but in
2752 most cases the automatic setup should do the right
2755 <para>The following variables may be set in the
2756 <filename>Makefile</filename> to affect how the automatic
2757 source file search is done:</para>
2761 <term><literal>ALL_DIRS</literal></term>
2762 <indexterm><primary><literal>ALL_DIRS</literal></primary>
2765 <para>Set to a list of directories to search in
2766 addition to the current directory for source
2772 <term><literal>EXCLUDE_SRCS</literal></term>
2773 <indexterm><primary><literal>EXCLUDE_SRCS</literal></primary>
2776 <para>Set to a list of source files (relative to the
2777 current directory) to omit from the automatic
2778 search. The source searching machinery is clever
2779 enough to know that if you exclude a source file
2780 from which other sources are derived, then the
2781 derived sources should also be excluded. For
2782 example, if you set <literal>EXCLUDED_SRCS</literal>
2783 to include <filename>Foo.y</filename>, then
2784 <filename>Foo.hs</filename> will also be
2790 <term><literal>EXTRA_SRCS</literal></term>
2791 <indexterm><primary><literal>EXCLUDE_SRCS</literal></primary>
2794 <para>Set to a list of extra source files (perhaps
2795 in directories not listed in
2796 <literal>ALL_DIRS</literal>) that should be
2802 <para>The results of the automatic source file search are
2803 placed in the following make variables:</para>
2807 <term><literal>SRCS</literal></term>
2808 <indexterm><primary><literal>SRCS</literal></primary></indexterm>
2810 <para>All source files found, sorted and without
2811 duplicates, including those which might not exist
2812 yet but will be derived from other existing sources.
2813 <literal>SRCS</literal> <emphasis>can</emphasis> be
2814 overriden if necessary, in which case the variables
2815 below will follow suit.</para>
2820 <term><literal>HS_SRCS</literal></term>
2821 <indexterm><primary><literal>HS_SRCS</literal></primary></indexterm>
2823 <para>all Haskell source files in the current
2824 directory, including those derived from other source
2825 files (eg. Happy sources also give rise to Haskell
2831 <term><literal>HS_OBJS</literal></term>
2832 <indexterm><primary><literal>HS_OBJS</literal></primary></indexterm>
2834 <para>Object files derived from
2835 <literal>HS_SRCS</literal>.</para>
2840 <term><literal>HS_IFACES</literal></term>
2841 <indexterm><primary><literal>HS_IFACES</literal></primary></indexterm>
2843 <para>Interface files (<literal>.hi</literal> files)
2844 derived from <literal>HS_SRCS</literal>.</para>
2849 <term><literal>C_SRCS</literal></term>
2850 <indexterm><primary><literal>C_SRCS</literal></primary></indexterm>
2852 <para>All C source files found.</para>
2857 <term><literal>C_OBJS</literal></term>
2858 <indexterm><primary><literal>C_OBJS</literal></primary></indexterm>
2860 <para>Object files derived from
2861 <literal>C_SRCS</literal>.</para>
2866 <term><literal>SCRIPT_SRCS</literal></term>
2867 <indexterm><primary><literal>SCRIPT_SRCS</literal></primary></indexterm>
2869 <para>All script source files found
2870 (<literal>.lprl</literal> files).</para>
2875 <term><literal>SCRIPT_OBJS</literal></term>
2876 <indexterm><primary><literal>SCRIPT_OBJS</literal></primary></indexterm>
2878 <para><quote>object</quote> files derived from
2879 <literal>SCRIPT_SRCS</literal>
2880 (<literal>.prl</literal> files).</para>
2885 <term><literal>HSC_SRCS</literal></term>
2886 <indexterm><primary><literal>HSC_SRCS</literal></primary></indexterm>
2888 <para>All <literal>hsc2hs</literal> source files
2889 (<literal>.hsc</literal> files).</para>
2894 <term><literal>HAPPY_SRCS</literal></term>
2895 <indexterm><primary><literal>HAPPY_SRCS</literal></primary></indexterm>
2897 <para>All <literal>happy</literal> source files
2898 (<literal>.y</literal> or <literal>.hy</literal> files).</para>
2903 <term><literal>OBJS</literal></term>
2904 <indexterm><primary>OBJS</primary></indexterm>
2906 <para>the concatenation of
2907 <literal>$(HS_OBJS)</literal>,
2908 <literal>$(C_OBJS)</literal>, and
2909 <literal>$(SCRIPT_OBJS)</literal>.</para>
2914 <para>Any or all of these definitions can easily be
2915 overriden by giving new definitions in your
2916 <filename>Makefile</filename>.</para>
2918 <para>What, exactly, does <filename>paths.mk</filename>
2919 consider a <quote>source file</quote> to be? It's based
2920 on the file's suffix (e.g. <filename>.hs</filename>,
2921 <filename>.lhs</filename>, <filename>.c</filename>,
2922 <filename>.hy</filename>, etc), but this is the kind of
2923 detail that changes, so rather than enumerate the source
2924 suffices here the best thing to do is to look in
2925 <filename>paths.mk</filename>.</para>
2930 <term><filename>opts.mk</filename></term>
2931 <indexterm><primary>opts.mk</primary></indexterm>
2933 <para>defines <command>make</command> variables for option
2934 strings to pass to each program. For example, it defines
2935 <constant>HC_OPTS</constant><indexterm><primary>HC_OPTS</primary></indexterm>,
2936 the option strings to pass to the Haskell compiler. See
2937 <Xref LinkEnd="sec-suffix">.</para>
2942 <term><filename>suffix.mk</filename></term>
2943 <indexterm><primary>suffix.mk</primary></indexterm>
2945 <para>defines standard pattern rules—see <Xref
2946 LinkEnd="sec-suffix">.</para>
2951 <para>Any of the variables and pattern rules defined by the
2952 boilerplate file can easily be overridden in any particular
2953 <filename>Makefile</filename>, because the boilerplate
2954 <literal>include</literal> comes first. Definitions after this
2955 <literal>include</literal> directive simply override the default
2956 ones in <filename>boilerplate.mk</filename>.</para>
2959 <sect2 id="sec-suffix">
2960 <title>Pattern rules and options</title>
2961 <indexterm><primary>Pattern rules</primary></indexterm>
2964 <filename>suffix.mk</filename><indexterm><primary>suffix.mk</primary></indexterm>
2965 defines standard <emphasis>pattern rules</emphasis> that say how
2966 to build one kind of file from another, for example, how to
2967 build a <filename>.o</filename> file from a
2968 <filename>.c</filename> file. (GNU <command>make</command>'s
2969 <emphasis>pattern rules</emphasis> are more powerful and easier
2970 to use than Unix <command>make</command>'s <emphasis>suffix
2971 rules</emphasis>.)</para>
2973 <para>Almost all the rules look something like this:</para>
2978 $(CC) $(CC_OPTS) -c $< -o $@
2981 <para>Here's how to understand the rule. It says that
2982 <emphasis>something</emphasis><filename>.o</filename> (say
2983 <filename>Foo.o</filename>) can be built from
2984 <emphasis>something</emphasis><filename>.c</filename>
2985 (<filename>Foo.c</filename>), by invoking the C compiler (path
2986 name held in <constant>$(CC)</constant>), passing to it
2987 the options <constant>$(CC_OPTS)</constant> and
2988 the rule's dependent file of the rule
2989 <literal>$<</literal> (<filename>Foo.c</filename> in
2990 this case), and putting the result in the rule's target
2991 <literal>$@</literal> (<filename>Foo.o</filename> in this
2994 <para>Every program is held in a <command>make</command>
2995 variable defined in <filename>mk/config.mk</filename>—look
2996 in <filename>mk/config.mk</filename> for the complete list. One
2997 important one is the Haskell compiler, which is called
2998 <constant>$(HC)</constant>.</para>
3000 <para>Every program's options are are held in a
3001 <command>make</command> variables called
3002 <constant><prog>_OPTS</constant>. the
3003 <constant><prog>_OPTS</constant> variables are
3004 defined in <filename>mk/opts.mk</filename>. Almost all of them
3005 are defined like this:</para>
3008 CC_OPTS = $(SRC_CC_OPTS) $(WAY$(_way)_CC_OPTS) $($*_CC_OPTS) $(EXTRA_CC_OPTS)
3011 <para>The four variables from which
3012 <constant>CC_OPTS</constant> is built have the following
3017 <term><constant>SRC_CC_OPTS</constant><indexterm><primary>SRC_CC_OPTS</primary></indexterm>:</term>
3019 <para>options passed to all C compilations.</para>
3024 <term><constant>WAY_<way>_CC_OPTS</constant>:</term>
3026 <para>options passed to C compilations for way
3027 <literal><way></literal>. For example,
3028 <constant>WAY_mp_CC_OPTS</constant>
3029 gives options to pass to the C compiler when compiling way
3030 <literal>mp</literal>. The variable
3031 <constant>WAY_CC_OPTS</constant> holds
3032 options to pass to the C compiler when compiling the
3033 standard way. (<Xref LinkEnd="sec-ways"> dicusses
3034 multi-way compilation.)</para>
3039 <term><constant><module>_CC_OPTS</constant>:</term>
3041 <para>options to pass to the C compiler that are specific
3042 to module <literal><module></literal>. For example,
3043 <constant>SMap_CC_OPTS</constant> gives the
3044 specific options to pass to the C compiler when compiling
3045 <filename>SMap.c</filename>.</para>
3050 <term><constant>EXTRA_CC_OPTS</constant><indexterm><primary>EXTRA_CC_OPTS</primary></indexterm>:</term>
3052 <para>extra options to pass to all C compilations. This
3053 is intended for command line use, thus:</para>
3056 gmake libHS.a EXTRA_CC_OPTS="-v"
3063 <sect2 id="sec-targets">
3064 <title>The main <filename>mk/target.mk</filename> file</title>
3065 <indexterm><primary>target.mk</primary></indexterm>
3067 <para><filename>target.mk</filename> contains canned rules for
3068 all the standard targets described in <Xref
3069 LinkEnd="sec-standard-targets">. It is complicated by the fact
3070 that you don't want all of these rules to be active in every
3071 <filename>Makefile</filename>. Rather than have a plethora of
3072 tiny files which you can include selectively, there is a single
3073 file, <filename>target.mk</filename>, which selectively includes
3074 rules based on whether you have defined certain variables in
3075 your <filename>Makefile</filename>. This section explains what
3076 rules you get, what variables control them, and what the rules
3077 do. Hopefully, you will also get enough of an idea of what is
3078 supposed to happen that you can read and understand any weird
3079 special cases yourself.</para>
3083 <term><constant>HS_PROG</constant><indexterm><primary>HS_PROG</primary></indexterm>.</term>
3085 <para>If <constant>HS_PROG</constant> is defined,
3086 you get rules with the following targets:</para>
3090 <term><filename>HS_PROG</filename><indexterm><primary>HS_PROG</primary></indexterm></term>
3092 <para>itself. This rule links
3093 <constant>$(OBJS)</constant> with the Haskell
3094 runtime system to get an executable called
3095 <constant>$(HS_PROG)</constant>.</para>
3100 <term><literal>install</literal><indexterm><primary>install</primary></indexterm></term>
3103 <constant>$(HS_PROG)</constant> in
3104 <constant>$(bindir)</constant>.</para>
3113 <term><constant>C_PROG</constant><indexterm><primary>C_PROG</primary></indexterm></term>
3115 <para>is similar to <constant>HS_PROG</constant>,
3116 except that the link step links
3117 <constant>$(C_OBJS)</constant> with the C
3118 runtime system.</para>
3123 <term><constant>LIBRARY</constant><indexterm><primary>LIBRARY</primary></indexterm></term>
3125 <para>is similar to <constant>HS_PROG</constant>,
3126 except that it links
3127 <constant>$(LIB_OBJS)</constant> to make the
3128 library archive <constant>$(LIBRARY)</constant>,
3129 and <literal>install</literal> installs it in
3130 <constant>$(libdir)</constant>.</para>
3135 <term><constant>LIB_DATA</constant><indexterm><primary>LIB_DATA</primary></indexterm></term>
3137 <para>…</para>
3142 <term><constant>LIB_EXEC</constant><indexterm><primary>LIB_EXEC</primary></indexterm></term>
3144 <para>…</para>
3149 <term><constant>HS_SRCS</constant><indexterm><primary>HS_SRCS</primary></indexterm>, <constant>C_SRCS</constant><indexterm><primary>C_SRCS</primary></indexterm>.</term>
3151 <para>If <constant>HS_SRCS</constant> is defined
3152 and non-empty, a rule for the target
3153 <literal>depend</literal> is included, which generates
3154 dependency information for Haskell programs. Similarly
3155 for <constant>C_SRCS</constant>.</para>
3160 <para>All of these rules are “double-colon” rules,
3164 install :: $(HS_PROG)
3165 ...how to install it...
3168 <para>GNU <command>make</command> treats double-colon rules as
3169 separate entities. If there are several double-colon rules for
3170 the same target it takes each in turn and fires it if its
3171 dependencies say to do so. This means that you can, for
3172 example, define both <constant>HS_PROG</constant> and
3173 <constant>LIBRARY</constant>, which will generate two rules for
3174 <literal>install</literal>. When you type <command>gmake
3175 install</command> both rules will be fired, and both the program
3176 and the library will be installed, just as you wanted.</para>
3179 <sect2 id="sec-subdirs">
3180 <title>Recursion</title>
3181 <indexterm><primary>recursion, in makefiles</primary></indexterm>
3182 <indexterm><primary>Makefile, recursing into subdirectories</primary></indexterm>
3184 <para>In leaf <filename>Makefile</filename>s the variable
3185 <constant>SUBDIRS</constant><indexterm><primary>SUBDIRS</primary></indexterm>
3186 is undefined. In non-leaf <filename>Makefile</filename>s,
3187 <constant>SUBDIRS</constant> is set to the list of
3188 sub-directories that contain subordinate
3189 <filename>Makefile</filename>s. <emphasis>It is up to you to
3190 set <constant>SUBDIRS</constant> in the
3191 <filename>Makefile</filename>.</emphasis> There is no automation
3192 here—<constant>SUBDIRS</constant> is too important to
3195 <para>When <constant>SUBDIRS</constant> is defined,
3196 <filename>target.mk</filename> includes a rather neat rule for
3197 the standard targets (<Xref LinkEnd="sec-standard-targets"> that
3198 simply invokes <command>make</command> recursively in each of
3199 the sub-directories.</para>
3201 <para><emphasis>These recursive invocations are guaranteed to
3202 occur in the order in which the list of directories is specified
3203 in <constant>SUBDIRS</constant>. </emphasis>This guarantee can
3204 be important. For example, when you say <command>gmake
3205 boot</command> it can be important that the recursive invocation
3206 of <command>make boot</command> is done in one sub-directory
3207 (the include files, say) before another (the source files).
3208 Generally, put the most independent sub-directory first, and the
3209 most dependent last.</para>
3212 <sect2 id="sec-ways">
3213 <title>Way management</title>
3214 <indexterm><primary>way management</primary></indexterm>
3216 <para>We sometimes want to build essentially the same system in
3217 several different “ways”. For example, we want to build GHC's
3218 <literal>Prelude</literal> libraries with and without profiling,
3219 so that there is an appropriately-built library archive to link
3220 with when the user compiles his program. It would be possible
3221 to have a completely separate build tree for each such “way”,
3222 but it would be horribly bureaucratic, especially since often
3223 only parts of the build tree need to be constructed in multiple
3227 <filename>target.mk</filename><indexterm><primary>target.mk</primary></indexterm>
3228 contains some clever magic to allow you to build several
3229 versions of a system; and to control locally how many versions
3230 are built and how they differ. This section explains the
3233 <para>The files for a particular way are distinguished by
3234 munging the suffix. The <quote>normal way</quote> is always
3235 built, and its files have the standard suffices
3236 <filename>.o</filename>, <filename>.hi</filename>, and so on.
3237 In addition, you can build one or more extra ways, each
3238 distinguished by a <emphasis>way tag</emphasis>. The object
3239 files and interface files for one of these extra ways are
3240 distinguished by their suffix. For example, way
3241 <literal>mp</literal> has files
3242 <filename>.mp_o</filename> and
3243 <filename>.mp_hi</filename>. Library archives have their
3244 way tag the other side of the dot, for boring reasons; thus,
3245 <filename>libHS_mp.a</filename>.</para>
3247 <para>A <command>make</command> variable called
3248 <constant>way</constant> holds the current way tag.
3249 <emphasis><constant>way</constant> is only ever set on the
3250 command line of <command>gmake</command></emphasis> (usually in
3251 a recursive invocation of <command>gmake</command> by the
3252 system). It is never set inside a
3253 <filename>Makefile</filename>. So it is a global constant for
3254 any one invocation of <command>gmake</command>. Two other
3255 <command>make</command> variables,
3256 <constant>way_</constant> and
3257 <constant>_way</constant> are immediately derived from
3258 <constant>$(way)</constant> and never altered. If
3259 <constant>way</constant> is not set, then neither are
3260 <constant>way_</constant> and
3261 <constant>_way</constant>, and the invocation of
3262 <command>make</command> will build the <quote>normal
3263 way</quote>. If <constant>way</constant> is set, then the other
3264 two variables are set in sympathy. For example, if
3265 <constant>$(way)</constant> is “<literal>mp</literal>”,
3266 then <constant>way_</constant> is set to
3267 “<literal>mp_</literal>” and
3268 <constant>_way</constant> is set to
3269 “<literal>_mp</literal>”. These three variables are
3270 then used when constructing file names.</para>
3272 <para>So how does <command>make</command> ever get recursively
3273 invoked with <constant>way</constant> set? There are two ways
3274 in which this happens:</para>
3278 <para>For some (but not all) of the standard targets, when
3279 in a leaf sub-directory, <command>make</command> is
3280 recursively invoked for each way tag in
3281 <constant>$(WAYS)</constant>. You set
3282 <constant>WAYS</constant> in the
3283 <filename>Makefile</filename> to the list of way tags you
3284 want these targets built for. The mechanism here is very
3285 much like the recursive invocation of
3286 <command>make</command> in sub-directories (<Xref
3287 LinkEnd="sec-subdirs">). It is up to you to set
3288 <constant>WAYS</constant> in your
3289 <filename>Makefile</filename>; this is how you control what
3290 ways will get built.</para>
3294 <para>For a useful collection of targets (such as
3295 <filename>libHS_mp.a</filename>,
3296 <filename>Foo.mp_o</filename>) there is a rule which
3297 recursively invokes <command>make</command> to make the
3298 specified target, setting the <constant>way</constant>
3299 variable. So if you say <command>gmake
3300 Foo.mp_o</command> you should see a recursive
3301 invocation <command>gmake Foo.mp_o way=mp</command>,
3302 and <emphasis>in this recursive invocation the pattern rule
3303 for compiling a Haskell file into a <filename>.o</filename>
3304 file will match</emphasis>. The key pattern rules (in
3305 <filename>suffix.mk</filename>) look like this:
3309 $(HC) $(HC_OPTS) $< -o $@
3316 <para>You can invoke <command>make</command> with a
3317 particular <literal>way</literal> setting yourself, in order
3318 to build files related to a particular
3319 <literal>way</literal> in the current directory. eg.
3325 will build files for the profiling way only in the current
3332 <title>When the canned rule isn't right</title>
3334 <para>Sometimes the canned rule just doesn't do the right thing.
3335 For example, in the <literal>nofib</literal> suite we want the
3336 link step to print out timing information. The thing to do here
3337 is <emphasis>not</emphasis> to define
3338 <constant>HS_PROG</constant> or
3339 <constant>C_PROG</constant>, and instead define a special
3340 purpose rule in your own <filename>Makefile</filename>. By
3341 using different variable names you will avoid the canned rules
3342 being included, and conflicting with yours.</para>
3346 <sect1 id="building-docs">
3347 <title>Building the documentation</title>
3349 <sect2 id="pre-supposed-doc-tools">
3350 <title>Tools for building the Documentation</title>
3352 <para>The following additional tools are required if you want to
3353 format the documentation that comes with the
3354 <literal>fptools</literal> projects:</para>
3358 <term>DocBook</term>
3359 <indexterm><primary>pre-supposed: DocBook</primary></indexterm>
3360 <indexterm><primary>DocBook, pre-supposed</primary></indexterm>
3362 <para>Much of our documentation is written in SGML, using
3363 the DocBook DTD. Instructions on installing and
3364 configuring the DocBook tools are below.</para>
3370 <indexterm><primary>pre-supposed: TeX</primary></indexterm>
3371 <indexterm><primary>TeX, pre-supposed</primary></indexterm>
3373 <para>A decent TeX distribution is required if you want to
3374 produce printable documentation. We recomment teTeX,
3375 which includes just about everything you need.</para>
3380 <term>Haddock</term>
3381 <indexterm><primary>Haddock</primary>
3384 <para>Haddock is a Haskell documentation tool that we use
3385 for automatically generating documentation from the
3386 library source code. It is an <literal>fptools</literal>
3387 project in itself. To build documentation for the
3388 libraries (<literal>fptools/libraries</literal>) you
3389 should check out and build Haddock in
3390 <literal>fptools/haddock</literal>. Haddock requires GHC
3398 <title>Installing the DocBook tools</title>
3401 <title>Installing the DocBook tools on Linux</title>
3403 <para>If you're on a recent RedHat system (7.0+), you probably
3404 have working DocBook tools already installed. The configure
3405 script should detect your setup and you're away.</para>
3407 <para>If you don't have DocBook tools installed, and you are
3408 using a system that can handle RedHat RPM packages, you can
3409 probably use the <ULink
3410 URL="http://sourceware.cygnus.com/docbook-tools/">Cygnus
3411 DocBook tools</ULink>, which is the most shrink-wrapped SGML
3412 suite that we could find. You need all the RPMs except for
3413 psgml (i.e. <Filename>docbook</Filename>,
3414 <Filename>jade</Filename>, <Filename>jadetex</Filename>,
3415 <Filename>sgmlcommon</Filename> and
3416 <Filename>stylesheets</Filename>). Note that most of these
3417 RPMs are architecture neutral, so are likely to be found in a
3418 <Filename>noarch</Filename> directory. The SuSE RPMs also
3419 work; the RedHat ones <Emphasis>don't</Emphasis> in RedHat 6.2
3420 (7.0 and later should be OK), but they are easy to fix: just
3422 <Filename>/usr/lib/sgml/stylesheets/nwalsh-modular/lib/dblib.dsl</Filename>
3423 to <Filename>/usr/lib/sgml/lib/dblib.dsl</Filename>. </para>
3427 <title>Installing DocBook on FreeBSD</title>
3429 <para>On FreeBSD systems, the easiest way to get DocBook up
3430 and running is to install it from the ports tree or a
3431 pre-compiled package (packages are available from your local
3432 FreeBSD mirror site).</para>
3434 <para>To use the ports tree, do this:
3436 $ cd /usr/ports/textproc/docproj
3439 This installs the FreeBSD documentation project tools, which
3440 includes everything needed to format the GHC
3441 documentation.</para>
3445 <title>Installing from binaries on Windows</title>
3447 <Para>It's a good idea to use Norman Walsh's <ULink
3448 URL="http://nwalsh.com/docbook/dsssl/doc/install.html">installation
3449 notes</ULink> as a guide. You should get version 3.1 of
3450 DocBook, and note that his file <Filename>test.sgm</Filename>
3451 won't work, as it needs version 3.0. You should unpack Jade
3452 into <Filename>\Jade</Filename>, along with the entities,
3453 DocBook into <Filename>\docbook</Filename>, and the DocBook
3454 stylesheets into <Filename>\docbook\stylesheets</Filename> (so
3455 they actually end up in
3456 <Filename>\docbook\stylesheets\docbook</Filename>).</para>
3461 <title>Installing the DocBook tools from source</title>
3466 <para>Install <ULink
3467 URL="http://openjade.sourceforge.net/">OpenJade</ULink>
3468 (Windows binaries are available as well as sources). If you
3469 want DVI, PS, or PDF then install JadeTeX from the
3470 <Filename>dsssl</Filename> subdirectory. (If you get the
3474 ! LaTeX Error: Unknown option implicit=false' for package hyperref'.
3477 your version of <Command>hyperref</Command> is out of date;
3478 download it from CTAN
3479 (<Filename>macros/latex/contrib/supported/hyperref</Filename>),
3480 and make it, ensuring that you have first removed or renamed
3481 your old copy. If you start getting file not found errors
3482 when making the test for <Command>hyperref</Command>, you
3483 can abort at that point and proceed straight to
3484 <Command>make install</Command>, or enter them as
3485 <Filename>../</Filename><Emphasis>filename</Emphasis>.)</para>
3487 <para>Make links from <Filename>virtex</Filename> to
3488 <Filename>jadetex</Filename> and
3489 <Filename>pdfvirtex</Filename> to
3490 <Filename>pdfjadetex</Filename> (otherwise DVI, PostScript
3491 and PDF output will not work). Copy
3492 <Filename>dsssl/*.{dtd,dsl}</Filename> and
3493 <Filename>catalog</Filename> to
3494 <Filename>/usr/[local/]lib/sgml</Filename>.</para>
3498 <title>DocBook and the DocBook stylesheets</title>
3500 <para>Get a Zip of <ULink
3501 URL="http://www.oasis-open.org/docbook/sgml/3.1/index.html">DocBook</ULink>
3502 and install the contents in
3503 <Filename>/usr/[local/]/lib/sgml</Filename>.</para>
3505 <para>Get the <ULink
3506 URL="http://nwalsh.com/docbook/dsssl/">DocBook
3507 stylesheets</ULink> and install in
3508 <Filename>/usr/[local/]lib/sgml/stylesheets</Filename>
3509 (thereby creating a subdirectory docbook). For indexing,
3510 copy or link <Filename>collateindex.pl</Filename> from the
3511 DocBook stylesheets archive in <Filename>bin</Filename> into
3512 a directory on your <Constant>PATH</Constant>.</para>
3514 <para>Download the <ULink
3515 URL="http://www.oasis-open.org/cover/ISOEnts.zip">ISO
3516 entities</ULink> into
3517 <Filename>/usr/[local/]lib/sgml</Filename>.</para>
3523 <title>Configuring the DocBook tools</title>
3525 <Para>Once the DocBook tools are installed, the configure script
3526 will detect them and set up the build system accordingly. If you
3527 have a system that isn't supported, let us know, and we'll try
3532 <title>Remaining problems</title>
3534 <para>If you install from source, you'll get a pile of warnings
3537 <Screen>DTDDECL catalog entries are not supported</Screen>
3539 every time you build anything. These can safely be ignored, but
3540 if you find them tedious you can get rid of them by removing all
3541 the <Constant>DTDDECL</Constant> entries from
3542 <Filename>docbook.cat</Filename>.</para>
3546 <title>Building the documentation</title>
3548 <para>To build documentation in a certain format, you can
3549 say, for example,</para>
3555 <para>to build HTML documentation below the current directory.
3556 The available formats are: <literal>dvi</literal>,
3557 <literal>ps</literal>, <literal>pdf</literal>,
3558 <literal>html</literal>, and <literal>rtf</literal>. Note that
3559 not all documentation can be built in all of these formats: HTML
3560 documentation is generally supported everywhere, and DocBook
3561 documentation might support the other formats (depending on what
3562 other tools you have installed).</para>
3564 <para>All of these targets are recursive; that is, saying
3565 <literal>make html</literal> will make HTML docs for all the
3566 documents recursively below the current directory.</para>
3568 <para>Because there are many different formats that the DocBook
3569 documentation can be generated in, you have to select which ones
3570 you want by setting the <literal>SGMLDocWays</literal> variable
3571 to a list of them. For example, in
3572 <filename>build.mk</filename> you might have a line:</para>
3575 SGMLDocWays = html ps
3578 <para>This will cause the documentation to be built in the requested
3579 formats as part of the main build (the default is not to build
3580 any documentation at all).</para>
3584 <title>Installing the documentation</title>
3586 <para>To install the documentation, use:</para>
3592 <para>This will install the documentation into
3593 <literal>$(datadir)</literal> (which defaults to
3594 <literal>$(prefix)/share</literal>). The exception is HTML
3595 documentation, which goes into
3596 <literal>$(datadir)/html</literal>, to keep things tidy.</para>
3598 <para>Note that unless you set <literal>$(SGMLDocWays)</literal>
3599 to a list of formats, the <literal>install-docs</literal> target
3600 won't do anything for SGML documentation.</para>
3606 <sect1 id="sec-porting-ghc">
3607 <title>Porting GHC</title>
3609 <para>This section describes how to port GHC to a currenly
3610 unsupported platform. There are two distinct
3611 possibilities:</para>
3615 <para>The hardware architecture for your system is already
3616 supported by GHC, but you're running an OS that isn't
3617 supported (or perhaps has been supported in the past, but
3618 currently isn't). This is the easiest type of porting job,
3619 but it still requires some careful bootstrapping. Proceed to
3620 <xref linkend="sec-booting-from-hc">.</para>
3624 <para>Your system's hardware architecture isn't supported by
3625 GHC. This will be a more difficult port (though by comparison
3626 perhaps not as difficult as porting gcc). Proceed to <xref
3627 linkend="unregisterised-porting">.</para>
3631 <sect2 id="sec-booting-from-hc">
3632 <title>Booting/porting from C (<filename>.hc</filename>) files</title>
3634 <indexterm><primary>building GHC from .hc files</primary></indexterm>
3635 <indexterm><primary>booting GHC from .hc files</primary></indexterm>
3636 <indexterm><primary>porting GHC</primary></indexterm>
3638 <para>Bootstrapping GHC on a system without GHC already
3639 installed is achieved by taking the intermediate C files (known
3640 as HC files) from a GHC compilation on a supported system to the
3641 target machine, and compiling them using gcc to get a working
3644 <para><emphasis>NOTE: GHC versions 5.xx were hard to bootstrap
3645 from C. We recommend using GHC 6.0.1 or
3646 later.</emphasis></para>
3648 <para>HC files are platform-dependent, so you have to get a set
3649 that were generated on similar hardware. There may be some
3650 supplied on the GHC download page, otherwise you'll have to
3651 compile some up yourself, or start from
3652 <emphasis>unregisterised</emphasis> HC files - see <xref
3653 linkend="unregisterised-porting">.</para>
3655 <para>The following steps should result in a working GHC build
3656 with full libraries:</para>
3660 <para>Unpack the HC files on top of a fresh source tree
3661 (make sure the source tree version matches the version of
3662 the HC files <emphasis>exactly</emphasis>!). This will
3663 place matching <filename>.hc</filename> files next to the
3664 corresponding Haskell source (<filename>.hs</filename> or
3665 <filename>.lhs</filename>) in the compiler subdirectory
3666 <filename>ghc/compiler</filename> and in the libraries
3667 (subdirectories of <filename>hslibs</filename> and
3668 <literal>libraries</literal>).</para>
3672 <para>The actual build process is fully automated by the
3673 <filename>hc-build</filename> script located in the
3674 <filename>distrib</filename> directory. If you eventually
3675 want to install GHC into the directory
3676 <replaceable>dir</replaceable>, the following
3677 command will execute the whole build process (it won't
3678 install yet):</para>
3681 foo% distrib/hc-build --prefix=<replaceable>dir</replaceable>
3683 <indexterm><primary>--hc-build</primary></indexterm>
3685 <para>By default, the installation directory is
3686 <filename>/usr/local</filename>. If that is what you want,
3687 you may omit the argument to <filename>hc-build</filename>.
3688 Generally, any option given to <filename>hc-build</filename>
3689 is passed through to the configuration script
3690 <filename>configure</filename>. If
3691 <filename>hc-build</filename> successfully completes the
3692 build process, you can install the resulting system, as
3702 <sect2 id="unregisterised-porting">
3703 <title>Porting GHC to a new architecture</title>
3705 <para>The first step in porting to a new architecture is to get
3706 an <firstterm>unregisterised</firstterm> build working. An
3707 unregisterised build is one that compiles via vanilla C only.
3708 By contrast, a registerised build uses the following
3709 architecture-specific hacks for speed:</para>
3713 <para>Global register variables: certain abstract machine
3714 <quote>registers</quote> are mapped to real machine
3715 registers, depending on how many machine registers are
3717 <filename>ghc/includes/MachRegs.h</filename>).</para>
3721 <para>Assembly-mangling: when compiling via C, we feed the
3722 assembly generated by gcc though a Perl script known as the
3723 <firstterm>mangler</firstterm> (see
3724 <filename>ghc/driver/mangler/ghc-asm.lprl</filename>). The
3725 mangler rearranges the assembly to support tail-calls and
3726 various other optimisations.</para>
3730 <para>In an unregisterised build, neither of these hacks are
3731 used — the idea is that the C code generated by the
3732 compiler should compile using gcc only. The lack of these
3733 optimisations costs about a factor of two in performance, but
3734 since unregisterised compilation is usually just a step on the
3735 way to a full registerised port, we don't mind too much.</para>
3737 <para>Notes on GHC portability in general: we've tried to stick
3738 to writing portable code in most parts of the system, so it
3739 should compile on any POSIXish system with gcc, but in our
3740 experience most systems differ from the standards in one way or
3741 another. Deal with any problems as they arise - if you get
3742 stuck, ask the experts on
3743 <email>glasgow-haskell-users@haskell.org</email>.</para>
3745 <para>Lots of useful information about the innards of GHC is
3746 available in the <ulink
3747 url="http://www.cse.unsw.edu.au/~chak/haskell/ghc/comm/">GHC
3748 Commentary</ulink>, which might be helpful if you run into some
3749 code which needs tweaking for your system.</para>
3752 <title>Cross-compiling to produce an unregisterised GHC</title>
3754 <para>In this section, we explain how to bootstrap GHC on a
3755 new platform, using unregisterised intermediate C files. We
3756 haven't put a great deal of effort into automating this
3757 process, for two reasons: it is done very rarely, and the
3758 process usually requires human intervention to cope with minor
3759 porting issues anyway.</para>
3761 <para>The following step-by-step instructions should result in
3762 a fully working, albeit unregisterised, GHC. Firstly, you
3763 need a machine that already has a working GHC (we'll call this
3764 the <firstterm>host</firstterm> machine), in order to
3765 cross-compile the intermediate C files that we will use to
3766 bootstrap the compiler on the <firstterm>target</firstterm>
3771 <para>On the target machine:</para>
3775 <para>Unpack a source tree (preferably a released
3776 version). We will call the path to the root of this
3777 tree <replaceable>T</replaceable>.</para>
3782 $ cd <replaceable>T</replaceable>
3783 $ ./configure --enable-hc-boot --enable-hc-boot-unregisterised
3786 <para>You might need to update
3787 <filename>configure.in</filename> to recognise the new
3788 architecture, and re-generate
3789 <filename>configure</filename> with
3790 <literal>autoreconf</literal>.</para>
3795 $ cd <replaceable>T</replaceable>/ghc/includes
3803 <para>On the host machine:</para>
3807 <para>Unpack a source tree (same released version). Call
3808 this directory <replaceable>H</replaceable>.</para>
3813 $ cd <replaceable>H</replaceable>
3820 <filename><replaceable>H</replaceable>/mk/build.mk</filename>,
3821 with the following contents:</para>
3824 GhcUnregisterised = YES
3825 GhcLibHcOpts = -O -H32m -keep-hc-files
3828 GhcWithNativeCodeGen = NO
3829 GhcWithInterpreter = NO
3830 GhcStage1HcOpts = -O -H32m -fasm
3831 GhcStage2HcOpts = -O -fvia-C -keep-hc-files
3837 <filename><replaceable>H</replaceable>/mk/config.mk</filename>:</para>
3840 <para>change <literal>TARGETPLATFORM</literal>
3841 appropriately, and set the variables involving
3842 <literal>TARGET</literal> to the correct values for
3843 the target platform. This step is necessary because
3844 currently <literal>configure</literal> doesn't cope
3845 with specifying different values for the
3846 <literal>--host</literal> and
3847 <literal>--target</literal> flags.</para>
3850 <para>copy <literal>LeadingUnderscore</literal>
3851 setting from target.</para>
3858 <filename><replaceable>T</replaceable>/ghc/includes/config.h</filename>
3860 <filename><replaceable>H</replaceable>/ghc/includes</filename>.
3861 Note that we are building on the host machine, using the
3862 target machine's <literal>config.h</literal> file. This
3863 is so that the intermediate C files generated here will
3864 be suitable for compiling on the target system.</para>
3869 <para>Touch <literal>config.h</literal>, just to make
3870 sure it doesn't get replaced during the build:</para>
3872 $ touch <replaceable>H</replaceable>/ghc/includes/config.h</screen>
3876 <para>Now build the compiler:</para>
3878 $ cd <replaceable>H</replaceable>/glafp-utils && make boot && make
3879 $ cd <replaceable>H</replaceable>/ghc && make boot && make
3881 <para>Don't worry if the build falls over in the RTS, we
3882 don't need the RTS yet.</para>
3887 $ cd <replaceable>H</replaceable>/libraries
3888 $& make boot && make
3894 $ cd <replaceable>H</replaceable>/ghc
3895 $ make boot stage=2 && make stage=2
3901 $ cd <replaceable>H</replaceable>/ghc/utils
3903 $ make -k HC=<replaceable>H</replaceable>/ghc/compiler/stage1/ghc-inplace \
3904 EXTRA_HC_OPTS='-O -fvia-C -keep-hc-files'
3910 $ cd <replaceable>H</replaceable>
3911 $ make hc-file-bundle Project=Ghc
3917 <filename><replaceable>H</replaceable>/*-hc.tar.gz</filename>
3918 to <filename><replaceable>T</replaceable>/..</filename>.</para>
3924 <para>On the target machine:</para>
3926 <para>At this stage we simply need to bootstrap a compiler
3927 from the intermediate C files we generated above. The
3928 process of bootstrapping from C files is automated by the
3929 script in <literal>distrib/hc-build</literal>, and is
3930 described in <xref linkend="sec-booting-from-hc">.</para>
3933 $ ./distrib/hc-build --enable-hc-boot-unregisterised
3936 <para>However, since this is a bootstrap on a new machine,
3937 the automated process might not run to completion the
3938 first time. For that reason, you might want to treat the
3939 <literal>hc-build</literal> script as a list of
3940 instructions to follow, rather than as a fully automated
3941 script. This way you'll be able to restart the process
3942 part-way through if you need to fix anything on the
3945 <para>Don't bother with running
3946 <literal>make install</literal> in the newly
3947 bootstrapped tree; just use the compiler in that tree to
3948 build a fresh compiler from scratch, this time without
3949 booting from C files. Before doing this, you might want
3950 to check that the bootstrapped compiler is generating
3951 working binaries:</para>
3955 main = putStrLn "Hello World!\n"
3957 $ <replaceable>T</replaceable>/ghc/compiler/ghc-inplace hello.hs -o hello
3962 <para>Once you have the unregisterised compiler up and
3963 running, you can use it to start a registerised port. The
3964 following sections describe the various parts of the
3965 system that will need architecture-specific tweaks in
3966 order to get a registerised build going.</para>
3973 <title>Porting the RTS</title>
3975 <para>The following files need architecture-specific code for a
3976 registerised build:</para>
3980 <term><filename>ghc/includes/MachRegs.h</filename></term>
3981 <indexterm><primary><filename>MachRegs.h</filename></primary>
3984 <para>Defines the STG-register to machine-register
3985 mapping. You need to know your platform's C calling
3986 convention, and which registers are generally available
3987 for mapping to global register variables. There are
3988 plenty of useful comments in this file.</para>
3992 <term><filename>ghc/includes/TailCalls.h</filename></term>
3993 <indexterm><primary><filename>TailCalls.h</filename></primary>
3996 <para>Macros that cooperate with the mangler (see <xref
3997 linkend="sec-mangler">) to make proper tail-calls
4002 <term><filename>ghc/rts/Adjustor.c</filename></term>
4003 <indexterm><primary><filename>Adjustor.c</filename></primary>
4007 <literal>foreign import "wrapper"</literal>
4009 <literal>foreign export dynamic</literal>).
4010 Not essential for getting GHC bootstrapped, so this file
4011 can be deferred until later if necessary.</para>
4015 <term><filename>ghc/rts/StgCRun.c</filename></term>
4016 <indexterm><primary><filename>StgCRun.c</filename></primary>
4019 <para>The little assembly layer between the C world and
4020 the Haskell world. See the comments and code for the
4021 other architectures in this file for pointers.</para>
4025 <term><filename>ghc/rts/MBlock.h</filename></term>
4026 <term><filename>ghc/rts/MBlock.c</filename></term>
4027 <indexterm><primary><filename>MBlock.h</filename></primary>
4029 <indexterm><primary><filename>MBlock.c</filename></primary>
4032 <para>These files are really OS-specific rather than
4033 architecture-specific. In <filename>MBlock.h</filename>
4034 is specified the absolute location at which the RTS
4035 should try to allocate memory on your platform (try to
4036 find an area which doesn't conflict with code or dynamic
4037 libraries). In <filename>Mblock.c</filename> you might
4038 need to tweak the call to <literal>mmap()</literal> for
4045 <sect3 id="sec-mangler">
4046 <title>The mangler</title>
4048 <para>The mangler is an evil Perl-script that rearranges the
4049 assembly code output from gcc to do two main things:</para>
4053 <para>Remove function prologues and epilogues, and all
4054 movement of the C stack pointer. This is to support
4055 tail-calls: every code block in Haskell code ends in an
4056 explicit jump, so we don't want the C-stack overflowing
4057 while we're jumping around between code blocks.</para>
4060 <para>Move the <firstterm>info table</firstterm> for a
4061 closure next to the entry code for that closure. In
4062 unregisterised code, info tables contain a pointer to the
4063 entry code, but in registerised compilation we arrange
4064 that the info table is shoved right up against the entry
4065 code, and addressed backwards from the entry code pointer
4066 (this saves a word in the info table and an extra
4067 indirection when jumping to the closure entry
4072 <para>The mangler is abstracted to a certain extent over some
4073 architecture-specific things such as the particular assembler
4074 directives used to herald symbols. Take a look at the
4075 definitions for other architectures and use these as a
4076 starting point.</para>
4080 <title>The native code generator</title>
4082 <para>The native code generator isn't essential to getting a
4083 registerised build going, but it's a desirable thing to have
4084 because it can cut compilation times in half. The native code
4085 generator is described in some detail in the <ulink
4086 url="http://www.cse.unsw.edu.au/~chak/haskell/ghc/comm/">GHC
4087 commentary</ulink>.</para>
4093 <para>To support GHCi, you need to port the dynamic linker
4094 (<filename>fptools/ghc/rts/Linker.c</filename>). The linker
4095 currently supports the ELF and PEi386 object file formats - if
4096 your platform uses one of these then things will be
4097 significantly easier. The majority of Unix platforms use the
4098 ELF format these days. Even so, there are some
4099 machine-specific parts of the ELF linker: for example, the
4100 code for resolving particular relocation types is
4101 machine-specific, so some porting of this code to your
4102 architecture will probaly be necessary.</para>
4104 <para>If your system uses a different object file format, then
4105 you have to write a linker — good luck!</para>
4111 <sect1 id="sec-build-pitfalls">
4112 <title>Known pitfalls in building Glasgow Haskell
4114 <indexterm><primary>problems, building</primary></indexterm>
4115 <indexterm><primary>pitfalls, in building</primary></indexterm>
4116 <indexterm><primary>building pitfalls</primary></indexterm></title>
4119 WARNINGS about pitfalls and known “problems”:
4128 One difficulty that comes up from time to time is running out of space
4129 in <literal>TMPDIR</literal>. (It is impossible for the configuration stuff to
4130 compensate for the vagaries of different sysadmin approaches to temp
4132 <indexterm><primary>tmp, running out of space in</primary></indexterm>
4134 The quickest way around it is <command>setenv TMPDIR /usr/tmp</command><indexterm><primary>TMPDIR</primary></indexterm> or
4135 even <command>setenv TMPDIR .</command> (or the equivalent incantation with your shell
4138 The best way around it is to say
4141 export TMPDIR=<dir>
4144 in your <filename>build.mk</filename> file.
4145 Then GHC and the other <literal>fptools</literal> programs will use the appropriate directory
4154 In compiling some support-code bits, e.g., in <filename>ghc/rts/gmp</filename> and even
4155 in <filename>ghc/lib</filename>, you may get a few C-compiler warnings. We think these
4163 When compiling via C, you'll sometimes get “warning: assignment from
4164 incompatible pointer type” out of GCC. Harmless.
4171 Similarly, <command>ar</command>chiving warning messages like the following are not
4175 ar: filename GlaIOMonad__1_2s.o truncated to GlaIOMonad_
4176 ar: filename GlaIOMonad__2_2s.o truncated to GlaIOMonad_
4186 In compiling the compiler proper (in <filename>compiler/</filename>), you <emphasis>may</emphasis>
4187 get an “Out of heap space” error message. These can vary with the
4188 vagaries of different systems, it seems. The solution is simple:
4195 If you're compiling with GHC 4.00 or later, then the
4196 <emphasis>maximum</emphasis> heap size must have been reached. This
4197 is somewhat unlikely, since the maximum is set to 64M by default.
4198 Anyway, you can raise it with the
4199 <option>-optCrts-M<size></option> flag (add this flag to
4200 <constant><module>_HC_OPTS</constant>
4201 <command>make</command> variable in the appropriate
4202 <filename>Makefile</filename>).
4209 For GHC < 4.00, add a suitable <option>-H</option> flag to the <filename>Makefile</filename>, as
4218 and try again: <command>gmake</command>. (see <Xref LinkEnd="sec-suffix"> for information about
4219 <constant><module>_HC_OPTS</constant>.)
4221 Alternatively, just cut to the chase:
4225 % make EXTRA_HC_OPTS=-optCrts-M128M
4234 If you try to compile some Haskell, and you get errors from GCC about
4235 lots of things from <filename>/usr/include/math.h</filename>, then your GCC was
4236 mis-installed. <command>fixincludes</command> wasn't run when it should've been.
4238 As <command>fixincludes</command> is now automagically run as part of GCC installation,
4239 this bug also suggests that you have an old GCC.
4247 You <emphasis>may</emphasis> need to re-<command>ranlib</command><indexterm><primary>ranlib</primary></indexterm> your libraries (on Sun4s).
4251 % cd $(libdir)/ghc-x.xx/sparc-sun-sunos4
4252 % foreach i ( `find . -name '*.a' -print` ) # or other-shell equiv...
4254 ? # or, on some machines: ar s $i
4259 We'd be interested to know if this is still necessary.
4267 GHC's sources go through <command>cpp</command> before being compiled, and <command>cpp</command> varies
4268 a bit from one Unix to another. One particular gotcha is macro calls
4273 SLIT("Hello, world")
4277 Some <command>cpp</command>s treat the comma inside the string as separating two macro
4278 arguments, so you get
4282 :731: macro `SLIT' used with too many (2) args
4286 Alas, <command>cpp</command> doesn't tell you the offending file!
4288 Workaround: don't put weird things in string args to <command>cpp</command> macros.
4299 <Sect1 id="winbuild"><Title>Notes for building under Windows</Title>
4302 This section summarises how to get the utilities you need on your
4303 Win95/98/NT/2000 machine to use CVS and build GHC. Similar notes for
4304 installing and running GHC may be found in the user guide. In general,
4305 Win95/Win98 behave the same, and WinNT/Win2k behave the same.
4306 You should read the GHC installation guide sections on Windows (in the user
4307 guide) before continuing to read these notes.
4311 <sect2 id="cygwin-and-mingw"><Title>Cygwin and MinGW</Title>
4313 <para> The Windows situation for building GHC is rather confusing. This section
4314 tries to clarify, and to establish terminology.</para>
4316 <sect3 id="ghc-mingw"><title>GHC-mingw</title>
4318 <para> <ulink url="http://www.mingw.org">MinGW (Minimalist GNU for Windows)</ulink>
4319 is a collection of header
4320 files and import libraries that allow one to use <command>gcc</command> and produce
4321 native Win32 programs that do not rely on any third-party DLLs. The
4322 current set of tools include GNU Compiler Collection (<command>gcc</command>), GNU Binary
4323 Utilities (Binutils), GNU debugger (Gdb), GNU make, and a assorted
4326 <para>The GHC that we distribute includes, inside the distribution itself, the MinGW <command>gcc</command>,
4327 <command>as</command>, <command>ld</command>, and a bunch of input/output libraries.
4328 GHC compiles Haskell to C (or to
4329 assembly code), and then invokes these MinGW tools to generate an executable binary.
4330 The resulting binaries can run on any Win32 system.
4332 <para> We will call a GHC that targets MinGW in this way <emphasis>GHC-mingw</emphasis>.</para>
4334 <para> The down-side of GHC-mingw is that the MinGW libraries do not support anything like the full
4335 Posix interface. So programs compiled with GHC-mingw cannot import the (Haskell) Posix
4336 library; they have to do
4337 their input output using standard Haskell I/O libraries, or native Win32 bindings.
4341 <sect3 id="ghc-cygwin"><title>GHC-cygwin</title>
4343 <para>There <emphasis>is</emphasis> a way to get the full Posix interface, which is to use Cygwin.
4344 <ulink url="http://www.cygwin.com">Cygwin</ulink> is a complete Unix simulation that runs on Win32.
4345 Cygwin comes with a shell, and all the usual Unix commands: <command>mv</command>, <command>rm</command>,
4346 <command>ls</command>, plus of course <command>gcc</command>, <command>ld</command> and so on.
4347 A C program compiled with the Cygwin <command>gcc</command> certainly can use all of Posix.
4349 <para>So why doesn't GHC use the Cygwin <command>gcc</command> and libraries? Because
4350 Cygwin comes with a DLL <emphasis>that must be linked with every runnable Cygwin-compiled program</emphasis>.
4351 A program compiled by the Cygwin tools cannot run at all unless Cygwin is installed.
4352 If GHC targeted Cygwin, users would have to install Cygwin just to run the Haskell programs
4353 that GHC compiled; and the Cygwin DLL would have to be in the DLL load path.
4354 Worse, Cygwin is a moving target. The name of the main DLL, <literal>cygwin1.dll</literal>
4355 does not change, but the implementation certainly does. Even the interfaces to functions
4356 it exports seem to change occasionally. So programs compiled by GHC might only run with
4357 particular versions of Cygwin. All of this seems very undesirable.
4360 Nevertheless, it is certainly possible to build a version of GHC that targets Cygwin;
4361 we will call that <emphasis>GHC-cygwin</emphasis>. The up-side of GHC-cygwin is
4362 that Haskell programs compiled by GHC-cygwin can import the (Haskell) Posix library.
4366 <sect3><title>HOST_OS vs TARGET_OS</title>
4369 In the source code you'll find various ifdefs looking like:
4371 #ifdef mingw32_HOST_OS
4377 #ifdef mingw32_TARGET_OS
4381 These macros are set by the configure script (via the file config.h).
4382 Which is which? The criterion is this. In the ifdefs in GHC's source code:
4385 The "host" system is the one on which GHC itself will be run.
4388 The "target" system is the one for which the program compiled by GHC will be run.
4391 For a stage-2 compiler, in which GHCi is available, the "host" and "target" systems must be the same.
4392 So then it doesn't really matter whether you use the HOST_OS or TARGET_OS cpp macros.
4397 <sect3><title>Summary</title>
4399 <para>Notice that "GHC-mingw" means "GHC that <emphasis>targets</emphasis> MinGW". It says nothing about
4400 how that GHC was <emphasis>built</emphasis>. It is entirely possible to have a GHC-mingw that was built
4401 by compiling GHC's Haskell sources with a GHC-cygwin, or vice versa.</para>
4403 <para>We distribute only a GHC-mingw built by a GHC-mingw; supporting
4404 GHC-cygwin too is beyond our resources. The GHC we distribute
4405 therefore does not require Cygwin to run, nor do the programs it
4406 compiles require Cygwin.</para>
4408 <para>The instructions that follow describe how to build GHC-mingw. It is
4409 possible to build GHC-cygwin, but it's not a supported route, and the build system might
4412 <para>In your build tree, you build a compiler called <Command>ghc-inplace</Command>. It
4413 uses the <Command>gcc</Command> that you specify using the
4414 <option>--with-gcc</option> flag when you run
4415 <Command>configure</Command> (see below).
4416 The makefiles are careful to use <Command>ghc-inplace</Command> (not <Command>gcc</Command>)
4417 to compile any C files, so that it will in turn invoke the right <Command>gcc</Command> rather that
4418 whatever one happens to be in your path. However, the makefiles do use whatever <Command>ld</Command>
4419 and <Command>ar</Command> happen to be in your path. This is a bit naughty, but (a) they are only
4420 used to glom together .o files into a bigger .o file, or a .a file,
4421 so they don't ever get libraries (which would be bogus; they might be the wrong libraries), and (b)
4422 Cygwin and Mingw use the same .o file format. So its ok.
4427 <Sect2><Title>Installing and configuring Cygwin</Title>
4429 <para>You don't need Cygwin to <emphasis>use</emphasis> GHC,
4430 but you do need it to <emphasis>build</emphasis> GHC.</para>
4432 <para> Install Cygwin from <ulink url="http://www.cygwin.com/">http://www.cygwin.com/</ulink>.
4433 The installation process is straightforward; we install it in <Filename>c:/cygwin</Filename>.
4434 During the installation dialogue, make sure that you select:
4435 <command>cvs</command>, <command>openssh</command>,
4436 <command>autoconf</command>,
4437 <command>binutils</command> (includes ld and (I think) ar),
4438 <command>gcc</command>,
4439 <command>flex</command>,
4440 <command>make</command>.
4443 <para> Now set the following user environment variables:
4446 <listitem><para> Add <filename>c:/cygwin/bin</filename> and <filename>c:/cygwin/usr/bin</filename> to your
4447 <constant>PATH</constant></para></listitem>
4451 Set <constant>MAKE_MODE</constant> to <Literal>UNIX</Literal>. If you
4452 don't do this you get very weird messages when you type
4453 <Command>make</Command>, such as:
4455 /c: /c: No such file or directory
4460 <listitem><para> Set <constant>SHELL</constant> to
4461 <Filename>c:/cygwin/bin/sh</Filename>. When you invoke a shell in Emacs, this
4462 <constant>SHELL</constant> is what you get.
4465 <listitem><para> Set <constant>HOME</constant> to point to your
4466 home directory. This is where, for example,
4467 <command>bash</command> will look for your <filename>.bashrc</filename>
4468 file. Ditto <command>emacs</command> looking for <filename>.emacsrc</filename>
4474 There are a few other things to do:
4478 By default, cygwin provides the command shell <filename>ash</filename>
4479 as <filename>sh.exe</filename>. We have often seen build-system problems that
4480 turn out to be due to bugs in <filename>ash</filename>
4482 and length of command lines). On the other hand <filename>bash</filename> seems
4484 So, in <filename>cygwin/bin</filename>
4485 remove the supplied <filename>sh.exe</filename> (or rename it as <filename>ash.exe</filename>),
4486 and copy <filename>bash.exe</filename> to <filename>sh.exe</filename>.
4487 You'll need to do this in Windows Explorer or the Windows <command>cmd</command> shell, because
4488 you can't rename a running program!
4494 Some script files used in the make system start with "<Command>#!/bin/perl</Command>",
4495 (and similarly for <Command>sh</Command>). Notice the hardwired path!
4496 So you need to ensure that your <Filename>/bin</Filename> directory has the following
4499 <listitem> <para><Command>sh</Command></para></listitem>
4500 <listitem> <para><Command>perl</Command></para></listitem>
4501 <listitem> <para><Command>cat</Command></para></listitem>
4503 All these come in Cygwin's <Filename>bin</Filename> directory, which you probably have
4504 installed as <Filename>c:/cygwin/bin</Filename>. By default Cygwin mounts "<Filename>/</Filename>" as
4505 <Filename>c:/cygwin</Filename>, so if you just take the defaults it'll all work ok.
4506 (You can discover where your Cygwin
4507 root directory <Filename>/</Filename> is by typing <Command>mount</Command>.)
4508 Provided <Filename>/bin</Filename> points to the Cygwin <Filename>bin</Filename>
4509 directory, there's no need to copy anything. If not, copy these binaries from the <filename>cygwin/bin</filename>
4510 directory (after fixing the <filename>sh.exe</filename> stuff mentioned in the previous bullet).
4516 <para>Finally, here are some things to be aware of when using Cygwin:
4518 <listitem> <para>Cygwin doesn't deal well with filenames that include
4519 spaces. "<filename>Program Files</filename>" and "<filename>Local files</filename>" are
4523 <listitem> <para> Cygwin implements a symbolic link as a text file with some
4524 magical text in it. So other programs that don't use Cygwin's
4525 I/O libraries won't recognise such files as symlinks.
4526 In particular, programs compiled by GHC are meant to be runnable
4527 without having Cygwin, so they don't use the Cygwin library, so
4528 they don't recognise symlinks.
4532 Win32 has a <command>find</command> command which is not the same as Cygwin's find.
4533 You will probably discover that the Win32 <command>find</command> appears in your <constant>PATH</constant>
4534 before the Cygwin one, because it's in the <emphasis>system</emphasis> <constant>PATH</constant>
4535 environment variable, whereas you have probably modified the <emphasis>user</emphasis> <constant>PATH</constant>
4536 variable. You can always invoke <command>find</command> with an absolute path, or rename it.
4543 <Sect2 id="configure-ssh"><Title>Configuring SSH</Title>
4545 <para><command>ssh</command> comes with Cygwin, provided you remember to ask for it when
4546 you install Cygwin. (If not, the installer lets you update easily.) Look for <command>openssh</command>
4547 (not ssh) in the Cygwin list of applications!</para>
4549 <para>There are several strange things about <command>ssh</command> on Windows that you need to know.
4553 The programs <command>ssh-keygen1</command>, <command>ssh1</command>, and <command>cvs</command>,
4554 seem to lock up <command>bash</command> entirely if they try to get user input (e.g. if
4555 they ask for a password). To solve this, start up <filename>cmd.exe</filename>
4556 and run it as follows:
4558 c:\tmp> set CYGWIN32=tty
4559 c:\tmp> c:/user/local/bin/ssh-keygen1
4564 <command>ssh</command> needs to access your directory <filename>.ssh</filename>, in your home directory.
4565 To determine your home directory <command>ssh</command> first looks in
4566 <filename>c:/cygwin/etc/passwd</filename> (or wherever you have Cygwin installed). If there's an entry
4567 there with your userid, it'll use that entry to determine your home directory, <emphasis>ignoring
4568 the setting of the environment variable $HOME</emphasis>. If the home directory is
4569 bogus, <command>ssh</command> fails horribly. The best way to see what is going on is to say
4571 ssh -v cvs.haskell.org
4573 which makes <command>ssh</command> print out information about its activity.
4575 <para> You can fix this problem, either by correcting the home-directory field in
4576 <filename>c:/cygwin/etc/passwd</filename>, or by simply deleting the entire entry for your userid. If
4577 you do that, <command>ssh</command> uses the $HOME environment variable instead.
4583 <para>To protect your
4584 <literal>.ssh</literal> from access by anyone else,
4585 right-click your <literal>.ssh</literal> directory, and
4586 select <literal>Properties</literal>. If you are not on
4587 the access control list, add yourself, and give yourself
4588 full permissions (the second panel). Remove everyone else
4589 from the access control list. Don't leave them there but
4590 deny them access, because 'they' may be a list that
4591 includes you!</para>
4595 <para>In fact <command>ssh</command> 3.6.1 now seems to <emphasis>require</emphasis>
4596 you to have Unix permissions 600 (read/write for owner only)
4597 on the <literal>.ssh/identity</literal> file, else it
4598 bombs out. For your local C drive, it seems that <literal>chmod 600 identity</literal> works,
4599 but on Windows NT/XP, it doesn't work on a network drive (exact dteails obscure).
4600 The solution seems to be to set the $CYGWIN environment
4601 variable to "<literal>ntsec neta</literal>". The $CYGWIN environment variable is discussed
4602 in <ulink url="http://cygwin.com/cygwin-ug-net/using-cygwinenv.html">the Cygwin User's Guide</ulink>,
4603 and there are more details in <ulink url="http://cygwin.com/faq/faq_4.html#SEC44">the Cygwin FAQ</ulink>.
4610 <Sect2><Title>Other things you need to install</Title>
4612 <para>You have to install the following other things to build GHC:
4616 Install an executable GHC, from <ulink url="http://www.haskell.org/ghc">http://www.haskell.org/ghc</ulink>.
4617 This is what you will use to compile GHC. Add it in your
4618 <constant>PATH</constant>: the installer tells you the path element
4619 you need to add upon completion.
4625 Install an executable Happy, from <ulink url="http://www.haskell.org/happy">http://www.haskell.org/happy</ulink>.
4626 Happy is a parser generator used to compile the Haskell grammar. Add it in your
4627 <constant>PATH</constant>.
4632 <para>Install Alex. This can be done by building from the
4633 source distribution in the usual way. Sources are
4634 available from <ulink
4635 url="http://www.haskell.org/alex">http://www.haskell.org/alex</ulink>.</para>
4639 <para>GHC uses the <emphasis>mingw</emphasis> C compiler to
4640 generate code, so you have to install that (see <xref linkend="cygwin-and-mingw">).
4641 Just pick up a mingw bundle at
4642 <ulink url="http://www.mingw.org/">http://www.mingw.org/</ulink>.
4643 We install it in <filename>c:/mingw</filename>.
4645 <para>Do <emphasis>not</emphasis> add any of the <emphasis>mingw</emphasis> binaries to your path.
4646 They are only going to get used by explicit access (via the --with-gcc flag you
4647 give to <Command>configure</Command> later). If you do add them to your path
4648 you are likely to get into a mess because their names overlap with Cygwin binaries.
4654 <para>We use <command>emacs</command> a lot, so we install that too.
4655 When you are in <filename>fptools/ghc/compiler</filename>, you can use
4656 "<literal>make tags</literal>" to make a TAGS file for emacs. That uses the utility
4657 <filename>fptools/ghc/utils/hasktags/hasktags</filename>, so you need to make that first.
4658 The most convenient way to do this is by going <literal>make boot</literal> in <filename>fptools/ghc</filename>.
4659 The <literal>make tags</literal> command also uses <command>etags</command>, which comes with <command>emacs</command>,
4660 so you will need to add <filename>emacs/bin</filename> to your <literal>PATH</literal>.
4666 <para> Finally, check out a copy of GHC sources from
4667 the CVS repository, following the instructions above (<xref linkend="cvs-access">).
4674 <Sect2><Title>Building GHC</Title>
4677 Now go read the documentation above on building from source (<xref linkend="sec-building-from-source">);
4678 the bullets below only tell
4679 you about Windows-specific wrinkles.</para>
4683 Run <Command>autoconf</Command> both in <filename>fptools</filename>
4684 and in <filename>fptools/ghc</filename>. If you omit the latter step you'll
4685 get an error when you run <filename>./configure</filename>:
4688 creating mk/config.h
4689 mk/config.h is unchanged
4691 running /bin/sh ./configure --cache-file=.././config.cache --srcdir=.
4692 ./configure: ./configure: No such file or directory
4693 configure: error: ./configure failed for ghc
4698 <listitem> <para><command>autoconf</command> seems to create the file <filename>configure</filename>
4699 read-only. So if you need to run autoconf again (which I sometimes do for safety's sake),
4702 /usr/bin/autoconf: cannot create configure: permission denied
4704 Solution: delete <filename>configure</filename> first.
4709 You either need to add <filename>ghc</filename> to your
4710 <constant>PATH</constant> before you invoke
4711 <Command>configure</Command>, or use the <Command>configure</Command>
4712 option <option>--with-ghc=c:/ghc/ghc-some-version/bin/ghc</option>.
4717 If you are paranoid, delete <filename>config.cache</filename> if it exists.
4718 This file occasionally remembers out-of-date configuration information, which
4719 can be really confusing.
4725 After <command>autoconf</command> run <command>./configure</command> in
4726 <filename>fptools/</filename> thus:
4729 ./configure --host=i386-unknown-mingw32 --with-gcc=c:/mingw/bin/gcc
4731 This is the point at which you specify that you are building GHC-mingw
4732 (see <xref linkend="ghc-mingw">). </para>
4734 <para> Both these options are important! It's possible to get into
4735 trouble using the wrong C compiler!</para>
4737 Furthermore, it's <emphasis>very important</emphasis> that you specify a
4738 full MinGW path for <command>gcc</command>, not a Cygwin path, because GHC (which
4739 uses this path to invoke <command>gcc</command>) is a MinGW program and won't
4740 understand a Cygwin path. For example, if you
4741 say <literal>--with-gcc=/mingw/bin/gcc</literal>, it'll be interpreted as
4742 <filename>/cygdrive/c/mingw/bin/gcc</filename>, and GHC will fail the first
4743 time it tries to invoke it. Worse, the failure comes with
4744 no error message whatsoever. GHC simply fails silently when first invoked,
4745 typically leaving you with this:
4747 make[4]: Leaving directory `/cygdrive/e/fptools-stage1/ghc/rts/gmp'
4748 ../../ghc/compiler/ghc-inplace -optc-mno-cygwin -optc-O
4749 -optc-Wall -optc-W -optc-Wstrict-prototypes -optc-Wmissing-prototypes
4750 -optc-Wmissing-declarations -optc-Winline -optc-Waggregate-return
4751 -optc-Wbad-function-cast -optc-Wcast-align -optc-I../includes
4752 -optc-I. -optc-Iparallel -optc-DCOMPILING_RTS
4753 -optc-fomit-frame-pointer -O2 -static
4754 -package-name rts -O -dcore-lint -c Adjustor.c -o Adjustor.o
4755 make[2]: *** [Adjustor.o] Error 1
4756 make[1]: *** [all] Error 1
4757 make[1]: Leaving directory `/cygdrive/e/fptools-stage1/ghc'
4758 make: *** [all] Error 1
4764 If you want to build GHC-cygwin (<xref linkend="ghc-cygwin">)
4765 you'll have to do something more like:
4767 ./configure --with-gcc=...the Cygwin gcc...
4772 <listitem><para> You almost certainly want to set
4776 in your <filename>build.mk</filename> configuration file (see <xref linkend="sec-build-config">).
4777 This tells the build system not to split each library into a myriad of little object files, one
4778 for each function. Doing so reduces binary sizes for statically-linked binaries, but on Windows
4779 it dramatically increases the time taken to build the libraries in the first place.
4783 <listitem><para> Do not attempt to build the documentation.
4784 It needs all kinds of wierd Jade stuff that we haven't worked out for
4785 Win32.</para></listitem>