Getting started with hacking on GHC ----------------------------------- So you've decided to hack on GHC, congratulations! We hope you have a rewarding experience. This file contains a few nuggets of information that will help get you started right away, and point you in the direction of more comprehensive documentation for later. Setting up your build --------------------- The GHC build tree is set up so that, by default, it builds a compiler ready for installing and using. That means full optimisation, and the build can take a *long* time. If you unpack your source tree and right away say "./configure; make", expect to have to wait a while. For hacking, you want the build to be quick - quick to build in the first place, and quick to rebuild after making changes. Tuning your build setup can make the difference between several hours to build GHC, and less than an hour. Here's how to do it. mk/build.mk is a GNU makefile that contains all your build settings. By default, this file doesn't exist, and all the parameters are set to their defaults in mk/config.mk (mk/config.mk is the place to look for *all* the things you might want to tune). A good mk/build.mk to start hacking on GHC is: ------ SRC_HC_OPTS = -H32m -O -fasm -Rghc-timing GhcStage1HcOpts = -O0 -DDEBUG GhcLibHcOpts = -O -fgenerics GhcLibWays = SplitObjs = NO ------ What do these options do? SRC_HC_OPTS = -H32m -O -fasm -Rghc-timing These options are added to the command line for all Haskell compilations. We turn on -fasm, because that halves compilation time at the expense of a few percent performance. -Rghc-timing prints out a line of timing info about each compilation. It's handy to keep an eye on. GhcStage1HcOpts = -O0 -DDEBUG The options for building the stage1 compiler (these come after SRC_HC_OPTS, so you can override settings from there). We turn off optimisation here, assuming you'll be modifying and testing stage1. With optimisation off, rebuilding GHC after modifying it will be *much* quicker, not only because the individual compilations will be quicker, but also there will be fewer dependencies between modules, so less stuff needs to be rebuilt after each modification. Also we turn on -DDEBUG, because that enables assertions and debugging code in the compiler itself. Turning on DEBUG makes the compiler about 30% slower. GhcLibHcOpts = -O -fgenerics You almost certainly want optimisation *on* when building libraries, otherwise the code you build with this compiler goes really slowly. -fgenerics add generics support to the libraries - you can turn this off if you like (it'll make the libraries a bit smaller), but you won't be able to use Generics in the code you build against these libraries. GhcLibWays = Normally the profiled libs are built. Setting GhcLibWays to empty disables this, so you only build the normal libs. SplitObjs = NO Object splitting causes each module to be split into smaller pieces in the final library, to reduce executable sizes when linking against the library. It can be quite time and memory-consuming, so turn it off when you're hacking. Actually building the bits -------------------------- To just build everything, from the top level: $ autoreconf $ ./configure $ make $ make install Building individual parts of the tree ------------------------------------- The first thing to understand is that the source tree is built in two passes. First 'make boot' builds dependencies and any other tools required as part of the build itself. For example, ghc/utils/genprimopcode is built as part of 'make boot', because it is required to preprocess ghc/compiler/prelude/primops.txt.pp. After 'make boot', 'make' will build everything. If you say 'make' from the very top-level, the build system will arrange to do the appropriate 'make boot' steps for you. If you just want to build in a subdirectory (eg. ghc), you have to do 'make boot' yourself. You don't need to 'make boot' after every single change, but you might want to do it to update dependencies, for example. Refining the setup ------------------ If you will be hacking mostly on libraries, then you probably want to build stage1 with optimisation, because you're only building it once but using it many times. GhcStage1HcOpts = -O If you are working on GHCi or Template Haskell, then you will be building and modifying the stage 2 compiler. Hence, you want to build stage 1 with, and stage 2 without, optimisation. GhcStage1HcOpts = -O GhcStage2HcOpts = -O0 -DDEBUG Take a look through mk/config.mk for more settings you might want to override in build.mk. Remember: don't modify config.mk directly (it gets overwritten when you run ./configure). Full optimisation ----------------- To turn up everything to the max, for running performance tests for example, try theses: SRC_HC_OPTS = -H64m -O2 GhcLibHcOpts = -O2 SplitObjs = YES You can even add some more aggresive options, such as -fliberate-case-threshold50, -funfolding-use-threshold50. Roadmap ------- A rough roadmap to the source tree: distrib materials for building distributions docs build system documentation ghc The GHC Compiler rts the runtime system and storage manager lib libraries used in GHC and its tools utils tools that come with GHC, and tools used in the build compiler the compiler itself driver various scripts, and package databases docs compiler documentation includes header files shipped with GHC glafp-utils tools for the build system libraries The hierarchical libraries nofib A benchmark suite testsuite The regression test suite Resources --------- The GHC Developer's Wiki The home for GHC Developers, with information on accessing the latest sources, the bug tracker, and further documentation on the code. http://hackage.haskell.org/trac/ghc The Building Guide Full documentation on the build system. http://www.haskell.org/ghc/docs/latest/html/building/index.html The GHC Commentary Notes on the internals and architecture of GHC. Much of this isn't up to date, but there is still lots of useful stuff in there. Read in conjunction with the source code. http://www.cse.unsw.edu.au/~chak/haskell/ghc/comm/ Mailing lists Ask on glasgow-haskell-users@haskell.org if you have difficulties. If you're working with the current CVS sources of GHC, then cvs-ghc@haskell.org might be a more appropriate (developers hang out here). See http://www.haskell.org/mailman/listinfo for subscription. Happy Hacking! --The GHC Team