Commit 1fb1ab5d authored by simonpj's avatar simonpj

[project @ 1997-03-14 07:52:06 by simonpj]

Major update to more-or-less 2.02
parent fa6fb09e
The Glasgow Haskell Compiler -- version 2.02
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
We are pleased to announce the first release of the Glasgow Haskell
Compiler (GHC, version 2.02) for *Haskell 1.4*. Sources and binaries
are freely available by anonymous FTP and on the World-Wide Web;
details below.
Haskell is "the" standard lazy functional programming language; the
current language version is 1.3, agreed in May, 1996. The Haskell
Report is online at
http://haskell.cs.yale.edu/1.4/haskell-report.html
GHC 2.02 is a beta-quality release:
* It is reliable.
It has been extensively tested against a large suite of Haskell 1.2
programs, but not so extensively tested against Haskell 1.4 programs
because we don't have a comprehensive set (Donations of Haskell 1.4
programs to our test suite are most welcome).
* It should generate good code.
All the optimisations that GHC 0.29 used to do are back in, with
the exception of specialisation. It ought to be the case that
GHC 2.02 outperforms GHC 0.29, because it has a much better
handle on cross-module inlining, but there's a good chance that
there are performance "holes" lurking. We have yet to make
a systematic comparison. (Please send us programs where 2.02
does noticeably worse than 0.29.)
* It is more expensive than it should be.
GHC 2.02 has received even less attention to its own performance.
At present it eats more space and time than GHC 0.29, especially
for very small programs. We'll work on this.
* A couple of Haskell 1.4 features are incompletely supported,
notably polymorphic strictness annotations, and Unicode.
If you want to use Haskell 1.4, this is a good moment to switch. If
you don't need the Haskell 1.4 extensions, then stay with GHC 0.29.
If you want to hack on GHC itself, then 2.02 is definitely for you.
The release notes comment further on this point.
GHC 2.02 is substantially changed from 2.01. Changes worth noting
include:
* The whole front end, which deals with the module system, has
been rewritten. The interface file format has changed.
* GHC 2.02 comes complete with Green Card, a C foreign language
interface for GHC. Green card is a pre-processor that
scans Haskell source files for Green Card directives, which
it expands into tons of "ccall" boilerplate that marshalls
your arguments to and from C.
* GHC 2.02 is available for Windows NT. From now on, Windows NT
will be a fully supported platform for GHC.
* GHC 2.02 supports full cross moudule inlining. Unlike 0.29 and
its predecessors, inlining can happen even if the inlined body
mentions a function or type that is not itself exported. This is
one place Haskell 1.4's new module system really pays off.
* Like 2.01, GHC 2.02 aborts a compilation if it decides that
nothing that the module imports *and acually uses* has changed.
This decision is now taken by the compiler itself, rather than
by a Perl script (as in 2.01) which sometimes got it wrong.
* The ghc/lib libraries are much more systematically organised.
* There's a completely new "make" system. This will mainly affect people
who want the source distribution, who will hopefully find it much, much,
easier than grappling with the old Jmakefiles. Even for binary
installation, the procedure is a little simpler, though.
Please see the release notes for a complete discussion of What's New.
To run this release, you need a machine with 16+MB memory (more if
building from sources), GNU C (`gcc'), and `perl'. We have seen GHC
2.01 work on these platforms: alpha-dec-osf2, hppa1.1-hp-hpux9,
sparc-sun-{sunos4,solaris2}, mips-sgi-irix5, and
i386-unknown-{linux,solaris2,freebsd}. Similar platforms should work
with minimal hacking effort. The installer's guide give a full
what-ports-work report.
Binaries are distributed in `bundles', e.g. a "profiling bundle" or a
"concurrency bundle" for your platform. Just grab the ones you need.
Once you have the distribution, please follow the pointers in
ghc/README to find all of the documentation about this release. NB:
preserve modification times when un-tarring the files (no `m' option
for tar, please)!
We run mailing lists for GHC users and bug reports; to subscribe, send
mail to majordomo@dcs.gla.ac.uk; the msg body should be:
subscribe glasgow-haskell-<which> Your Name <your-email@where.you.are>
Please send bug reports about GHC to glasgow-haskell-bugs@dcs.gla.ac.uk.
Simon Peyton Jones
Dated: February 1997
Relevant URLs on the World-Wide Web:
GHC home page http://www.dcs.gla.ac.uk/fp/software/ghc/
Glasgow FP group page http://www.dcs.gla.ac.uk/fp/
comp.lang.functional FAQ http://www.cs.nott.ac.uk/Department/Staff/mpj/faq.html
======================================================================
How to get GHC 2.01:
This release is available by anonymous FTP from the main Haskell
archive sites, in the directory pub/haskell/glasgow:
ftp.dcs.gla.ac.uk (130.209.240.50)
ftp.cs.chalmers.se (129.16.227.140)
haskell.cs.yale.edu (128.36.11.43)
The Glasgow site is mirrored by src.doc.ic.ac.uk (146.169.43.1), in
computing/programming/languages/haskell/glasgow.
These are the available files (.gz files are gzipped) -- some are `on
demand', ask if you don't see them:
ghc-2.01-src.tar.gz The source distribution; about 3MB.
ghc-2.01.ANNOUNCE This file.
ghc-2.01.{README,RELEASE-NOTES} From the distribution; for those who
want to peek before FTPing...
ghc-2.01-ps-docs.tar.gz Main GHC documents in PostScript format; in
case your TeX setup doesn't agree with our
DVI files...
ghc-2.01-<platform>.tar.gz Basic binary distribution for a particular
<platform>. Unpack and go: you can compile
and run Haskell programs with nothing but one
of these files. NB: does *not* include
profiling (see below).
<platform> ==> alpha-dec-osf2
hppa1.1-hp-hpux9
i386-unknown-freebsd
i386-unknown-linux
i386-unknown-solaris2
m68k-sun-sunos4
mips-sgi-irix5
sparc-sun-sunos4
sparc-sun-solaris2
ghc-2.01-<bundle>-<platform>.tar.gz
<platform> ==> as above
<bundle> ==> prof (profiling)
conc (concurrent Haskell)
par (parallel)
gran (GranSim parallel simulator)
ticky (`ticky-ticky' counts -- for implementors)
prof-conc (profiling for "conc[urrent]")
prof-ticky (ticky for "conc[urrent]")
ghc-2.01-hc-files.tar.gz Basic set of intermediate C (.hc) files for the
compiler proper, the prelude, and `Hello,
world'. Used for bootstrapping the system.
About 4MB.
ghc-2.01-<bundle>-hc-files.tar.gz Further sets of .hc files, for
building other "bundles", e.g., profiling.
ghc-2.01-hi-files-<blah>.tar.gz Sometimes it's more convenient to
use a different set of interface files than
the ones in *-src.tar.gz. (The installation
guide will advise you of this.)
#-----------------------------------------------------------------------------
# $Id: Makefile,v 1.2 1996/11/21 16:45:54 simonm Exp $
#################################################################################
#
# fptools/Makefile
#
# This is the main Makefile for fptools.
#
#################################################################################
TOP = .
SUBDIRS = glafp-utils ghc
include $(TOP)/mk/gen.mk
include $(TOP)/mk/subdir.mk
TOP=.
include $(TOP)/mk/boilerplate.mk
SRC_DIST_DIR=$(shell pwd)/$(SRC_DIST_NAME)
line = @echo "------------------------------------------------------------------------------"
SUBDIRS = $(ProjectsToBuild)
boot ::
@echo "Bootstrapping $(PROJECTNAME)..."
$(line)
@echo "Booting glafp-utils"
$(line)
@$(MAKE) -C glafp-utils boot
#
# Files to include in fptools source distribution
#
SRC_DIST_DIRS += mk $(ProjectsToBuild)
SRC_DIST_FILES += configure.in config.guess config.sub configure README ANNOUNCE NEWS INSTALL Makefile
$(line)
@echo "Booting ghc"
$(line)
@$(MAKE) -C ghc boot
@echo "Done!"
#
# Making a binary distribution
#
BIN_DIST_TMPDIR=$(shell pwd)
BIN_DIST_NAME=fptools
#
# list of toplevel `projects' to include in binary distrib.
#
BIN_DIST_DIRS=ghc
binary-dist:: binary-dist-pre
BIN_DIST_TOP= distrib/Makefile-bin.in \
distrib/configure-bin.in \
distrib/README \
distrib/INSTALL \
distrib/ANNOUNCE
binary-dist::
@for i in $(BIN_DIST_TOP); do \
@echo cp $$i $(BIN_DIST_TMPDIR)/$(BIN_DIST_NAME) \
cp $$i $(BIN_DIST_TMPDIR)/$(BIN_DIST_NAME) \
done;
#
# Creating and copying the documentation into the bin-dist tree.
#
binary-dist::
$(MKDIRHIER) $(BIN_DIST_TMPDIR)/$(BIN_DIST_NAME)/html
$(MKDIRHIER) $(BIN_DIST_TMPDIR)/$(BIN_DIST_NAME)/dvi
$(MKDIRHIER) $(BIN_DIST_TMPDIR)/$(BIN_DIST_NAME)/info
@echo "Making html documentation.."
$(MAKE) -C docs --no-print-directory $(MFLAGS) html
cp docs/html/* $(BIN_DIST_TMPDIR)/$(BIN_DIST_NAME)/html
@echo "Making dvi files.."
$(MAKE) -C docs --no-print-directory $(MFLAGS) dvi
cp docs/*.dvi $(BIN_DIST_TMPDIR)/$(BIN_DIST_NAME)/dvi
@echo "Making info files.."
$(MAKE) -C docs --no-print-directory $(MFLAGS) info
cp docs/*.info* $(BIN_DIST_TMPDIR)/$(BIN_DIST_NAME)/info
dist :: dist-pre
include $(TOP)/mk/target.mk
dist :: dist-post
#
# Automatically remake update configuration files
# (from autoconf manual)
#
configure: configure.in
autoconf
#
# autoheader might not change config.h.in, so touch a stamp file.
#
mk/config.h.in: mk/stamp-h.in
mk/stamp-h.in: configure.in
autoheader
echo timestamp > mk/stamp-h.in
mk/config.h: mk/stamp-h
mk/stamp-h: mk/config.h.in config.status
./config.status
config.status: configure
./config.status --recheck
.PHONY: config
config: config.status
@:
......@@ -4,7 +4,7 @@ dnl
dnl * INITIAL SETUP, CHOICE OF PLATFORM(S)
#!/bin/sh
#
# (c) The AQUA Project, Glasgow University, 1994-1995
# (c) The AQUA Project, Glasgow University, 1994-1997
#
# Configure script for the Glasgow functional programming tools
# (created automagically by autoconf...do not edit by hand)
......@@ -12,163 +12,51 @@ dnl * INITIAL SETUP, CHOICE OF PLATFORM(S)
# Do "./configure --help" to see what flags are available.
# (Better yet, read the documentation!)
#
AC_INIT(mk/platform.mk.in)
# First off, a distrib sanity check..
AC_INIT(mk/config.mk.in)
# -------------------------------------------------------------------------
# Prepare to generate the following header files
#
AC_CONFIG_HEADER(ghc/includes/config.h literate/lit-deatify/config.h)
#
AC_CONFIG_HEADER(mk/config.h)
# No, we don't do `--srcdir'...
if test x"$srcdir" != 'x.' ; then
echo "This configuration does not support the \`--srcdir' option."
echo "This configuration does not support the \`--srcdir' option.."
exit 1
fi
#
# Remove some automounter nonsense (Glasgow specific gruff)
#
hardtop=`pwd`
hardtop=`echo $hardtop | sed 's|^/tmp_mnt/|/|' | sed 's|^/export/|/|' | sed 's|^/grasp_tmp|/local/grasp_tmp|'`
hardtop=`echo $hardtop | sed 's|^/tmp_mnt/|/|' | sed 's|^/grasp_tmp|/local/grasp_tmp|'`
#OLD: hardtop=`echo $hardtop | sed 's|^/tmp_mnt/|/|' | sed 's|^/export/|/|' | sed 's|^/grasp_tmp|/local/grasp_tmp|'`
echo ''
echo "*** The top of your build tree is: $hardtop"
AC_SUBST(hardtop)
# -------------------------------------------------------------------------
dnl ** choose what blobs to build (ghc,hslibs,haggis,happy,nofib,????)
# set to the name for the dir if doing it, otherwise empty
DoingHsLibs=''
DoingNoFib=''
DoingHappy=''
DoingHaggis=''
MkWorldSetup='std'
AC_ARG_ENABLE(hslibs,
[
**********************************************************************
* Configuration options for the Glasgow functional-programming tools *
**********************************************************************
First, select *which* of the tools you want to build, with
--{enable,disable}-{hslibs,nofib,happy,haggis}. (Currently, you have to
build ghc.)
Second, you may set one of a few applies-in-all-cases options. For
example, --with-tmpdir=/usr/tmp.
Then you may set various options which are specifically for the tools you
choose in step 1. For GHC, perhaps --enable-concurrent. For NoFib,
perhaps --enable-all-tests. And so on.
The rest of this message lists all of the configure options. If the option
is enabled by default, the message says how to disable it. And vice versa.
If you are confused, don't forget the installation documents that came with
the software!
*******************************************************************
** FOR SELECTING WHICH GLASGOW FP TOOLS TO BUILD:]
[--enable-hslibs build suite of Haskell libraries],
[case "$enableval" in
yes) DoingHsLibs='hslibs'
;;
no) DoingHsLibs=''
;;
*) echo "I don't understand this option: --enable-hslibs=$enableval"
exit 1
;;
esac])
if test "xxx$DoingHsLibs" = 'xxxhslibs' -a \( ! -d hslibs \) ; then
DoingHsLibs=''
echo 'Doing --disable-hslibs, as there is no hslibs directory'
fi
AC_ARG_ENABLE(nofib,
[--enable-nofib build NoFib suite as part of Glasgow FP tools],
[case "$enableval" in
yes) DoingNoFib='nofib'
;;
no) DoingNoFib=''
;;
*) echo "I don't understand this option: --enable-nofib=$enableval"
exit 1
;;
esac])
if test "xxx$DoingNoFib" = 'xxxnofib' -a \( ! -d nofib \) ; then
DoingNoFib=''
echo 'Doing --disable-nofib, as there is no nofib directory'
fi
AC_ARG_ENABLE(happy,
[--enable-happy build Happy parser-generator as part of Glasgow FP tools],
[case "$enableval" in
yes) DoingHappy='happy'
;;
no) DoingHappy=''
;;
*) echo "I don't understand this option: --enable-happy=$enableval"
exit 1
;;
esac])
if test "xxx$DoingHappy" = 'xxxhappy' -a \( ! -d happy \) ; then
DoingHappy=''
echo 'Doing --disable-happy, as there is no happy directory'
fi
AC_ARG_ENABLE(haggis,
[--enable-haggis build Haggis GUI toolkit as part of Glasgow FP tools],
[case "$enableval" in
yes) DoingHaggis='haggis'
;;
no) DoingHaggis=''
;;
*) echo "I don't understand this option: --enable-haggis=$enableval"
exit 1
;;
esac])
if test "xxx$DoingHaggis" = 'xxxhaggis' -a \( ! -d haggis \) ; then
DoingHaggis=''
echo 'Doing --disable-haggis, as there is no haggis directory'
fi
AC_SUBST(DoingHsLibs)
AC_SUBST(DoingNoFib)
AC_SUBST(DoingHappy)
AC_SUBST(DoingHaggis)
# -------------------------------------------------------------------------
dnl ** choose host(/target/build) platform
#
# Guess host/target/build platform(s) if necessary.
# Partly stolen from GCC "configure".
#
if test "x$target" = xNONE ; then
if test "x$nonopt" != xNONE ; then
target=$nonopt
else
# This way of testing the result of a command substitution is
# defined by Posix.2 (section 3.9.1) as well as traditional shells.
if target=`$srcdir/config.guess` ; then
echo "Configuring for a ${target} host." 1>&2
else
echo 'Config.guess failed to determine the host type. You need \
to specify one.' 1>&2
if test -r config.status ; then
tail +2 config.status 1>&2
fi
exit 1
fi
fi
fi
AC_CANONICAL_SYSTEM
# "$host" defaults to "$target"
if test "x$host" = xNONE ; then
host=$target
fi
# "$build" defaults to "$host"
if test "x$build" = xNONE ; then
build=$host
else
echo "This configuration does not support the \`--build' option."
exit 1
fi
#if test "x$build" = xNONE ; then
# build=$host
#else
# echo "This configuration does not support the \`--build' option."
# exit 1
#fi
dnl ** canonicalize platform names
# Canonicali[sz]e those babies
......@@ -181,22 +69,23 @@ if test x"$TargetPlatform" != x"$HostPlatform" ; then
exit 1
fi
#
# The following will be more difficult when we *are* cross-compiling.
# Suitable names to slam in *_CPP are in platform.h.in.
# We also record the architecture, vendor, and operating system (OS)
# separately.
case $HostPlatform in
alpha-dec-osf[[1234]]*)
HostPlatform=alpha-dec-osf1 # canonicalise for our purposes
HostPlatform=alpha-dec-osf1 # canonicalise for our purposes
TargetPlatform=alpha-dec-osf1 # this will work for now... (hack)
BuildPlatform=alpha-dec-osf1 #hack
BuildPlatform=alpha-dec-osf1 # hack
HostPlatform_CPP='alpha_dec_osf1'
HostArch_CPP='alpha'
HostVendor_CPP='dec'
HostOS_CPP='osf1'
;;
hppa1.1-hp-hpux*)
HostPlatform=hppa1.1-hp-hpux # canonicalise for our purposes (hack)
HostPlatform=hppa1.1-hp-hpux # canonicalise for our purposes (hack)
TargetPlatform=hppa1.1-hp-hpux
BuildPlatform=hppa1.1-hp-hpux
HostPlatform_CPP='hppa1_1_hp_hpux'
......@@ -205,7 +94,7 @@ hppa1.1-hp-hpux*)
HostOS_CPP='hpux'
;;
i[[3456]]86-*-linuxaout*)
HostPlatform=i386-unknown-linuxaout # hack again
HostPlatform=i386-unknown-linuxaout # hack again
TargetPlatform=i386-unknown-linuxaout
BuildPlatform=i386-unknown-linuxaout
HostPlatform_CPP='i386_unknown_linuxaout'
......@@ -249,6 +138,15 @@ i[[3456]]86-*-solaris2*)
HostVendor_CPP='unknown'
HostOS_CPP='solaris2'
;;
i[[3456]]86-*-cygwin32*)
HostPlatform=i386-unknown-cygwin32 # hack again
TargetPlatform=i386-unknown-cygwin32
BuildPlatform=i386-unknown-cygwin32
HostPlatform_CPP='i386_unknown_cygwin32'
HostArch_CPP='i386'
HostVendor_CPP='unknown'
HostOS_CPP='cygwin32'
;;
m68k-next-nextstep2)
HostPlatform_CPP='m68k_next_nextstep2'
HostArch_CPP='m68k'
......@@ -326,12 +224,9 @@ sparc-sun-solaris2*)
exit 1
;;
esac
test -n "$verbose" && echo "Host platform set to $HostPlatform"
test -n "$verbose" -a x"$HostPlatform" != x"$TargetPlatform" \
&& echo "Target platform set to $TargetPlatform"
test -n "$verbose" -a x"$BuildPlatform" != x"$HostPlatform" \
&& echo "Build platform set to $BuildPlatform"
echo "Which we'll canonicalise into: $HostPlatform"
test x"$HostPlatform" != x"$TargetPlatform" && echo "Target platform set to $TargetPlatform"
test x"$BuildPlatform" != x"$HostPlatform" && echo "Build platform set to $BuildPlatform"