ghc-hetmet.git
23 years ago[project @ 2001-03-29 10:45:58 by simonmar]
simonmar [Thu, 29 Mar 2001 10:45:58 +0000 (10:45 +0000)]
[project @ 2001-03-29 10:45:58 by simonmar]
allow VERSION_SYMLINKS_FOR to be configured

23 years ago[project @ 2001-03-29 10:44:29 by simonmar]
simonmar [Thu, 29 Mar 2001 10:44:29 +0000 (10:44 +0000)]
[project @ 2001-03-29 10:44:29 by simonmar]
update for GHC 5.00

23 years ago[project @ 2001-03-29 08:03:47 by qrczak]
qrczak [Thu, 29 Mar 2001 08:03:47 +0000 (08:03 +0000)]
[project @ 2001-03-29 08:03:47 by qrczak]
integralEnumFrom*To don't need Bounded.

23 years ago[project @ 2001-03-29 08:03:21 by qrczak]
qrczak [Thu, 29 Mar 2001 08:03:21 +0000 (08:03 +0000)]
[project @ 2001-03-29 08:03:21 by qrczak]
Allow specifying output filenames explicitly.
Add -D --define option.

23 years ago[project @ 2001-03-29 00:01:18 by qrczak]
qrczak [Thu, 29 Mar 2001 00:01:18 +0000 (00:01 +0000)]
[project @ 2001-03-29 00:01:18 by qrczak]
Transform *.hsc into *.hs and optionally Hs*.h  and Hs*.c
(used to be          *.hs and optionally *.hs.h and *.hs.c).

Old names interacted badly with Makefile rules of the form
'%: %.o' and looked ugly.

23 years ago[project @ 2001-03-28 22:14:02 by qrczak]
qrczak [Wed, 28 Mar 2001 22:14:02 +0000 (22:14 +0000)]
[project @ 2001-03-28 22:14:02 by qrczak]
Fix Enum instances for types larger than Int.

23 years ago[project @ 2001-03-28 18:53:39 by qrczak]
qrczak [Wed, 28 Mar 2001 18:53:39 +0000 (18:53 +0000)]
[project @ 2001-03-28 18:53:39 by qrczak]
Typo.

23 years ago[project @ 2001-03-28 18:48:46 by qrczak]
qrczak [Wed, 28 Mar 2001 18:48:46 +0000 (18:48 +0000)]
[project @ 2001-03-28 18:48:46 by qrczak]
HS_CHAR_MAX is 0x10FFFF.

23 years ago[project @ 2001-03-28 18:48:22 by qrczak]
qrczak [Wed, 28 Mar 2001 18:48:22 +0000 (18:48 +0000)]
[project @ 2001-03-28 18:48:22 by qrczak]
s/ISO-8879-1/ISO-8859-1/

23 years ago[project @ 2001-03-28 16:51:02 by simonmar]
simonmar [Wed, 28 Mar 2001 16:51:03 +0000 (16:51 +0000)]
[project @ 2001-03-28 16:51:02 by simonmar]
Cleaning up error reporting, 2nd attempt.

  - The UserError name is already bagged by Exception (for
    userErrors).  So we use ProgramError instead, which is
    more appropriate.

  - some previously UserErrors are now CmdLineErrors.  GHCi
    catches CmdLineErrors and prints them without the "ghc: "
    prefix.

23 years ago[project @ 2001-03-28 16:45:35 by simonmar]
simonmar [Wed, 28 Mar 2001 16:45:35 +0000 (16:45 +0000)]
[project @ 2001-03-28 16:45:35 by simonmar]
Clean up GHC's error reporting

  - the GhcException type has some more constructors: CmdLineError,
    ProgramError, and InstallationError.  OtherError has gone.

  - most error messages should begin with "<location>:".  When the
    error is on the command-line or in GHC itself, <location> is
    "ghc", for consistency with std Unix semantics.

23 years ago[project @ 2001-03-28 16:40:48 by simonmar]
simonmar [Wed, 28 Mar 2001 16:40:48 +0000 (16:40 +0000)]
[project @ 2001-03-28 16:40:48 by simonmar]
add performGC

23 years ago[project @ 2001-03-28 16:38:39 by simonmar]
simonmar [Wed, 28 Mar 2001 16:38:39 +0000 (16:38 +0000)]
[project @ 2001-03-28 16:38:39 by simonmar]
use newStablePtr instead of deprecated makeStablePtr.

23 years ago[project @ 2001-03-28 16:36:04 by simonmar]
simonmar [Wed, 28 Mar 2001 16:36:04 +0000 (16:36 +0000)]
[project @ 2001-03-28 16:36:04 by simonmar]
add a simple test for IOExts.performGC

23 years ago[project @ 2001-03-28 16:35:39 by simonmar]
simonmar [Wed, 28 Mar 2001 16:35:39 +0000 (16:35 +0000)]
[project @ 2001-03-28 16:35:39 by simonmar]
Add a simple test for performGC.

23 years ago[project @ 2001-03-28 16:33:55 by simonmar]
simonmar [Wed, 28 Mar 2001 16:33:55 +0000 (16:33 +0000)]
[project @ 2001-03-28 16:33:55 by simonmar]
remove this outdated, nonworking test

23 years ago[project @ 2001-03-28 16:20:00 by simonmar]
simonmar [Wed, 28 Mar 2001 16:20:00 +0000 (16:20 +0000)]
[project @ 2001-03-28 16:20:00 by simonmar]
Use fromIntegral instead of fromInt/toInt.

Marcin: this test shows up some problems with enumFrom & friends on
the sized Integral types.  Could you take a look?  Cheers.

23 years ago[project @ 2001-03-28 15:44:57 by simonmar]
simonmar [Wed, 28 Mar 2001 15:44:57 +0000 (15:44 +0000)]
[project @ 2001-03-28 15:44:57 by simonmar]
- Note the difference in the size of Char between GHC & Haskell 98.

- Don't lie about support for arbitrary-sized tuples.

- Just for fun, document some of the more obscure differences between
  GHC syntax and Haskell 98.

23 years ago[project @ 2001-03-28 15:16:32 by simonmar]
simonmar [Wed, 28 Mar 2001 15:16:32 +0000 (15:16 +0000)]
[project @ 2001-03-28 15:16:32 by simonmar]
maxBound :: Char is 0x10ffff now.

23 years ago[project @ 2001-03-28 14:30:10 by simonmar]
simonmar [Wed, 28 Mar 2001 14:31:53 +0000 (14:31 +0000)]
[project @ 2001-03-28 14:30:10 by simonmar]
update expected output

23 years ago[project @ 2001-03-28 14:26:34 by simonmar]
simonmar [Wed, 28 Mar 2001 14:26:34 +0000 (14:26 +0000)]
[project @ 2001-03-28 14:26:34 by simonmar]
fix some test problems

23 years ago[project @ 2001-03-28 14:09:41 by simonmar]
simonmar [Wed, 28 Mar 2001 14:09:41 +0000 (14:09 +0000)]
[project @ 2001-03-28 14:09:41 by simonmar]
fix a cache problem with the Happy configure macro.

23 years ago[project @ 2001-03-28 11:01:19 by simonmar]
simonmar [Wed, 28 Mar 2001 11:01:19 +0000 (11:01 +0000)]
[project @ 2001-03-28 11:01:19 by simonmar]
Clean up GHC's error reporting.

  - the GhcException type has some more constructors: CmdLineError,
    UserError, and InstallationError.  OtherError has gone.

  - most error messages should begin with "<location>:".  When the
    error is on the command-line or in GHC itself, <location> is
    "ghc", for consistency with std Unix semantics.

  - GHCi no longer prints a superfluous "ghc: " before certain error
    messages.

23 years ago[project @ 2001-03-28 10:34:15 by simonmar]
simonmar [Wed, 28 Mar 2001 10:34:15 +0000 (10:34 +0000)]
[project @ 2001-03-28 10:34:15 by simonmar]
Document the use of :set with command-line options a little better.

23 years ago[project @ 2001-03-28 05:07:34 by kglynn]
kglynn [Wed, 28 Mar 2001 05:07:34 +0000 (05:07 +0000)]
[project @ 2001-03-28 05:07:34 by kglynn]
Cpr needs the correct arity for an imported function.  In some cases
the arity reported in the interface can be too small.

In this test case the function is returning a newtype,  and the newtype
is hiding a function.  The arity in the interface file says 0.  It is
triggered by compiling the cpr001.hs test files with profiling on. The code
comes from a bug report.

I should look for a better case,  because the profiling flaw which reveals
this problem will hopefully go away some time.

23 years ago[project @ 2001-03-27 16:55:03 by simonmar]
simonmar [Tue, 27 Mar 2001 16:55:03 +0000 (16:55 +0000)]
[project @ 2001-03-27 16:55:03 by simonmar]
CmdLineOpts:
- add dopt_set and dopt_unset utils.

InteractiveUI:
- unset -fwarn-unused-binds before evaluating a statement,
  to avoid bogus warnings about new bindings.

23 years ago[project @ 2001-03-27 16:33:17 by rrt]
rrt [Tue, 27 Mar 2001 16:33:17 +0000 (16:33 +0000)]
[project @ 2001-03-27 16:33:17 by rrt]
Make it work on Windows with recent GCCs.

23 years ago[project @ 2001-03-27 16:32:46 by rrt]
rrt [Tue, 27 Mar 2001 16:32:46 +0000 (16:32 +0000)]
[project @ 2001-03-27 16:32:46 by rrt]
Tweaks to make it work a bit better.

23 years ago[project @ 2001-03-27 16:10:13 by simonmar]
simonmar [Tue, 27 Mar 2001 16:10:13 +0000 (16:10 +0000)]
[project @ 2001-03-27 16:10:13 by simonmar]
shell-wrapper for ghc-pkg.

23 years ago[project @ 2001-03-27 14:10:34 by simonmar]
simonmar [Tue, 27 Mar 2001 14:10:34 +0000 (14:10 +0000)]
[project @ 2001-03-27 14:10:34 by simonmar]
- add --show-package option to display the config info for a particular
  package.

- add --field info, for use with --show-package, which displays the
  contents of the given field in the package config only.  The field
  is displayed one string per line.

23 years ago[project @ 2001-03-27 14:05:09 by simonpj]
simonpj [Tue, 27 Mar 2001 14:05:09 +0000 (14:05 +0000)]
[project @ 2001-03-27 14:05:09 by simonpj]
Print minimal import operators correctly

23 years ago[project @ 2001-03-27 13:38:03 by simonmar]
simonmar [Tue, 27 Mar 2001 13:38:03 +0000 (13:38 +0000)]
[project @ 2001-03-27 13:38:03 by simonmar]
- add ghc-pkg-$(ProjectVersion) and ghc-pkg-inplace scripts
- add ghc-pkg script link
- rename ghc-pkg to ghc-pkg.bin and put it in $(libdir)
- remove the hacky passing of $(libdir) to the binary

23 years ago[project @ 2001-03-27 12:59:18 by simonmar]
simonmar [Tue, 27 Mar 2001 12:59:18 +0000 (12:59 +0000)]
[project @ 2001-03-27 12:59:18 by simonmar]
wibble

23 years ago[project @ 2001-03-27 11:58:54 by simonmar]
simonmar [Tue, 27 Mar 2001 11:58:54 +0000 (11:58 +0000)]
[project @ 2001-03-27 11:58:54 by simonmar]
bootstrap wibbles

23 years ago[project @ 2001-03-27 11:38:19 by simonmar]
simonmar [Tue, 27 Mar 2001 11:38:19 +0000 (11:38 +0000)]
[project @ 2001-03-27 11:38:19 by simonmar]
updates to work with GHC 5, I hope...

23 years ago[project @ 2001-03-27 11:37:39 by simonmar]
simonmar [Tue, 27 Mar 2001 11:37:39 +0000 (11:37 +0000)]
[project @ 2001-03-27 11:37:39 by simonmar]
non-absolute paths didn't work.

23 years ago[project @ 2001-03-27 10:33:24 by simonmar]
simonmar [Tue, 27 Mar 2001 10:33:24 +0000 (10:33 +0000)]
[project @ 2001-03-27 10:33:24 by simonmar]
complain if the -B<dir> flag is missing.

23 years ago[project @ 2001-03-27 10:33:03 by simonmar]
simonmar [Tue, 27 Mar 2001 10:33:03 +0000 (10:33 +0000)]
[project @ 2001-03-27 10:33:03 by simonmar]
we don't have a default TopDir any more.

23 years ago[project @ 2001-03-27 10:32:42 by simonmar]
simonmar [Tue, 27 Mar 2001 10:32:42 +0000 (10:32 +0000)]
[project @ 2001-03-27 10:32:42 by simonmar]
Config.hs no longer contains any absolute directories.  The only
absolute directory it knows about is always passed in from the
mini-driver in the -B<dir> flag.

This setup is pretty fragile for booting from .hc anyway because
Config.hs contains platform-specific info, but it'll work enough to
get the compiler booted.

23 years ago[project @ 2001-03-27 09:44:56 by simonmar]
simonmar [Tue, 27 Mar 2001 09:44:56 +0000 (09:44 +0000)]
[project @ 2001-03-27 09:44:56 by simonmar]
use fewer absolute paths

23 years ago[project @ 2001-03-27 09:38:26 by simonmar]
simonmar [Tue, 27 Mar 2001 09:38:26 +0000 (09:38 +0000)]
[project @ 2001-03-27 09:38:26 by simonmar]
fixes for linking the compiler

23 years ago[project @ 2001-03-27 09:38:02 by simonmar]
simonmar [Tue, 27 Mar 2001 09:38:02 +0000 (09:38 +0000)]
[project @ 2001-03-27 09:38:02 by simonmar]
bootstrapping support

23 years ago[project @ 2001-03-27 09:36:53 by simonmar]
simonmar [Tue, 27 Mar 2001 09:36:53 +0000 (09:36 +0000)]
[project @ 2001-03-27 09:36:53 by simonmar]
fix GHCBIN for ghci-inplace

23 years ago[project @ 2001-03-27 08:29:07 by qrczak]
qrczak [Tue, 27 Mar 2001 08:29:07 +0000 (08:29 +0000)]
[project @ 2001-03-27 08:29:07 by qrczak]
Install package.conf in the right place.

23 years ago[project @ 2001-03-27 00:43:04 by qrczak]
qrczak [Tue, 27 Mar 2001 00:43:04 +0000 (00:43 +0000)]
[project @ 2001-03-27 00:43:04 by qrczak]
Fix making ghc-inplace.

23 years ago[project @ 2001-03-26 22:42:41 by qrczak]
qrczak [Mon, 26 Mar 2001 22:42:42 +0000 (22:42 +0000)]
[project @ 2001-03-26 22:42:41 by qrczak]
Fix installation of the compiler binary and its drivers.

23 years ago[project @ 2001-03-26 16:59:20 by simonmar]
simonmar [Mon, 26 Mar 2001 16:59:20 +0000 (16:59 +0000)]
[project @ 2001-03-26 16:59:20 by simonmar]
small hacks for bootstrapping

23 years ago[project @ 2001-03-26 16:58:09 by simonmar]
simonmar [Mon, 26 Mar 2001 16:58:09 +0000 (16:58 +0000)]
[project @ 2001-03-26 16:58:09 by simonmar]
nearly there: linking works now

23 years ago[project @ 2001-03-26 16:56:55 by simonmar]
simonmar [Mon, 26 Mar 2001 16:56:55 +0000 (16:56 +0000)]
[project @ 2001-03-26 16:56:55 by simonmar]
we have BinDistShScripts now

23 years ago[project @ 2001-03-26 16:56:27 by simonmar]
simonmar [Mon, 26 Mar 2001 16:56:27 +0000 (16:56 +0000)]
[project @ 2001-03-26 16:56:27 by simonmar]
remove unnecessary junk

23 years ago[project @ 2001-03-26 16:55:37 by simonmar]
simonmar [Mon, 26 Mar 2001 16:55:37 +0000 (16:55 +0000)]
[project @ 2001-03-26 16:55:37 by simonmar]
Move the -monly-n-regs flags into the Makefile, necessary for .hc
bootstrapping.

23 years ago[project @ 2001-03-26 16:54:34 by simonmar]
simonmar [Mon, 26 Mar 2001 16:54:34 +0000 (16:54 +0000)]
[project @ 2001-03-26 16:54:34 by simonmar]
Moved into ghci subdirectory.

23 years ago[project @ 2001-03-26 16:54:10 by simonmar]
simonmar [Mon, 26 Mar 2001 16:54:10 +0000 (16:54 +0000)]
[project @ 2001-03-26 16:54:10 by simonmar]
Remove unused file.

23 years ago[project @ 2001-03-26 16:53:35 by simonmar]
simonmar [Mon, 26 Mar 2001 16:53:35 +0000 (16:53 +0000)]
[project @ 2001-03-26 16:53:35 by simonmar]
- make the pkgconf program independent of the location of the build
  tree - instead of Config.hs, we pass the configuration parameters in
  as options.  This is necessary for .hc bootstrapping.

- remove the ghci driver stuff, it moved to the ghci subdirectory.

- remove ghc5.hs

23 years ago[project @ 2001-03-26 16:52:03 by simonmar]
simonmar [Mon, 26 Mar 2001 16:52:03 +0000 (16:52 +0000)]
[project @ 2001-03-26 16:52:03 by simonmar]
if the STOLEN_X86_REGS arg is missing, assume 4.

23 years ago[project @ 2001-03-26 16:51:26 by simonmar]
simonmar [Mon, 26 Mar 2001 16:51:27 +0000 (16:51 +0000)]
[project @ 2001-03-26 16:51:26 by simonmar]
mini-drivers for GHC 5  (ghci moved from ghc/driver to ghc/driver/ghci, ghc
is new).

23 years ago[project @ 2001-03-26 13:43:05 by simonmar]
simonmar [Mon, 26 Mar 2001 13:43:05 +0000 (13:43 +0000)]
[project @ 2001-03-26 13:43:05 by simonmar]
oops, STABLE_NAME_info is stg_STABLE_NAME_info in the HEAD.

23 years ago[project @ 2001-03-26 13:31:46 by simonmar]
simonmar [Mon, 26 Mar 2001 13:31:46 +0000 (13:31 +0000)]
[project @ 2001-03-26 13:31:46 by simonmar]
StableNames weren't being initialized properly (the CCS field for
profiling was left out).

23 years ago[project @ 2001-03-26 12:28:15 by simonmar]
simonmar [Mon, 26 Mar 2001 12:28:15 +0000 (12:28 +0000)]
[project @ 2001-03-26 12:28:15 by simonmar]
Simplify the foreign-export stub processing.

  - DynFlags now has fields for the stub.h and stub.c filenames, for
    consistency with the normal hsc output file name.

  - codeOutput puts the stubs into these files rather than dreaming
    up new temporary names for them

  - now we don't have to move the stubs into the right place in
    DriverPipeline.

  - we do however have to inject the correct #includes into the stub.c
    file when it is generated: I'm now injecting the same includes as
    the .hc file gets plus "RtsAPI.h", which is probably more correct
    than the hacky hardcoded "Stg.h" we had before.

23 years ago[project @ 2001-03-25 19:30:23 by qrczak]
qrczak [Sun, 25 Mar 2001 19:30:23 +0000 (19:30 +0000)]
[project @ 2001-03-25 19:30:23 by qrczak]
Preprocessing hack made less ugly by moving 'tail' from usage to definition.

23 years ago[project @ 2001-03-25 13:52:46 by qrczak]
qrczak [Sun, 25 Mar 2001 13:52:46 +0000 (13:52 +0000)]
[project @ 2001-03-25 13:52:46 by qrczak]
Remove bogus comment.

23 years ago[project @ 2001-03-25 13:29:54 by simonmar]
simonmar [Sun, 25 Mar 2001 13:29:54 +0000 (13:29 +0000)]
[project @ 2001-03-25 13:29:54 by simonmar]
- include :undef in the :help message
- don't read ./.ghci if it is the same file as $HOME/.ghci

23 years ago[project @ 2001-03-25 11:08:42 by panne]
panne [Sun, 25 Mar 2001 11:08:42 +0000 (11:08 +0000)]
[project @ 2001-03-25 11:08:42 by panne]
OK, here comes the third try to get a simple string from the Makefile
into the Haskell sources in a "portable" way... Perhaps we should use
compiler/main/Config.hs's technique here.

23 years ago[project @ 2001-03-25 09:57:24 by qrczak]
qrczak [Sun, 25 Mar 2001 09:57:26 +0000 (09:57 +0000)]
[project @ 2001-03-25 09:57:24 by qrczak]
Promote (//) from a function to IArray method with an inefficient
default definition. This completely compatible change allows efficient
implementations of (//) for particular types.

Explicitly define efficient (//) for Array and UArray.

Use unsafeFreeze* instead of freeze* in (//), accum and accumArray
for Array and IArray.

Remove showList in instance Show Array (the default definition does
the job).

Add Eq, Ord and Show instances for UArray. (Would be simpler if types
in instance contexts were not required to be type variables. I didn't
use -fallow-undecidable-instances but defined instances for individual
element types separately.)

Add unsafe{Freeze,Thaw}/{STArray,IOArray,IOUArray} rules.

Fix thaw/IOUArray rule (nobody uses IOUArrays in the ST monad!).

23 years ago[project @ 2001-03-24 19:50:56 by qrczak]
qrczak [Sat, 24 Mar 2001 19:50:56 +0000 (19:50 +0000)]
[project @ 2001-03-24 19:50:56 by qrczak]
Typo in 2147483648.

23 years ago[project @ 2001-03-24 18:34:05 by qrczak]
qrczak [Sat, 24 Mar 2001 18:34:05 +0000 (18:34 +0000)]
[project @ 2001-03-24 18:34:05 by qrczak]
Fix passing library dir to ghc-pkg.

23 years ago[project @ 2001-03-24 14:58:04 by simonmar]
simonmar [Sat, 24 Mar 2001 14:58:04 +0000 (14:58 +0000)]
[project @ 2001-03-24 14:58:04 by simonmar]
Turn CSE off when compiling main/DriverState for now.  It interacts
badly with our global variable hacks, commoning up several of them into
single variables :-(

We have a proposed fix for this, which is to make the NOINLINE pragma
also mean "NO SHARING", but it isn't committed yet.

23 years ago[project @ 2001-03-23 17:41:32 by simonmar]
simonmar [Fri, 23 Mar 2001 17:41:32 +0000 (17:41 +0000)]
[project @ 2001-03-23 17:41:32 by simonmar]
Add overlapping string pattern test.

23 years ago[project @ 2001-03-23 17:34:44 by simonmar]
simonmar [Fri, 23 Mar 2001 17:34:44 +0000 (17:34 +0000)]
[project @ 2001-03-23 17:34:44 by simonmar]
Fix the long-standing overlapping patterns bug with string-literal
patterns.

23 years ago[project @ 2001-03-23 17:04:56 by simonmar]
simonmar [Fri, 23 Mar 2001 17:04:56 +0000 (17:04 +0000)]
[project @ 2001-03-23 17:04:56 by simonmar]
can't inject the stub.h include here, it's too late.

23 years ago[project @ 2001-03-23 17:04:26 by simonmar]
simonmar [Fri, 23 Mar 2001 17:04:26 +0000 (17:04 +0000)]
[project @ 2001-03-23 17:04:26 by simonmar]
don't forget to inject a #include for the stub.h file.

23 years ago[project @ 2001-03-23 16:36:20 by simonmar]
simonmar [Fri, 23 Mar 2001 16:36:23 +0000 (16:36 +0000)]
[project @ 2001-03-23 16:36:20 by simonmar]
Changes to support bootstrapping the compiler from .hc files.  It's
not quite working yet, but it's not far off.

  - the biggest change is that any injected #includes are now placed in
    the .hc file at generation time, rather than compilation time.  I
    can't see any reason not to do this - it makes it clear by looking at
    the .hc file which files are being #included, it means one less
    temporary file at compilation time, and it means the .hc file is more
    standalone.

  - all the gruesomeness is in mk/bootstrap.mk, which handles building
    .hc files without a ghc driver.

23 years ago[project @ 2001-03-23 16:25:02 by simonmar]
simonmar [Fri, 23 Mar 2001 16:25:02 +0000 (16:25 +0000)]
[project @ 2001-03-23 16:25:02 by simonmar]
add ghc-pkg

23 years ago[project @ 2001-03-23 15:25:30 by simonmar]
simonmar [Fri, 23 Mar 2001 15:25:30 +0000 (15:25 +0000)]
[project @ 2001-03-23 15:25:30 by simonmar]
document :def and :undef

23 years ago[project @ 2001-03-23 14:52:30 by simonpj]
simonpj [Fri, 23 Mar 2001 14:52:30 +0000 (14:52 +0000)]
[project @ 2001-03-23 14:52:30 by simonpj]
Add comment

23 years ago[project @ 2001-03-23 14:52:17 by simonpj]
simonpj [Fri, 23 Mar 2001 14:52:17 +0000 (14:52 +0000)]
[project @ 2001-03-23 14:52:17 by simonpj]
Improve printing slightly

23 years ago[project @ 2001-03-23 14:51:54 by simonpj]
simonpj [Fri, 23 Mar 2001 14:51:54 +0000 (14:51 +0000)]
[project @ 2001-03-23 14:51:54 by simonpj]
Fix renamer error that prevented PrelBase compiling

23 years ago[project @ 2001-03-23 13:00:39 by rrt]
rrt [Fri, 23 Mar 2001 13:00:39 +0000 (13:00 +0000)]
[project @ 2001-03-23 13:00:39 by rrt]
Include config.h, not Stg.h

23 years ago[project @ 2001-03-23 12:29:07 by simonmar]
simonmar [Fri, 23 Mar 2001 12:29:07 +0000 (12:29 +0000)]
[project @ 2001-03-23 12:29:07 by simonmar]
Add test for :def, :undef

23 years ago[project @ 2001-03-23 12:12:18 by simonmar]
simonmar [Fri, 23 Mar 2001 12:12:18 +0000 (12:12 +0000)]
[project @ 2001-03-23 12:12:18 by simonmar]
make :def work again

23 years ago[project @ 2001-03-23 12:11:26 by simonmar]
simonmar [Fri, 23 Mar 2001 12:11:26 +0000 (12:11 +0000)]
[project @ 2001-03-23 12:11:26 by simonmar]
unused import

23 years ago[project @ 2001-03-23 10:47:21 by simonpj]
simonpj [Fri, 23 Mar 2001 10:47:21 +0000 (10:47 +0000)]
[project @ 2001-03-23 10:47:21 by simonpj]
Comments, plus dead code elim

23 years ago[project @ 2001-03-23 10:46:27 by simonpj]
simonpj [Fri, 23 Mar 2001 10:46:27 +0000 (10:46 +0000)]
[project @ 2001-03-23 10:46:27 by simonpj]
-----------------------------------
Correct size calculations for INLINE notes
-----------------------------------

Amazingly, CoreUnfold.size_up wasn't treating an InlineMe note
as small.  (InlineMe was only dealt with at top level.)  As
the comment now says:

  -- This can be important.  If you have an instance decl like this:
  --  instance Foo a => Foo [a] where
  --    {-# INLINE op1, op2 #-}
  --    op1 = ...
  --    op2 = ...
  -- then we'll get a dfun which is a pair of two INLINE lambdas

This fixes a problem Manuel encountered.

23 years ago[project @ 2001-03-23 10:44:08 by simonpj]
simonpj [Fri, 23 Mar 2001 10:44:08 +0000 (10:44 +0000)]
[project @ 2001-03-23 10:44:08 by simonpj]
-----------------------------------
Remove a redundant test in WorkWrap
-----------------------------------

We were making a worker/wrapper for an INLINE
function when it wasn't necessary, and that's a Bad Idea.
As the comment with WorkWrap.tryWW now says:

-- It's very important to refrain from w/w-ing an INLINE function
-- If we do so by mistake we transform
-- f = __inline (\x -> E)
-- into
-- f = __inline (\x -> case x of (a,b) -> fw E)
-- fw = \ab -> (__inline (\x -> E)) (a,b)
-- and the original __inline now vanishes, so E is no longer
-- inside its __inline wrapper.  Death!  Disaster!

There was one case when we did w/w it (to do with coercions),
but it turned out to be a vestige, as the OUT OF DATE NOTE says.

23 years ago[project @ 2001-03-23 09:33:22 by qrczak]
qrczak [Fri, 23 Mar 2001 09:33:22 +0000 (09:33 +0000)]
[project @ 2001-03-23 09:33:22 by qrczak]
hsc2hs lexical rules clarified.

23 years ago[project @ 2001-03-22 12:12:23 by simonmar]
simonmar [Thu, 22 Mar 2001 12:12:23 +0000 (12:12 +0000)]
[project @ 2001-03-22 12:12:23 by simonmar]
Move the bug reporting section earlier, and mention the SourceForge
bug tracker.

Update the release notes.

I think I'm just about done in here...

23 years ago[project @ 2001-03-22 11:20:36 by simonpj]
simonpj [Thu, 22 Mar 2001 11:20:36 +0000 (11:20 +0000)]
[project @ 2001-03-22 11:20:36 by simonpj]
Improve error message for ambiguity

23 years ago[project @ 2001-03-22 11:19:28 by simonpj]
simonpj [Thu, 22 Mar 2001 11:19:28 +0000 (11:19 +0000)]
[project @ 2001-03-22 11:19:28 by simonpj]
Improve error reporting for signatures not in scope

23 years ago[project @ 2001-03-22 11:00:54 by simonpj]
simonpj [Thu, 22 Mar 2001 11:00:54 +0000 (11:00 +0000)]
[project @ 2001-03-22 11:00:54 by simonpj]
Add a bit more info about -fno-implicit-prelude

23 years ago[project @ 2001-03-22 03:51:08 by hwloidl]
hwloidl [Thu, 22 Mar 2001 03:51:13 +0000 (03:51 +0000)]
[project @ 2001-03-22 03:51:08 by hwloidl]
-*- outline -*-
Time-stamp: <Thu Mar 22 2001 03:50:16 Stardate: [-30]6365.79 hwloidl>

This commit covers changes in GHC to get GUM (way=mp) and GUM/GdH (way=md)
working. It is a merge of my working version of GUM, based on GHC 4.06,
with GHC 4.11. Almost all changes are in the RTS (see below).

GUM is reasonably stable, we used the 4.06 version in large-ish programs for
recent papers. Couple of things I want to change, but nothing urgent.
GUM/GdH has just been merged and needs more testing. Hope to do that in the
next weeks. It works in our working build but needs tweaking to run.
GranSim doesn't work yet (*sigh*). Most of the code should be in, but needs
more debugging.

ToDo: I still want to make the following minor modifications before the release
- Better wrapper skript for parallel execution [ghc/compiler/main]
- Update parallel docu: started on it but it's minimal [ghc/docs/users_guide]
- Clean up [nofib/parallel]: it's a real mess right now (*sigh*)
- Update visualisation tools (minor things only IIRC) [ghc/utils/parallel]
- Add a Klingon-English glossary

* RTS:

Almost all changes are restricted to ghc/rts/parallel and should not
interfere with the rest. I only comment on changes outside the parallel
dir:

- Several changes in Schedule.c (scheduling loop; createThreads etc);
  should only affect parallel code
- Added ghc/rts/hooks/ShutdownEachPEHook.c
- ghc/rts/Linker.[ch]: GUM doesn't know about Stable Names (ifdefs)!!
- StgMiscClosures.h: END_TSO_QUEUE etc now defined here (from StgMiscClosures.hc)
                     END_ECAF_LIST was missing a leading stg_
- SchedAPI.h: taskStart now defined in here; it's only a wrapper around
              scheduleThread now, but might use some init, shutdown later
- RtsAPI.h: I have nuked the def of rts_evalNothing

* Compiler:

- ghc/compiler/main/DriverState.hs
  added PVM-ish flags to the parallel way
  added new ways for parallel ticky profiling and distributed exec

- ghc/compiler/main/DriverPipeline.hs
  added a fct run_phase_MoveBinary which is called with way=mp after linking;
  it moves the bin file into a PVM dir and produces a wrapper script for
  parallel execution
  maybe cleaner to add a MoveBinary phase in DriverPhases.hs but this way
  it's less intrusive and MoveBinary makes probably only sense for mp anyway

* Nofib:

- nofib/spectral/Makefile, nofib/real/Makefile, ghc/tests/programs/Makefile:
  modified to skip some tests if HWL_NOFIB_HACK is set; only tmp to record
  which test prgs cause problems in my working build right now

23 years ago[project @ 2001-03-21 15:33:47 by simonmar]
simonmar [Wed, 21 Mar 2001 15:33:47 +0000 (15:33 +0000)]
[project @ 2001-03-21 15:33:47 by simonmar]
remove bits left over from STG hugs.

23 years ago[project @ 2001-03-21 15:14:27 by simonmar]
simonmar [Wed, 21 Mar 2001 15:14:27 +0000 (15:14 +0000)]
[project @ 2001-03-21 15:14:27 by simonmar]
tidy up, indent, and make this match reality

23 years ago[project @ 2001-03-21 14:26:30 by simonmar]
simonmar [Wed, 21 Mar 2001 14:26:30 +0000 (14:26 +0000)]
[project @ 2001-03-21 14:26:30 by simonmar]
Fix for a harmless assertion failure (reachable_from included package
modules).

Also, clean up the downsweep code, and make it more efficient.

23 years ago[project @ 2001-03-21 14:25:37 by simonmar]
simonmar [Wed, 21 Mar 2001 14:25:37 +0000 (14:25 +0000)]
[project @ 2001-03-21 14:25:37 by simonmar]
GHCi test updates

23 years ago[project @ 2001-03-21 11:17:00 by sewardj]
sewardj [Wed, 21 Mar 2001 11:17:00 +0000 (11:17 +0000)]
[project @ 2001-03-21 11:17:00 by sewardj]
Implement tagToEnum# for the bytecode system.  Blargh.  We spot tail-calls
   tagToEnum# <type> arg
and emit code to push the arg, then do a bytecode test-sequence to
determine what value it is, push the relevant constructor, and merge
control flow again, at a label which does the normal tail-call
sequence: slide the constructor down to the sequel and enter it.

Blargyle, or as some would say, barferama.

23 years ago[project @ 2001-03-21 10:56:04 by sewardj]
sewardj [Wed, 21 Mar 2001 10:56:04 +0000 (10:56 +0000)]
[project @ 2001-03-21 10:56:04 by sewardj]
RTS support for the ugly tagToEnum# hack.  Actually a very general
thing -- just a bytecode unconditional jump, so we can do more general
control-flow in BCOs.

23 years ago[project @ 2001-03-20 21:21:46 by panne]
panne [Tue, 20 Mar 2001 21:21:46 +0000 (21:21 +0000)]
[project @ 2001-03-20 21:21:46 by panne]
SuSE's Version 7.1 once again broke DocBook, so as a last resort we
use a home-made CATALOG.   >:-(

23 years ago[project @ 2001-03-20 15:57:30 by simonmar]
simonmar [Tue, 20 Mar 2001 15:57:30 +0000 (15:57 +0000)]
[project @ 2001-03-20 15:57:30 by simonmar]
play quoting games to hopefully get this to build with both ghc-4.08.2
and ghc-4.11+.

23 years ago[project @ 2001-03-20 15:37:52 by simonmar]
simonmar [Tue, 20 Mar 2001 15:37:52 +0000 (15:37 +0000)]
[project @ 2001-03-20 15:37:52 by simonmar]
- add index terms
- elaborate the description of the 'it' variable

23 years ago[project @ 2001-03-20 15:37:05 by simonmar]
simonmar [Tue, 20 Mar 2001 15:37:05 +0000 (15:37 +0000)]
[project @ 2001-03-20 15:37:05 by simonmar]
markup fix.