GitLab

- give rts_n_free_capabilities an interpretation
  in threaded mode (possible values: 0,1)
- noFreeCapabilities() -? noCapabilities()

Switch over to the new hierarchical libraries
---------------------------------------------

This commit reorganises our libraries to use the new hierarchical
module namespace extension.

The basic story is this:

   - fptools/libraries contains the new hierarchical libraries.
     Everything in here is "clean", i.e. most deprecated stuff has
     been removed.

	- fptools/libraries/base is the new base package
	  (replacing "std") and contains roughly what was previously
	  in std, lang, and concurrent, minus deprecated stuff.
	  Things that are *not allowed* in libraries/base include:
		Addr, ForeignObj, ByteArray, MutableByteArray,
		_casm_, _ccall_, ``'', PrimIO

	  For ByteArrays and MutableByteArrays we use UArray and
	  STUArray/IOUArray respectively now.

	  Modules previously called PrelFoo are now under
	  fptools/libraries/GHC.  eg. PrelBase is now GHC.Base.

	- fptools/libraries/haskell98 provides the Haskell 98 std.
	  libraries (Char, IO, Numeric etc.) as a package.  This
	  package is enabled by default.

	- fptools/libraries/network is a rearranged version of
	  the existing net package (the old package net is still
	  available; see below).

	- Other packages will migrate to fptools/libraries in
	  due course.

     NB. you need to checkout fptools/libraries as well as
     fptools/hslibs now.  The nightly build scripts will need to be
     tweaked.

   - fptools/hslibs still contains (almost) the same stuff as before.
     Where libraries have moved into the new hierarchy, the hslibs
     version contains a "stub" that just re-exports the new version.
     The idea is that code will gradually migrate from fptools/hslibs
     into fptools/libraries as it gets cleaned up, and in a version or
     two we can remove the old packages altogether.

   - I've taken the opportunity to make some changes to the build
     system, ripping out the old hslibs Makefile stuff from
     mk/target.mk; the new package building Makefile code is in
     mk/package.mk (auto-included from mk/target.mk).

     The main improvement is that packages now register themselves at
     make boot time using ghc-pkg, and the monolithic package.conf
     in ghc/driver is gone.

     I've updated the standard packages but haven't tested win32,
     graphics, xlib, object-io, or OpenGL yet.  The Makefiles in
     these packages may need some further tweaks, and they'll need
     pkg.conf.in files added.

   - Unfortunately all this rearrangement meant I had to bump the
     interface-file version and create a bunch of .hi-boot-6 files :-(

Fix bug in previous commit

fromInt ==> fromIntegral

bring stuff in parallel/ into scope when running 'mkdependC'

comment wibble

This fixes a problem with the recent clean up in the parsing of type
declarations.  The cleaned up version was correct by H98 (I think), but
unfortunately won't compile the Prelude anymore.

I made the definition a little bit more liberal as follows:
- In `tycl_hdr' allow `gtycon' instead of just `tycon' for the defined type
  constructor.  This allows beyond H98 special syntax like "[]" etc as well as
  tycon ops in parenthesis.  Moreover, it allows qualified names, but the
  renamer flags the latter as errors.
- Allow `gtycon' instead of only `qtycon' for the class name in definitions of
  the form `Eq a => T a' to avoid a whole list of s/r conflicts.  (I have *not*
  checked that the renamer flags misuse, but I would expect that it does.)

ToDo: I think, the renamer should raise an error for all these additional
      forms *unless* -fglasgow-exts is given.

In PrelBase, I needed to replace the infix notation by prefix notation in the
definition of `:+:' and `:*:'.

- Addr ==> Ptr
- Don't directly import PrelIOBase & PrelBase

fromInt ==> fromIntegral

Wibble

Fix lexer bug: we didn't look far enough ahead when parsing 1.0e+x
(i.e. the e+x shouldn't be treated as an exponent).

Documentation for kinded declarations

----------------------------------
	Implement kinded type declarations
	----------------------------------

This commit allows the programmer to supply kinds in
	* data decls
	* type decls
	* class decls
	* 'forall's in types

e.g. 	data T (x :: *->*) = MkT

        type Composer c = forall (x :: * -> *) (y :: * -> *) (z :: * -> *).
		          (c y z) -> (c x y) -> (c x z);

This is occasionally useful.

It turned out to be convenient to add the form

	(type :: kind)

to the syntax of types too, so you can put kind signatures in types as well.

fromInt ==> fromIntegral

Documentation for the parallel array extension merged earlier.  (This
documents Milestone 1 as well as some of the basics of flattening.)

Add a missing dependency (submitted by Volker Stolz; thanks)

------------------------------
		Towards kinded data type decls
		------------------------------

Move towards being able to have 'kinded' data type decls.
The burden of this commit, though, is to tidy up the parsing
of data type decls.  Previously we had

	data ctype '=' constrs

where the 'ctype' is a completetely general polymorphic type.
	forall a. (Eq a) => T a

Then a separate function checked that it was of a suitably restricted
form.  The reason for this is the usual thing --- it's hard to tell
when looking at

	data Eq a => T a = ...

whether you are reading the data type or the context when you have
only got as far as 'Eq a'.

However, the 'ctype' trick doesn't work if we want to allow

	data T (a :: * -> *) = ...

So we have to parse the data type decl in a more serious way.
That's what this commit does, and it makes the grammar look much nicer.
The main new producion is tycl_hdr.

Remove unused rules

Updated to the use of [~N] in rules implementing foldr/build.

*******************************
		       * Merging from ghc-ndp-branch *
		       *******************************

This commit merges the current state of the "parallel array extension" and
includes the following:

* (Almost) completed Milestone 1:
  - The option `-fparr' activates the H98 extension for parallel arrays.
  - These changes have a high likelihood of conflicting (in the CVS sense)
    with other changes to GHC and are the reason for merging now.
  - ToDo: There are still some (less often used) functions not implemented in
	  `PrelPArr' and a mechanism is needed to automatically import
	  `PrelPArr' iff `-fparr' is given.  Documentation that should go into
	  the Commentary is currently in `ghc/compiler/ndpFlatten/TODO'.

* Partial Milestone 2:
  - The option `-fflatten' activates the flattening transformation and `-ndp'
    selects the "ndp" way (where all libraries have to be compiled with
    flattening).  The way option `-ndp' automagically turns on `-fparr' and
    `-fflatten'.
  - Almost all changes are in the new directory `ndpFlatten' and shouldn't
    affect the rest of the compiler.  The only exception are the options and
    the points in `HscMain' where the flattening phase is called when
    `-fflatten' is given.
  - This isn't usable yet, but already implements function lifting,
    vectorisation, and a new analysis that determines which parts of a module
    have to undergo the flattening transformation.  Missing are data structure
    and function specialisation, the unboxed array library (including fusion
    rules), and lots of testing.

I have just run the regression tests on the thing without any problems.  So,
it seems, as if we haven't broken anything crucial.

Remove dependency on package text.

Avoid using findPS/packString where simple string matching will do.

Don't use RegexString to parse OPTIONS pragmas, instead do the
matching by hand, thereby removing a dependency on RegexString (and
hence Regex).

Back out rev 1.28.  In retrospect saving the pre-messed-with %esp in a
global location is thread/reentrancy unsafe and a brain-dead thing to
do, and we can't think of a clean way to fix this.  So am backing it out.

Improve wording slightly

naive script for preparing a binary-dist tree for the installer

declare MainCapability + simplify releaseCapability()

Describe ultrakludge^Hclever engineering which makes IO types work in the
interpreter.

Add many details about bytecode generation, the interpreter, and
compiled/interpreted code interop.

Remove duplication of CVS info, and improve Win32 notes

-------------------------------------------
	Improve the "stragely-kinded tyvar" problem
	-------------------------------------------

When the type checker finds a type variable with no binding,
which means it can be instantiated with an arbitrary type, it
usually instantiates it to Void.  Eg.

	length []
===>
	length Void (Nil Void)

But in really obscure programs, the type variable might have
a kind other than *, so we need to invent a suitably-kinded type.

This commit uses
	Void for kind *
	List for kind *->*
	Tuple for kind *->...*->*

which deals with most cases.  (Previously, it only dealt with
kind *.)

In the other cases, it just makes up a TyCon with a suitable
kind.  If this gets into an interface file, anyone reading that
file won't understand it.  This is fixable (by making the client
of the interface file make up a TyCon too) but it is tiresome and
never happens, so I am leaving it.

Most of the added lines are comments.

--------------------------------------------------
	Slurp in a few more instance decls with ghc --make
	--------------------------------------------------

ghc --make wasn't slurping in quite enough instance decls.
The relevant comment is in RnIfaces; the new part is marked.
George Russel's Uniform showed this up.

We slurp in an instance decl from the gated instance pool iff

	all its gates are either in the gates of the module,
	or are a previously-loaded tycon or class.

The latter constraint is because there might have been an instance
decl slurped in during an earlier compilation, like this:

	instance Foo a => Baz (Maybe a) where ...

In the module being compiled we might need (Baz (Maybe T)), where T
is defined in this module, and hence we need (Foo T).  So @Foo@ becomes
a gate.  But there's no way to 'see' that.

NEW:  More generally, types might be involved as well:
NEW:  	instance Foo2 T a => Baz2 a where ...
NEW:
NEW: Now we must treat T as a gate too, as well as Foo.  So the solution
NEW: we adopt is:
NEW:
NEW: 	we simply treat all previously-loaded
NEW: 	tycons and classes as gates.
NEW:
NEW: This gloss only affects ghc --make and ghc --interactive.

----------------------------------------------------
	Make TcType.match and TcUnify.uUnboundVar kind-aware
	----------------------------------------------------

George Russel had apparently-overlapping (ha) instance decls like

	instance .. => C (a b) where
	instance .. => C (x y) where

But the a,b and x,y were different kinds!  Turned out that TcType.unify
was kind-aware (so we didn't report a duplicate instance decl, but TcType.match
was not (so we simply selected the wrong one, and got a mis-kinded constraint
popping up from the ".."  part.  Very exciting to track down.

I also make the ordinary unification kind-aware in the same way.  It's
quite legitimate to attempt to unify, say,
	(a b)    with     (c d)
but the unification should fail if a's kind differs from c's.
(There was a kind of debug warning before, but it's actually not an error
in the compiler... so it should just make unification fail gracefully.)

Better pretty printing

clean $(HC_OBJS) too

- update some comments

- split up DERIVED_SRCS into separate variables so we can refer to
  them separately in the specification of the various CLEAN_FILES
  variables.

- only clean happy-generated files in maintainer-clean mode

Little bit more doc on suspended_ccalling_threads

unbreak stdcall handling (caused by recent change to Outputable instance for CCallConv)

Eliminate all vestiages of UsageTy, in preparation for
	Keith's new version.  Hurrah!

	Keith: LBVarInfo and usOnce,usMany are still there,
	because I know you have eliminated LBVarInfo, and I didn't
	want to cause unnecessary conflicts.

x86 only: make %esp be 8-aligned before entering HC-world code.
This avoids misalignment penalties for C doubles stored on the C
stack.  A quick test using nofib/imaginary/rfib shows that getting
this wrong increases run time by about 10% on our 1 GHz PIII.