As suggested by SimonM, moved an overview over the three components of the FFI
into the FFI document included as Chapter 7 into the GHC user guide.

Open questions:

* How can I make cross references to the doc in hslibs/?  There are already
  references to hslibs sections in other parts of ghc's user guide, but they
  are not correctly resolved when I generate html from the sgml files.

* There is now a (little) overlap with material in 6.5 (GHC Language Features:
  The foreign interface).  I would say, we could actually leave this as it is,
  as 6.5 goes on to describe GHC specifics.

fix -fno-asm-mangling flag

Force binds with coreBindsSize after every simplifier iteration.
Significantly reduces space use, especially with -O.  This could
probably be done more cleanly.

remove duplicates from the list of ways, so that eg. ghc -prof -prof
is legal.

Don't try to finalize DEAD_WEAKs.  This fix has been sitting in my
tree for so long, I can't remember whether it was a real bugfix or
just a cleanup, oh well.

Set the context switch flag to zero when running a new thread; the
timer interrupt now initiates a context switch.

context switch on a timer interrupt.

remove ToDo about mkdependHS and -osuf.

mkdependHS doesn't understand the -o option anymore; instead it grokss
ghc's -osuf option.  We never really used this, and it behaves
strangely when mixed with -s, but at least now it doesn't conflict
with GHC's -opt<blah> options.

- print the version number as x.yy.z (i.e. replace ", patchlevel z"
with ".z"), but omit the .z if z == 0.

- add --numeric-version flag which prints x.yy.z without the verbage.

- fix -optCrts (was being overriden by -optC)
- fix -osuf with -split-objs
- fix -odir

* makeForeignObj => newForeignObj

* Tried to make type mapping table look less strange, but this
  succeeded only a little bit. Could somebody else have a look?

oops, backout accidental commit

Updated the building from .hc section.

*MERGE*

Time fixes from Michael Weber <michael.weber@post.rwth-aachen.de>:

  * `toClockTime' previously didn't honor the `tz' field of a
    `CalendarTime', which led to time warping when applying

      => (toUTCTime (toClockTime ... (toUTCTime (toClockTime someTime) ... )))

    continuously.

    Now it accepts at least <local>- and UTC-encoded `CalendarTime's
    (TODO: test, whether all timezones work) and converts them
    correctly to <secs from epoch>-format (which is always UTC, as one
    might have guessed).

  * `addToClockTime' now works.

    Previously, `tz' seconds were added(!) when used like:

      => addToClockTime noTimeDiff someTime

    which is clearly wrong.
    Now, the following (hopefully) always holds

      => someTime == (addToClockTime noTimeDiff someTime)

  * `diffClockTimes' works correctly, and is the dual to
    `addToClockTime', i.e.

      => diff == ((addToClockTime diff someTime) `diffClockTimes` someTime)

    should now hold for all diff, someTime

    Previously, it reports ugly diffs at {min,hour,day,...}-breaks,
    for example:

      => "2000/06/18 01:00 UTC"
            `diffClockTimes` "2000/06/17 23:00 UTC" == 1 day, -22 hours

    whereas now it emits "7200 secs". This number can be converted
    with `normalizeTimeDiff' to "2 hours".

  * added `normalizeTimeDiff', which calculates year, month, days,
    etc. out of an unnormalized `TimeDiff' (generated by
    `diffClockTimes', for example)

  * `formatTimeDiff': added the missing "%c" case. The
    format is proprietary, though... Is there a nicer one?

ignore -no-link-chk for backwards compat.

remove tests subdir

Give 'ghc -E' the traditional behaviour: the output is dumped to
stdout.  The new driver also generates <file>.cpp.

Changed SUBDIRS from users_guide to set (the former doesn't build any more
except as part of the latter).

Added Eq and Ord instances for ArrayException

mention cvs-ghc@haskell.org

Add FreeBSD-compatible catalog.  This one works on my FreeBSD 4.0 box
with the following packages installed:

	docbook-3.1
	dsssl-docbook-modular-1.52
	jade-1.2.1
	iso8879-1986

alternatively, installing the 'docproj' meta-package gets everything
you need.

Script that automates building GHC from .hc files.

?? Shall I edit the corresponding section in building.sgml to advise the use
?? of this script (thus replacing the outdated description of the manual
?? steps - which is wrong anyway).

*** MERGE WITH 4.07 ***

* Fix the ambiguity check in TcMonotype.lhs so that
  it is not carried out for types read from interface 
  files.  Some workers may get ambiguous types but that
  does not matter, and should not make compilation fail.
  More detail in the comments with TcMonoType.tc_type_kind
  (the HsForAll case)

* Don't create specialisations for type applications 
  where there's a matching rule.  The rule should
  clearly take precedence.  (Bug reported by Sven.)
  I havn't tested this fix.

* Run the occurrence analyser after tidyCore, so that
  occurrence info (notably dead-var info) is correct
  for the code generators.  This should fix Erik's problem,
  but again I've not tested the fix.  The extra call 
  is in Main.lhs

* Fix CoreToStg so that it can handle an StgLam in mkStgCase.
  This only shows up in a wierd case, documented in 
  CoreToStg.mkStgCase

* SGML_DOC can now hold a _list_ of sgml documentations to build
* both building.sgml and ffi-art.sgml are now build in docs/

*MERGE*

Fix sparc bustage following latest round of NCG hacking (reg-alloc stuff).
Still won't work, but at least should compile again.

instance Eq (HsDecl name pat) : add default case

Quick workaround for Reuben's M$ configuration problems

sigh, fix the ordering of the phases so that splitting works again.

Added new primitives and bytecodes that support
code generation for XMLambda. All additions are
surrounded by #ifdef XMLAMBDA.

Most important additions:
- Rows (n-tuples) which are implemented on top of Frozen Mutarrays
- Inj (variant sums), which is implemented using a new constructor
called Inj which contains both the value and an unboxed int
which represents the index.

Added definition of int64 to make it compilable with both gcc and VisualC++

Added functions to the bytecode assembler that
support code generation for Xmlambda. All additions for
Xmlambda are surrounded by #ifdef XMLAMBDA.

Added definition of int64 to make it compilable with both gcc and VisualC++

urk! the arity of a record selector Id didn't take into account any
dictionary arguments due to the context on the datatype...

(fixes bug on H/OpenGL reported by Sven)

Emit slightly better x86 floating point code for comparisons, +, -,
* and /, in the common case where one of the source fake FP regs
is the same as the destination reg.

Fixed typo: .hs left out of .SUFFIXES list.

Major thing: new register allocator.  Brief description follows.
Should correctly handle code with loops in, even though we don't
generate any such at the moment.  A lot of comments.  The previous
machinery for spilling is retained, as is the idea of a fast-and-easy
initial allocation attempt intended to deal with the majority of code
blocks (about 60% on x86) very cheaply.  Many comments explaining
in detail how it works :-)

The Stix inliner is now on by default.  Integer code seems to run
within about 1% of that -fvia-C.  x86 fp code is significantly worse,
up to about 30% slower, depending on the amount of fp activity.

Minor thing: lazyfication of the top-level NCG plumbing, so that the
NCG doesn't require any greater residency than compiling to C, just a
bit more time.  Created lazyThenUs and lazyMapUs for this purpose.

The new allocator is somewhat, although not catastophically, slower
than the old one.  Fixing of the long-standing NCG space leak more
than makes up for it; overall hsc run-time is down about 5%, due to
significantly reduced GC time.

--------------------------------------------------------------------

Instructions are numbered sequentially, starting at zero.

A flow edge (FE) is a pair of insn numbers (MkFE Int Int) denoting
a possible flow of control from the first insn to the second.

The input to the register allocator is a list of instructions, which
mention Regs.  A Reg can be a RealReg -- a real machine reg -- or a
VirtualReg, which carries a unique.  After allocation, all the
VirtualReg references will have been converted into RealRegs, and
possibly some spill code will have been inserted.

The heart of the register allocator works in four phases.

1.  (find_flow_edges) Calculate all the FEs for the code list.
    Return them not as a [FE], but implicitly, as a pair of
    Array Int [Int], being the successor and predecessor maps
    for instructions.

2.  (calc_liveness) Returns a FiniteMap FE RegSet.  For each
    FE, indicates the set of registers live on that FE.  Note
    that the set includes both RealRegs and VirtualRegs.  The
    former appear because the code could mention fixed register
    usages, and we need to take them into account from the start.

3.  (calc_live_range_sets) Invert the above mapping, giving a
    FiniteMap Reg FeSet, indicating, for each virtual and real
    reg mentioned in the code, which FEs it is live on.

4.  (calc_vreg_to_rreg_mapping) For virtual reg, try and find
    an allocatable real register for it.  Each real register has
    a "current commitment", indicating the set of FEs it is
    currently live on.  A virtual reg v can be assigned to
    real reg r iff v's live-fe-set does not intersect with r's
    current commitment fe-set.  If the assignment is made,
    v's live-fe-set is union'd into r's current commitment fe-set.
    There is also the minor restriction that v and r must be of
    the same register class (integer or floating).

    Once this mapping is established, we simply apply it to the
    input insns, and that's it.

    If no suitable real register can be found, the vreg is mapped
    to itself, and we deem allocation to have failed.  The partially
    allocated code is returned.  The higher echelons of the allocator
    (doGeneralAlloc and runRegAlloc) then cooperate to insert spill
    code and re-run allocation, until a successful allocation is found.

add missing default case in eq_btype

isNeverInlinePrag returned False for a NOINLINE pragma, which seems
wrong to me.  4.07 also has this bug, and it's why Memo is going wrong.

generate _stub.o files properly when there's a -o flag on the command
line, and in the presence of -split-objs.

don't use $(WithGhcHc) to build the driver, use $(HC) instead.