GitLab

NO_INSTALL_LIBRARY=YES

Fix library installation: we now install libraries *unless*
NO_INSTALL_LIBRARY=YES.

GC changes: instead of threading old-generation mutable lists
through objects in the heap, keep it in a separate flat array.

This has some advantages:

  - the IND_OLDGEN object is now only 2 words, so the minimum
    size of a THUNK is now 2 words instead of 3.  This saves
    some amount of allocation (about 2% on average according to
    my measurements), and is more friendly to the cache by
    squashing objects together more.

  - keeping the mutable list separate from the IND object
    will be necessary for our multiprocessor implementation.

  - removing the mut_link field makes the layout of some objects
    more uniform, leading to less complexity and special cases.

  - I also unified the two mutable lists (mut_once_list and mut_list)
    into a single mutable list, which lead to more simplifications
    in the GC.

xrefs to Cabal docs

Fix previous commit. Got distracted, thought I had already tested.

Fix Mac OS X-specific breakage introduced by rev. 1.67 of ghc/includes/Stg.h.
Adding the used attribute to static functions caused gcc to emit
Apple-specific .no_dead_strip directives, which messed things up.

Add a special case for Mac OS X to FP_PROG_AR_NEEDS_RANLIB.
On Mac OS X, we need to run ranlib after installing .a files because
they contain a timestamp which will be outdated after installing.

We could do a proper test for that, but it would contain the command
'sleep 6' and still apply only to one platform.

PowerPC: mangler trouble again

PowerPC: #ifdef darwin_REGS -> #if darwin_REGS

Keep .type/.size directives for _entry symbols, but rename them to
_info.  This helps cachegrind to find where my instructions are going.

Add comment

oops, fix building of GHCi libraries.  We only want to avoid the
*install* if this isn't a package.

Only build GHCi libs, install libs, and make Haddock docs for packages
(not ordinary libraries, like libghccompat.a).

Happy 1.15 is now required

Remove trace in mkGenericRhs (merge to STABLE)

Fix doc building for binary distributions

Missing commit from the module A export list commit

-----------------------------------------------------
	Switch off derivable type class generation by default
	-----------------------------------------------------

	Merge to STABLE

This commit switches off the generation of the support code for 'Derivable
Type Classes' by default.  Use -fgenerics to switch it on.

Motivation: seldom used, and there's a nasty blowup in Core types for data
types that have lots of constructors or lots of fields.

We now put -fgenerics in the GhcLibOpts, so that the libraries have the
support code, as before.

------------------------------
	Improve unused-import warnings
	------------------------------

	Merge to STABLE

This commit improves the warning messages for unused imports, in the
case where the 'module A' form is used in an export list.

In doing this, I've realised that the unused-import checking is deficient
in several ways.  At least

	a) it doesn't recognise that there might be several import statements
	   for the same module (TcRnTypes.imp_mods has only one entry per
	   module

	b) it doesn't understand about qualified modules at all

	c) even more fundamentally, it starts from the used Names,
	   but if the module mentions (say) aliases M.f and N.f for
	   the same Name, then two imports might be necessary for it

I'm not going to fix these problems now; this message just records them.

Mention that GHC is required for building GHC.

Point directly to the Building Guide from here.

Better documentation for unboxed types; merge to STABLE

Wibble to message

------------------------------------------------------
	Report top-level implicit parameter errors more nicely
	------------------------------------------------------

Consider
    module Main where

	main = let ?x = 5 in print foo

	foo = woggle 3

	woggle :: (?x :: Int) => Int -> Int
	woggle y = ?x + y

GHC's current rules say that 'foo' is monomorphic, so we get
	foo :: Int
but we also get an unbound top-level constraint (?x::Int).  GHC 6.2 emits a
message like:
     Unbound implicit parameter (?x::Int)
     arising from use of `woggle' at ...

The point is that THERE IS NO WAY FOR THIS CONSTRAINT TO GET BOUND,
because we don't have a top-level binding form for implicit parameters.
So it's stupid for 'foo' to be monomorphic.

This commit improves matters by giving a much nicer error message:

     Implicit parameters escape from the monomorphic top-level binding(s) of `foo':
       ?x::Int arising from use of `woggle' at tcfail130.hs:10:6-11
     Probably fix: add type signatures for the top-level binding(s)
     When generalising the type(s) for `foo'

Respect --exclude-module in ghc -M; some tidying up as well

Comments and type sig

The HEAD is now 6.5

In the fragments of Haskell code which setup buffering and flushing
for the std Handles in GHCi, qualify names with System.IO rather than IO.

Need to add -I. when compiling with GHC 6.3+

Fix another substitution-related bug in CoreLint

Move the in_haskell check to somewhere that works for the threaded RTS too.

add a prototype

Make setProgArgv() set the RTS's idea of the prog name too.

Document dependency printing

Dump package config only with -v4
Dump module dependencies (ghc -M) with -v2

Tidy output

Make sure that the argument of DataToTag is evaluated; rather a horrible piece of code, I fear

Import trimming

Add a helpful sanity check: if we try to re-enter the scheduler from
Haskell code without going via suspendThread()/resumeThread(), such as
when you have a 'foreign import unsafe' that should be 'safe', then you
get an error message.

include ghcplatform.h