More to allow infix exprssions on LHS of rule

Allow infix exprssions on LHS of rule

Comment wibble

Oops, exprIsConApp_maybe should ignore only InlineMe notes.

Add test to check contexts on existential types in the interpreter.

Remove list of packages to install on Windows; the new story is that you
install all of Cygwin.

A bit more about flags for generics

Close files eagerly after analysing their imports.

Add zlib to list of required packages for building.

Implement bci_STKCHECK.

Not quite sure what this test does, but I've had it lying around for a while.

Add an item about the networking library rewrite.

Start a new directory for networking tests, and add a simple test
which starts two threads which communicate via a socket.

The socket subdirectory already contains some tests, but they're about
6 years old and don't appear to be designed to run without user
intervention.

Ignore tyvars on variables bound by case alts when generating unpack code.

Tests for BlockedOnDeadMVar and ThreadKilled exceptions (they generate
no output by default in forked threads).

Don't include the win32 and com packages unless we're building for Windows.

fix HC bootstrapping after LibsReadline no longer has the -l prefixes.

Inline instance dictionary functions.
Remove {-# INLINE instance #-} support and uses.

Add some {-# SCC #-} annotations, and fix a space leak.

SCC notes move inside other notes.

Fix the inlining bug: exprIsConApp_maybe didn't recognize constructor
applications inside notes.

I'm not sure if it's the right place to fix. Should notes be ignored
in other uses of collectArgs? Perhaps they should be ignored inside
the loop of collectArgs? And why the function marked as inline was
not inlined even though it's not a constructor application?

Remove instructions about making /etc/passwd, as it happens automatically
when installing Cygwin.

Say not to install the cygwin package, rather than to install it.

Plumb LibsReadline into extra_libraries for -package util, rather than
extra_ld_opts, so that GHCi knows to load them too.

o=>O:-)

Fix typo

Fix minor typo in generics stuff; could merge with branch

add -fwarn-misc.

document -fwarn-misc.

The convention for warnings currently is: standard warnings indicate
"probable bugs", whereas -W warnings indicate "style problems".
I think -fwarn-misc is more suited to -W.

Added cygwin to list of required packages, and said why gdbm is needed (for
CVS).

HSnet_cbits R.I.P.
(BTW, that's what I meant with "monolithic approach" :-)

Fix test so that it doesnt use undecidable instances

-----------------------------
	Better filtering for warnings
	-----------------------------

* Add Opt_WarnMisc, to enable warnings not otherwise covered by Opt_Warn*
  in the renamer

* Add RnMonad.ifOptRn :: DynFlag -> RnM d a -> RnM d ()
  and use it many places instead of the clumsy direct code

Add comments in tcInstType

Add comments

Improve error message

Makefile.config should be in a source dist.

better "#define IN_STG_CODE 0" before including the standard HC header
in a foreign export dynamic stub, because this isn't really HC code.

Fix the documentation about the -i<dirs> option.  These are scanned
left to right, but before the package directories.  Also add a
suitably severe warning about having a module with the same name as a
package module.