GitLab

This patch embodies many, many changes to the contraint solver, which
make it simpler, more robust, and more beautiful.  But it has taken
me ages to get right. The forcing issue was some obscure programs
involving recursive dictionaries, but these eventually led to a
massive refactoring sweep.

Main changes are:
 * No more "frozen errors" in the monad.  Instead "insoluble
   constraints" are now part of the WantedConstraints type.

 * The WantedConstraint type is a product of bags, instead of (as
   before) a bag of sums.  This eliminates a good deal of tagging and
   untagging.

 * This same WantedConstraints data type is used
     - As the way that constraints are gathered
     - As a field of an implication constraint
     - As both argument and result of solveWanted
     - As the argument to reportUnsolved

 * We do not generate any evidence for Derived constraints. They are
   purely there to allow "impovement" by unifying unification
   variables.

 * In consequence, nothing is ever *rewritten* by a Derived
   constraint.  This removes, by construction, all the horrible
   potential recursive-dictionary loops that were making us tear our
   hair out.  No more isGoodRecEv search either. Hurrah!

 * We add the superclass Derived constraints during canonicalisation,
   after checking for duplicates.  So fewer superclass constraints
   are generated than before.

 * Skolem tc-tyvars no longer carry SkolemInfo.  Instead, the
   SkolemInfo lives in the GivenLoc of the Implication, where it
   can be tidied, zonked, and substituted nicely.  This alone is
   a major improvement.

 * Tidying is improved, so that we tend to get t1, t2, t3, rather
   than t1, t11, t111, etc

   Moreover, unification variables are always printed with a digit
   (thus a0, a1, etc), so that plain 'a' is available for a skolem
   arising from a type signature etc. In this way,
     (a) We quietly say which variables are unification variables,
         for those who know and care
     (b) Types tend to get printed as the user expects.  If he writes
             f :: a -> a
             f = ...blah...
         then types involving 'a' get printed with 'a', rather than
         some tidied variant.

 * There are significant improvements in error messages, notably
   in the "Cannot deduce X from Y" messages.

This patch fixes Trac #4875.  The main point is to do dependency
analysis on type and class declarations, and kind-check them in
dependency order, so as to improve error messages.

This patch means that a few programs that would typecheck before won't
typecheck any more; but before we were (naughtily) going beyond
Haskell 98 without any language-extension flags, and Trac #4875
convinces me that doing so is a Bad Idea.

Here's an example that won't typecheck any more
       data T a b = MkT (a b)
       type F k = T k Maybe

If you look at T on its own you'd default 'a' to kind *->*;
and then kind-checking would fail on F.

But GHC currently accepts this program beause it looks at
the *occurrences* of T.

For single-character identifiers we now don't make any suggestions
See comments in Util.fuzzyLookup

We need the unfolding even for a *recursive* function (indeed
that's the point) and I was using the wrong function to get it
(idUnfolding rather than realIdUnfolding).

They're a little nicer now, and a regression in the cygwin build is
fixed (the $i in the destination wasn't surviving being passed through
cygpath).

As far as I can tell, it is the responsibility of the program to save
and restore its own terminal settings across a suspend/foreground, the
shell doesn't do it (which seems odd).  So I've added a signal handler
for SIGTSTP to the RTS which will save and restore the terminal
settings iff we modified them with hSetBuffering or hSetEcho (we
already restore them at exit time in these cases).

cygwin's /bin/install doesn't set file modes correctly if the
destination path is a C: style path:

$ /bin/install -c -m 644 foo /cygdrive/c/cygwin/home/ian/foo2
$ /bin/install -c -m 644 foo c:/cygwin/home/ian/foo3
$ ls -l foo*
-rw-r--r-- 1 ian None 0 2011-01-06 18:28 foo
-rw-r--r-- 1 ian None 0 2011-01-06 18:29 foo2
-rwxrwxrwx 1 ian None 0 2011-01-06 18:29 foo3

This causes problems for bindisttest/checkBinaries.sh which then
thinks that e.g. the userguide HTML files are binaries.

We therefore use a /cygdrive path if we are on cygwin

This patch adds support for multiline commands in GHCi.

The first line of input is lexed.  If there is an active
layout context once the lexer reaches the end of file, the
user is prompted for more input.

Multiline input is exited by an empty line and can be escaped 
with a user interrupt.

Multiline mode is toggled with `:set +m`

Using Haskell conditionals means the compiler sees all the code, so
there should be less rot of code specific to uncommon arches. Code
for other platforms should still be optimised away, although if we want
to support targetting other arches then we'll need to compile it
for-real anyway.

This patch fixes what seems to be the last problem with the --with-gcc
option.  On OS X, we need to pass the path to gcc to dtrace as the
preprocessor.  (Internally, dtrace on OS X sets the default preprocessor
to /usr/bin/gcc.)  ATM, dtrace is only supported on OS X, so we don't
need any conditionalization.  If dtrace is ported to other platforms,
we might need to change this. However, usage on other platforms will
probably be similar to OS X, since many of Apple's changes are to
use the gnu toolchain instead of the Sun toolchain.
  

This small change means that if you have
     type Age = Int
and you try to unify Age and Bool, you'll get a complaint about
not matching Age and Bool, rather than Int and Bool.  See the notes
with Trac #4525

..so that you get a more helpful message when
you mis-spell a module name in an 'import'.

Validates, but not fully tested.

Based on Max's patch in Trac #2442, but heavily refactored.

...so that you get helpful suggestions when you mis-spell a name
Based on Max's patch in Trac #2442, but heavily refactored.

Plus, I changed quite a bit of layout to make the lines shorter.

For a long time an 'mdo' expression has had a SyntaxTable
attached to it.  However, we're busy deprecating SyntaxTables
in favour of rebindable syntax attached to individual Stmts,
and MDoExpr was totally inconsistent with DoExpr in this
regard.

This patch tidies it all up.  Now there's no SyntaxTable on
MDoExpr, and 'modo' is generally handled much more like 'do'.

There is resulting small change in behaviour: now MonadFix is
required only if you actually *use* recursion in mdo. This
seems consistent with the implicit dependency analysis that
is done for mdo.

Still to do:
  * Deal with #4148 (this patch is on the way)
  * Get rid of the last remaining SyntaxTable on HsCmdTop

This fixes a somewhat obscure situation in which an
over-optimistic use of "occurs once" led to an infinite
sequence of simplifier iterations.  Se Note [Cascading inlines]
for the details.

This showed up when compiling rather large DPH programs, which
run lots of iterations of the simplifier, which in turn made
compilation take much longer than necessary.