Signed-off-by: Austin Seipp <austin@well-typed.com>

Signed-off-by: Austin Seipp <austin@well-typed.com>

Signed-off-by: Austin Seipp <austin@well-typed.com>

See merge commit 66693401



This commit also adds a check for a left-over testsuite/.git folder to sync-all

This way, the first time sync-all is called after updating to a
post-testsuite-merge (see #8545) state of ghc.git, the sync-all script
aborts with an error message if a `testsuite/.git` folder is detected and
thus forces the user to take action.

Signed-off-by: Herbert Valerio Riedel <hvr@gnu.org>

This commit performs a subtree merge of testsuite.git into ghc.git;
The next commit will adapt `sync-all` et al. to the new situation.

At the time of merge, testsuite.git was at commit
[998a816ae89c4fd573f4abd7c6abb346cf7ee9af/testsuite]

The following steps have been used to accomplish this merge:

1. Clone a fresh testsuite.git copy (& cd into)

2. Remove accidentally committed binary files from history

 git filter-branch \
  --index-filter "git rm -r --cached --ignore-unmatch \
    tests/haddock/should_compile_flag_nohaddock/a.out \
    tests/haddock/should_compile_noflag_nohaddock/a.out \
    tests/ghc-regress/haddock/should_compile_flag_nohaddock/a.out \
    tests/ghc-regress/haddock/should_compile_noflag_nohaddock/a.out \
    tests/ghc-regress/dph/diophantine/dph-diophantine-fast \
    tests/ghc-regress/dph/diophantine/dph-diophantine-opt \
    tests/ghc-regress/dph/primespj/dph-primespj-fast \
    tests/ghc-regress/dph/quickhull/dph-quickhull-fast \
    tests/ghc-regress/dph/smvm/dph-smvm \
    tests/ghc-regress/dph/sumnats/dph-sumnats \
    tests/ghc-regress/dph/words/dph-words-fast \
    tests/ghc-regress/plugins/plugins01" \
   HEAD

3. Rename all paths in testsuite.git to be prefixed with `testsuite/`

 git filter-branch -f --prune-empty --tree-filter \
  "mkdir -p testsuite; \
   git ls-tree --name-only \$GIT_COMMIT | xargs -I files mv files testsuite/"

4. cd into ghc/ checkout, and perform subtree merge of testsuite into ghc
   (see also http://nuclearsquid.com/writings/subtree-merging-and-you/

)

 cd ../ghc/
 git remote add -f testsuite ../testsuite/.git
 git merge -s ours --no-commit testsuite/master
 git read-tree --prefix=/ -u testsuite/master
 git commit

Signed-off-by: Herbert Valerio Riedel <hvr@gnu.org>

Richard, you may want to regenerate the PDF
as I get

! LaTeX Error: File `stmaryrd.sty' not found.

with my setup and do not want to mess things up.

Also for me ?? appear twice on top of the page 13
and I have tp 'rm core-spec.pdf; make' to get it
right.

Signed-off-by: Herbert Valerio Riedel <hvr@gnu.org>

Signed-off-by: Herbert Valerio Riedel <hvr@gnu.org>

This change is just some documentation around ignoring the context
of an enclosing instance when processing `deriving` clauses of an
associated data instance.

We do *not* propagate kind information from an instance declaration's
members back into the instance head.

If we don't do this, then in various GHC API scenarios
(which use runGhc) with tracing/debugging/ASSERTs on,
we try to read those unsafe global dynamic flags and
find them uninitialised.

Previously they just used a fixed width of 100, ignoring
-dppr-cols.  I think this dates back to a time when
the flag didn't exist, or wasn't conveniently available.

Thanks to Andrew Gibiansky for pointing this out.

Fixes Trac #8616

Explain why defaultDmd resTypeArgDmd are similar, but both needed, and
apply slight code cosmetics.

Mostly improvements, happily

The basic idea here is simple, and described in Note [The interactive package]
in HscTypes, which starts thus:

    Note [The interactive package]
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    Type and class declarations at the command prompt are treated as if
    they were defined in modules
       interactive:Ghci1
       interactive:Ghci2
       ...etc...
    with each bunch of declarations using a new module, all sharing a
    common package 'interactive' (see Module.interactivePackageId, and
    PrelNames.mkInteractiveModule).

    This scheme deals well with shadowing.  For example:

       ghci> data T = A
       ghci> data T = B
       ghci> :i A
       data Ghci1.T = A  -- Defined at <interactive>:2:10

    Here we must display info about constructor A, but its type T has been
    shadowed by the second declaration.  But it has a respectable
    qualified name (Ghci1.T), and its source location says where it was
    defined.

    So the main invariant continues to hold, that in any session an original
    name M.T only refers to oe unique thing.  (In a previous iteration both
    the T's above were called :Interactive.T, albeit with different uniques,
    which gave rise to all sorts of trouble.)

This scheme deals nicely with the original problem.  It allows us to
eliminate a couple of grotseque hacks
  - Note [Outputable Orig RdrName] in HscTypes
  - Note [interactive name cache] in IfaceEnv
(both these comments have gone, because the hacks they describe are no
longer necessary). I was also able to simplify Outputable.QueryQualifyName,
so that it takes a Module/OccName as args rather than a Name.

However, matters are never simple, and this change took me an
unreasonably long time to get right.  There are some details in
Note [The interactive package] in HscTypes.

Sometimes we actually have a good SrcSpan for the type constructor
and reporting that is better than just reporting which module it
was defined on

I did quite a bit of restructuring, as well as adding
the note specifically referred to in #8622

Trac #8651 revealed that my previous fix (itself in response to #8644)
wasn't quite right.  The plan, using the CtOrigin to identify
constraints arising from flattening, is described in TcSimplify,
Note [When does an implication have given equalities?]

Signed-off-by: Austin Seipp <austin@well-typed.com>

This resulted form a discussion about #5916.

Signed-off-by: Austin Seipp <austin@well-typed.com>

When reading the program from standard input, runghc did not properly
handle the --ghc-arg= escape for arguments to ghc which do not start
with a dash, since arguments were processed twice and the first time the
--ghc-arg= was stripped.  Now arguments are only processed once.  For
backwards compatibility, a prefix of --ghc-arg=--ghc-arg= is allowed
since this prefix will work on both old and new versions of ghc.

This fixes #8601



Signed-off-by: Austin Seipp <austin@well-typed.com>

Signed-off-by: Austin Seipp <austin@well-typed.com>

Signed-off-by: Austin Seipp <austin@well-typed.com>

Previously a few symbols weren't flipped from %function to %object
as the section splitter was emitting them without processes. This
may be a bug in itself but for now let's just work around the issue
but rewriting all symbol `.types`.

Signed-off-by: Austin Seipp <austin@well-typed.com>

Signed-off-by: Austin Seipp <austin@well-typed.com>

Signed-off-by: Austin Seipp <austin@well-typed.com>