GitLab

Since we can't run stage 2 on the host.

This required a little adjustment in zonkQuantifiedTyVars

The "consistency" in this case is beteween the instance head
and the associated type instance head, which is made trickier
by the presence of kind variables that are not explicitly mentioned
in the class head.

See Note [Checking consistent instantiation] in TcInstDcls

This fixes Trac #7282.

Function responsible for parsing the static flags, that were spread
across two modules (StaticFlags and StaticFlagParser), are now
in one file. This is analogous to dynamic flags parsing, which is
also contained within a single module.
Signed-off-by: David Terei <davidterei@gmail.com>

* Move tidyType and friends from TcType to TypeRep
  (It was always wrong to have it in TcType.)

* Move mkCoAxBranch and friends from FamInst to Coercion

* Move pprCoAxBranch and friends from FamInstEnv to Coercion

No change in functionality, though there might be a little
wibble in error message output, because I combined two different
functions both called pprCoAxBranch!

Quite a bit of tidying up here; the fix to #7524 is actually
only a small part.

* Be fully clear that the cab_tvs in a CoAxBranch are not
  fresh.  See Note [CoAxBranch type variables] in CoAxiom.

* Use CoAxBranch to replace the ATDfeault type in Class.
  CoAxBranch is perfect here.  This change allowed me to
  delete quite a bit of boilerplate code, including the
  corresponding IfaceSynType.

* Tidy up the construction of CoAxBranches, and when FamIntBranch is
  freshened.  The latter onw happens only in FamInst.newFamInst.

* Tidy the tyvars of a CoAxBranch when we build them, done in
  FamInst.mkCoAxBranch.  See Note [Tidy axioms when we build them]
  in that module.  This is what fixes #7524.

Much niceer now.

The bug where TSOs were unconditionally kept on the mutable list was #1589
which was fixed in 04cddd33.
Curiously enough, the commit that changed this comment
0417404f occurred *after* this
change was made; I can only assume Simon Marlow accidentally forgot
that he had fixed this bug. :-)
Signed-off-by: Edward Z. Yang <ezyang@mit.edu>

The .git file in a submodule's root directory is not necessarily a
directory, as is the case for the 'time' submodule.

This fixes #7577, where some './sync-all' operations don't detect
that the 'time' submodule is present in the repository.

Makes it clearer whether it succeeded when redirecting output

It thought that something impossible was happening when they
were involved.

This program:

    main :: IO ()
    main = do re "[^ ]" "\207"
              re "[ ]" "\207"
              re " " "\207"

    re :: String -> String -> IO ()
    re r str = let r' = makeRegex r :: Regex
                   res = matchM r' str :: Maybe (String, String, String, [String])
               in print res

prints

    Nothing
    Nothing
    Nothing

for me with Posix, but

    Just ("","\207","",[])
    Nothing
    Nothing

with PCRE. This was causing compare to fail with

    Tar line doesn't parse: "drwxrwxr-x simonmar/GHC       0 2012-12-08 21:35 ghc-7.6.1.20121207/libraries/haskeline/tests/dummy-\206\188\206\177\207\\302\\203/"

on the GHC source tarball.

- -float-abi=hard also works for arm v6
- pass -mattr=+vfp2 if we have VFP2

Submitted by: Karel Gardas <karel.gardas@centrum.cz>

Fixes Trac #7601

It should be the case that either an entire mutually recursive
group of data type declarations can be promoted, or none of them.
It's really odd to promote some data constructors of a type but
not others. Eg
  data T a = T1 a | T2 Int
Here T1 is sort-of-promotable but T2 isn't (becuase Int isn't
promotable).

This patch makes it all-or-nothing. At the same time I've made
the TyCon point to its promoted cousin (via the tcPromoted field
of an AlgTyCon), as well as vice versa (via the ty_con field of
PromotedTyCon).

The inference for the group is done in TcTyDecls, the same place
that infers which data types are recursive, another global question.

Signed-off-by: David Terei <davidterei@gmail.com>