GitLab

This implements the ideas originally put forward in
"System FC with Explicit Kind Equality" (ICFP'13).

There are several noteworthy changes with this patch:
 * We now have casts in types. These change the kind
   of a type. See new constructor `CastTy`.

 * All types and all constructors can be promoted.
   This includes GADT constructors. GADT pattern matches
   take place in type family equations. In Core,
   types can now be applied to coercions via the
   `CoercionTy` constructor.

 * Coercions can now be heterogeneous, relating types
   of different kinds. A coercion proving `t1 :: k1 ~ t2 :: k2`
   proves both that `t1` and `t2` are the same and also that
   `k1` and `k2` are the same.

 * The `Coercion` type has been significantly enhanced.
   The documentation in `docs/core-spec/core-spec.pdf` reflects
   the new reality.

 * The type of `*` is now `*`. No more `BOX`.

 * Users can write explicit kind variables in their code,
   anywhere they can write type variables. For backward compatibility,
   automatic inference of kind-variable binding is still permitted.

 * The new extension `TypeInType` turns on the new user-facing
   features.

 * Type families and synonyms are now promoted to kinds. This causes
   trouble with parsing `*`, leading to the somewhat awkward new
   `HsAppsTy` constructor for `HsType`. This is dispatched with in
   the renamer, where the kind `*` can be told apart from a
   type-level multiplication operator. Without `-XTypeInType` the
   old behavior persists. With `-XTypeInType`, you need to import
   `Data.Kind` to get `*`, also known as `Type`.

 * The kind-checking algorithms in TcHsType have been significantly
   rewritten to allow for enhanced kinds.

 * The new features are still quite experimental and may be in flux.

 * TODO: Several open tickets: #11195, #11196, #11197, #11198, #11203.

 * TODO: Update user manual.

Tickets addressed: #9017, #9173, #7961, #10524, #8566, #11142.
Updates Haddock submodule.

This is the second attempt at merging D757.

This patch implements the idea floated in Trac #9858, namely that we
should generate type-representation information at the data type
declaration site, rather than when solving a Typeable constraint.

However, this turned out quite a bit harder than I expected. I still
think it's the right thing to do, and it's done now, but it was quite
a struggle.

See particularly

 * Note [Grand plan for Typeable] in TcTypeable (which is a new module)
 * Note [The overall promotion story] in DataCon (clarifies existing
stuff)

The most painful bit was that to generate Typeable instances (ie
TyConRepName bindings) for every TyCon is tricky for types in ghc-prim
etc:

 * We need to have enough data types around to *define* a TyCon
 * Many of these types are wired-in

Also, to minimise the code generated for each data type, I wanted to
generate pure data, not CAFs with unpackCString# stuff floating about.

Performance
~~~~~~~~~~~
Three perf/compiler tests start to allocate quite a bit more. This isn't
surprising, because they all allocate zillions of data types, with
practically no other code, esp. T1969

 * T1969:    GHC allocates 19% more
 * T4801:    GHC allocates 13% more
 * T5321FD:  GHC allocates 13% more
 * T9675:    GHC allocates 11% more
 * T783:     GHC allocates 11% more
 * T5642:    GHC allocates 10% more

I'm treating this as acceptable. The payoff comes in Typeable-heavy
code.

Remaining to do
~~~~~~~~~~~~~~~

 * I think that "TyCon" and "Module" are over-generic names to use for
   the runtime type representations used in GHC.Typeable. Better might
be
   "TrTyCon" and "TrModule". But I have not yet done this

 * Add more info the the "TyCon" e.g. source location where it was
   defined

 * Use the new "Module" type to help with Trac Trac #10068

 * It would be possible to generate TyConRepName (ie Typeable
   instances) selectively rather than all the time. We'd need to persist
   the information in interface files. Lacking a motivating reason I
have
   not done this, but it would not be difficult.

Refactoring
~~~~~~~~~~~
As is so often the case, I ended up refactoring more than I intended.
In particular

 * In TyCon, a type *family* (whether type or data) is repesented by a
   FamilyTyCon
     * a algebraic data type (including data/newtype instances) is
       represented by AlgTyCon This wasn't true before; a data family
       was represented as an AlgTyCon. There are some corresponding
       changes in IfaceSyn.

     * Also get rid of the (unhelpfully named) tyConParent.

 * In TyCon define 'Promoted', isomorphic to Maybe, used when things are
   optionally promoted; and use it elsewhere in GHC.

 * Cleanup handling of knownKeyNames

 * Each TyCon, including promoted TyCons, contains its TyConRepName, if
   it has one. This is, in effect, the name of its Typeable instance.

Updates haddock submodule

Test Plan: Let Harbormaster validate

Reviewers: austin, hvr, goldfire

Subscribers: goldfire, thomie

Differential Revision: https://phabricator.haskell.org/D1404

GHC Trac Issues: #9858

This reverts commit bef2f03e.

This merge was botched

Also reverts haddock submodule.

This patch implements the idea floated in Trac #9858, namely that we
should generate type-representation information at the data type
declaration site, rather than when solving a Typeable constraint.

However, this turned out quite a bit harder than I expected. I still
think it's the right thing to do, and it's done now, but it was quite
a struggle.

See particularly

 * Note [Grand plan for Typeable] in TcTypeable (which is a new module)
 * Note [The overall promotion story] in DataCon (clarifies existing stuff)

The most painful bit was that to generate Typeable instances (ie
TyConRepName bindings) for every TyCon is tricky for types in ghc-prim
etc:

 * We need to have enough data types around to *define* a TyCon
 * Many of these types are wired-in

Also, to minimise the code generated for each data type, I wanted to
generate pure data, not CAFs with unpackCString# stuff floating about.

Performance
~~~~~~~~~~~
Three perf/compiler tests start to allocate quite a bit more. This isn't
surprising, because they all allocate zillions of data types, with
practically no other code, esp. T1969

 * T3294:   GHC allocates 110% more (filed #11030 to track this)
 * T1969:   GHC allocates 30% more
 * T4801:   GHC allocates 14% more
 * T5321FD: GHC allocates 13% more
 * T783:    GHC allocates 12% more
 * T9675:   GHC allocates 12% more
 * T5642:   GHC allocates 10% more
 * T9961:   GHC allocates 6% more

 * T9203:   Program allocates 54% less

I'm treating this as acceptable. The payoff comes in Typeable-heavy
code.

Remaining to do
~~~~~~~~~~~~~~~

 * I think that "TyCon" and "Module" are over-generic names to use for
   the runtime type representations used in GHC.Typeable. Better might be
   "TrTyCon" and "TrModule". But I have not yet done this

 * Add more info the the "TyCon" e.g. source location where it was
   defined

 * Use the new "Module" type to help with Trac Trac #10068

 * It would be possible to generate TyConRepName (ie Typeable
   instances) selectively rather than all the time. We'd need to persist
   the information in interface files. Lacking a motivating reason I have
   not done this, but it would not be difficult.

Refactoring
~~~~~~~~~~~
As is so often the case, I ended up refactoring more than I intended.
In particular

 * In TyCon, a type *family* (whether type or data) is repesented by a
   FamilyTyCon
     * a algebraic data type (including data/newtype instances) is
       represented by AlgTyCon This wasn't true before; a data family
       was represented as an AlgTyCon. There are some corresponding
       changes in IfaceSyn.

     * Also get rid of the (unhelpfully named) tyConParent.

 * In TyCon define 'Promoted', isomorphic to Maybe, used when things are
   optionally promoted; and use it elsewhere in GHC.

 * Cleanup handling of knownKeyNames

 * Each TyCon, including promoted TyCons, contains its TyConRepName, if
   it has one. This is, in effect, the name of its Typeable instance.

Requires update of the haddock submodule.

Differential Revision: https://phabricator.haskell.org/D757

This implements DuplicateRecordFields, the first part of the
OverloadedRecordFields extension, as described at
https://ghc.haskell.org/trac/ghc/wiki/Records/OverloadedRecordFields/DuplicateRecordFields

This includes fairly wide-ranging changes in order to allow multiple
records within the same module to use the same field names.  Note that
it does *not* allow record selector functions to be used if they are
ambiguous, and it does not have any form of type-based disambiguation
for selectors (but it does for updates). Subsequent parts will make
overloading selectors possible using orthogonal extensions, as
described on the wiki pages.  This part touches quite a lot of the
codebase, and requires changes to several GHC API datatypes in order
to distinguish between field labels (which may be overloaded) and
selector function names (which are always unique).

The Haddock submodule has been adapted to compile with the GHC API
changes, but it will need further work to properly support modules
that use the DuplicateRecordFields extension.

Test Plan: New tests added in testsuite/tests/overloadedrecflds; these
will be extended once the other parts are implemented.

Reviewers: goldfire, bgamari, simonpj, austin

Subscribers: sjcjoosten, haggholm, mpickering, bgamari, tibbe, thomie,
goldfire

Differential Revision: https://phabricator.haskell.org/D761

Comes with Haddock submodule update.
Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Summary:
DeriveGeneric generates some data types (for data type constructors and for
selectors of those constructors) and instances for those types. This patch
changes name generation for these new types to make it working with data types
with same names imported from different modules and with data types with same
names imported from same modules(using module imports).

Bonus content:

- Some refactoring in `TcGenGenerics.metaTyConsToDerivStuff` to remove some
  redundant partial function applications and to remove a duplicated function.
- Remove some unused names from `OccName`. (those were used for an old
  implementation of `DeriveGeneric`)

Reviewers: kosmikus, simonpj, dreixel, ezyang, bgamari, austin

Reviewed By: bgamari, austin

Subscribers: ezyang, thomie

Differential Revision: https://phabricator.haskell.org/D1081

GHC Trac Issues: #10487

Left in by c89bd681, and otherwise rather annoying during the build!
Signed-off-by: Austin Seipp <austin@well-typed.com>

In the test program from comment:3 of Trac #10370, it turned out
that 25% of all compile time was going in OccName.tidyOccName!

It was all becuase the algorithm for finding an unused OccName
had a quadratic case.

This patch fixes it.  THe effect is pretty big:

Before:
	total time  =       34.30 secs   (34295 ticks @ 1000 us, 1 processor)
	total alloc = 15,496,011,168 bytes  (excludes profiling overheads)

After
	total time  =       25.41 secs   (25415 ticks @ 1000 us, 1 processor)
	total alloc = 11,812,744,816 bytes  (excludes profiling overheads)

Summary: It looks like during .lhs -> .hs switch the comments were not updated. So doing exactly that.

Reviewers: austin, jstolarek, hvr, goldfire

Reviewed By: austin, jstolarek

Subscribers: thomie, goldfire

Differential Revision: https://phabricator.haskell.org/D621

GHC Trac Issues: #9986

The purpose of silent superclass parameters was to solve the
awkward problem of superclass dictinaries being bound to bottom.
See THE PROBLEM in Note [Recursive superclasses] in TcInstDcls

Although the silent-superclass idea worked,

  * It had non-local consequences, and had effects even in Haddock,
    where we had to discard silent parameters before displaying
    instance declarations

  * It had unexpected peformance costs, shown up by Trac #3064 and its
    test case.  In monad-transformer code, when constructing a Monad
    dictionary you had to pass an Applicative dictionary; and to
    construct that you neede a Functor dictionary. Yet these extra
    dictionaries were often never used.  (All this got much worse when
    we added Applicative as a superclass of Monad.) Test T3064
    compiled *far* faster after silent superclasses were eliminated.

  * It introduced new bugs.  For example SilentParametersOverlapping,
    T5051, and T7862, all failed to compile because of instance overlap
    directly because of the silent-superclass trick.

So this patch takes a new approach, which I worked out with Dimitrios
in the closing hours before Christmas.  It is described in detail
in THE PROBLEM in Note [Recursive superclasses] in TcInstDcls.

Seems to work great!

Quite a bit of knock-on effect

 * The main implementation work is in tcSuperClasses in TcInstDcls
   Everything else is fall-out

 * IdInfo.DFunId no longer needs its n-silent argument
   * Ditto IDFunId in IfaceSyn
   * Hence interface file format changes

 * Now that DFunIds do not have silent superclass parameters, printing
   out instance declarations is simpler. There is tiny knock-on effect
   in Haddock, so that submodule is updated

 * I realised that when computing the "size of a dictionary type"
   in TcValidity.sizePred, we should be rather conservative about
   type functions, which can arbitrarily increase the size of a type.
   Hence the new datatype TypeSize, which has a TSBig constructor for
   "arbitrarily big".

 * instDFunType moves from TcSMonad to Inst

 * Interestingly, CmmNode and CmmExpr both now need a non-silent
   (Ord r) in a couple of instance declarations. These were previously
   silent but must now be explicit.

 * Quite a bit of wibbling in error messages

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

This commit also refactors a bunch of lexeme-oriented code into
a new module Lexeme, and includes a submodule update for haddock.

Summary: Using `dropWhileEndLE` tends to be faster and easier to read
than the `reverse . dropWhile p . reverse` idiom. This also cleans up
some other, nearby, messes. Fix #9616 (incorrect number formatting
potentially leading to incorrect numbers in output).

Test Plan: Run validate

Reviewers: thomie, rwbarton, nomeata, austin

Reviewed By: nomeata, austin

Subscribers: simonmar, ezyang, carter, thomie

Projects: #ghc

Differential Revision: https://phabricator.haskell.org/D259

GHC Trac Issues: #9623, #9616

Conflicts:
	compiler/basicTypes/OccName.lhs

This reverts commit 2a885688.

Summary: Using `dropWhileEndLE` tends to be faster and easier to read
than the `reverse . dropWhile p . reverse` idiom. This also cleans up
some other, nearby, messes. Fix #9616 (incorrect number formatting
potentially leading to incorrect numbers in output).

Test Plan: Run validate

Reviewers: thomie, rwbarton, nomeata, austin

Reviewed By: nomeata, austin

Subscribers: simonmar, ezyang, carter, thomie

Projects: #ghc

Differential Revision: https://phabricator.haskell.org/D259

GHC Trac Issues: #9623, #9616

Conflicts:
	compiler/basicTypes/OccName.lhs

Removed (pprEqPred (coercionKind co)) in favor of
(pprType (coercionType co)).

Also had to make "~R#" a *symbolic* identifier and BuiltInSyntax
to squelch prefix notation and module prefixes in output. These
changes are both sensible independent of #9062.

instead of just one matching directly. This is an alternative way to fix
ticket #9177.

and the other way around. This fixes #9177.

All the initial work on this was done fy 'archblob' (fcsernik@gmail.com);
thank you!

I reviewed the patch, started some tidying, up and then ended up in a huge
swamp of changes, not all of which I can remember now.  But:

* To suppress kind arguments when we have -fno-print-explicit-kinds,
    - IfaceTyConApp argument types are in a tagged list IfaceTcArgs

* To allow overloaded types to be printed with =>, add IfaceDFunTy to IfaceType.

* When printing data/type family instances for the user, I've made them
  print out an informative RHS, which is a new feature. Thus
        ghci> info T
        data family T a
        data instance T Int = T1 Int Int
        data instance T Bool = T2

* In implementation terms, pprIfaceDecl has just one "context" argument,
  of type IfaceSyn.ShowSub, which says
       - How to print the binders of the decl
         see note [Printing IfaceDecl binders] in IfaceSyn
       - Which sub-comoponents (eg constructors) to print

* Moved FastStringEnv from RnEnv to OccName

It all took a ridiculously long time to do.  But it's done!

In some cases, the layout of the LANGUAGE/OPTIONS_GHC lines has been
reorganized, while following the convention, to

- place `{-# LANGUAGE #-}` pragmas at the top of the source file, before
  any `{-# OPTIONS_GHC #-}`-lines.

- Moreover, if the list of language extensions fit into a single
  `{-# LANGUAGE ... -#}`-line (shorter than 80 characters), keep it on one
  line. Otherwise split into `{-# LANGUAGE ... -#}`-lines for each
  individual language extension. In both cases, try to keep the
  enumeration alphabetically ordered.
  (The latter layout is preferable as it's more diff-friendly)

While at it, this also replaces obsolete `{-# OPTIONS ... #-}` pragma
occurences by `{-# OPTIONS_GHC ... #-}` pragmas.

This fixes Trac #8954.

There were actually three places where tuple occ-names
were parsed:
  - IfaceEnv.lookupOrigNameCache
  - Convert.isBuiltInOcc
  - OccName.isTupleOcc_maybe

I combined all three into TysWiredIn.isBuiltInOcc_maybe
Much nicer.

This is so that generated names like e.g. workers don't show up as
infix operators when using something like -ddump-simpl.

we want a rule "map coerce = coerce" to match the core generated for
"map Age" (this is #2110).

This patch implements Pattern Synonyms (enabled by -XPatternSynonyms),
allowing y ou to assign names to a pattern and abstract over it.

The rundown is this:

  * Named patterns are introduced by the new 'pattern' keyword, and can
    be either *unidirectional* or *bidirectional*. A unidirectional
    pattern is, in the simplest sense, simply an 'alias' for a pattern,
    where the LHS may mention variables to occur in the RHS. A
    bidirectional pattern synonym occurs when a pattern may also be used
    in expression context.

  * Unidirectional patterns are declared like thus:

        pattern P x <- x:_

    The synonym 'P' may only occur in a pattern context:

        foo :: [Int] -> Maybe Int
        foo (P x) = Just x
        foo _     = Nothing

  * Bidirectional patterns are declared like thus:

        pattern P x y = [x, y]

    Here, P may not only occur as a pattern, but also as an expression
    when given values for 'x' and 'y', i.e.

        bar :: Int -> [Int]
        bar x = P x 10

  * Patterns can't yet have their own type signatures; signatures are inferred.

  * Pattern synonyms may not be recursive, c.f. type synonyms.

  * Pattern synonyms are also exported/imported using the 'pattern'
    keyword in an import/export decl, i.e.

        module Foo (pattern Bar) where ...

    Note that pattern synonyms share the namespace of constructors, so
    this disambiguation is required as a there may also be a 'Bar'
    type in scope as well as the 'Bar' pattern.

  * The semantics of a pattern synonym differ slightly from a typical
    pattern: when using a synonym, the pattern itself is matched,
    followed by all the arguments. This means that the strictness
    differs slightly:

        pattern P x y <- [x, y]

        f (P True True) = True
        f _             = False

        g [True, True] = True
        g _            = False

    In the example, while `g (False:undefined)` evaluates to False,
    `f (False:undefined)` results in undefined as both `x` and `y`
    arguments are matched to `True`.

For more information, see the wiki:

    https://ghc.haskell.org/trac/ghc/wiki/PatternSynonyms
    https://ghc.haskell.org/trac/ghc/wiki/PatternSynonyms/ImplementationReviewed-by: Simon Peyton Jones <simonpj@microsoft.com>
Signed-off-by: Austin Seipp <austin@well-typed.com>

The build system would've complained loudly about these inclusions if it
weren't for #8527.

This major patch implements the cardinality analysis described
in our paper "Higher order cardinality analysis". It is joint
work with Ilya Sergey and Dimitrios Vytiniotis.

The basic is augment the absence-analysis part of the demand
analyser so that it can tell when something is used
	 never
	 at most once
 	 some other way

The "at most once" information is used
    a) to enable transformations, and
       in particular to identify one-shot lambdas
    b) to allow updates on thunks to be omitted.

There are two new flags, mainly there so you can do performance
comparisons:
    -fkill-absence   stops GHC doing absence analysis at all
    -fkill-one-shot  stops GHC spotting one-shot lambdas
                     and single-entry thunks

The big changes are:

* The Demand type is substantially refactored.  In particular
  the UseDmd is factored as follows
      data UseDmd
        = UCall Count UseDmd
        | UProd [MaybeUsed]
        | UHead
        | Used

      data MaybeUsed = Abs | Use Count UseDmd

      data Count = One | Many

  Notice that UCall recurses straight to UseDmd, whereas
  UProd goes via MaybeUsed.

  The "Count" embodies the "at most once" or "many" idea.

* The demand analyser itself was refactored a lot

* The previously ad-hoc stuff in the occurrence analyser for foldr and
  build goes away entirely.  Before if we had build (\cn -> ...x... )
  then the "\cn" was hackily made one-shot (by spotting 'build' as
  special.  That's essential to allow x to be inlined.  Now the
  occurrence analyser propagates info gotten from 'build's stricness
  signature (so build isn't special); and that strictness sig is
  in turn derived entirely automatically.  Much nicer!

* The ticky stuff is improved to count single-entry thunks separately.

One shortcoming is that there is no DEBUG way to spot if an
allegedly-single-entry thunk is acually entered more than once.  It
would not be hard to generate a bit of code to check for this, and it
would be reassuring.  But it's fiddly and I have not done it.

Despite all this fuss, the performance numbers are rather under-whelming.
See the paper for more discussion.

       nucleic2          -0.8%    -10.9%      0.10      0.10     +0.0%
         sphere          -0.7%     -1.5%      0.08      0.08     +0.0%
--------------------------------------------------------------------------------
            Min          -4.7%    -10.9%     -9.3%     -9.3%    -50.0%
            Max          -0.4%     +0.5%     +2.2%     +2.3%     +7.4%
 Geometric Mean          -0.8%     -0.2%     -1.3%     -1.3%     -1.8%

I don't quite know how much credence to place in the runtime changes,
but movement seems generally in the right direction.

This showed up when looking at some type error messages.
We were tidying some open types in a way that mapped two
distinct variables to the same thing.  Urk!

Mostly d -> g (matching DynFlag -> GeneralFlag).
Also renamed if* to when*, matching the Haskell if/when names