5.2. Warnings and sanity-checking

GHC has a number of options that select which types of non-fatal error messages, otherwise known as warnings, can be generated during compilation. Some options control individual warnings and others control collections of warnings. Use -W⟨wflag⟩ to turn on an individual warning or a collection, or use -Wno-⟨wflag⟩ to turn it off. Use -Werror to make all warnings into fatal errors, or -Werror=⟨wflag⟩ to make a specific warning into an error. Reverse this with -Wwarn to make all warnings non-fatal, or -Wwarn=⟨wflag⟩ to make a specific warning non-fatal.

Note

In GHC < 8 the syntax for -W⟨wflag⟩ was -fwarn-⟨wflag⟩ (e.g. -fwarn-incomplete-patterns). This spelling is deprecated, but still accepted for backwards compatibility. Likewise, -Wno-⟨wflag⟩ used to be fno-warn-⟨wflag⟩ (e.g. -fno-warn-incomplete-patterns).

5.2.1. Warning groups

The following flags are simple ways to select standard “packages” of warnings. They can be reversed using -Wno-⟨group⟩, which has the same effect as -Wno-... for every individual warning in the group.

-Wdefault
Since:8.0

By default, you get a standard set of warnings which are generally likely to indicate bugs in your program. These are:

-W

Provides the standard warnings plus

-Wextra

Alias for -W

-Wall

Turns on all warning options that indicate potentially suspicious code. They include all warnings in -Wextra, plus:

-Weverything
Since:8.0

Turns on every single warning supported by the compiler.

-Wcompat
Since:8.0

Turns on warnings that will be enabled by default in the future, but remain off in normal compilations for the time being. This allows library authors eager to make their code future compatible to adapt to new features before they even generate warnings.

This currently enables

-w

Turns off all warnings, including the standard ones and those that -Wall doesn’t enable.

-Wnot

Deprecated alias for -w

When a warning is emitted, the specific warning flag which controls it is shown, but the group can optionally be shown as well:

-fshow-warning-groups
Default:off

When showing which flag controls a warning, also show the respective warning group flag(s) that warning is contained in.

5.2.2. Treating warnings as fatal errors

These options control which warnings are considered fatal and cause compilation to abort.

-Werror
Since:6.8 (-Wwarn)

Makes any warning into a fatal error. Useful so that you don’t miss warnings when doing batch compilation. To reverse -Werror and stop treating any warnings as errors use -Wwarn, or use -Wwarn=⟨wflag⟩ to stop treating specific warnings as errors.

-Werror=⟨wflag⟩
Implies:-W⟨wflag⟩

Makes a specific warning into a fatal error. The warning will be enabled if it hasn’t been enabled yet. Can be reversed with -Wwarn=⟨wflag⟩.

-Werror=⟨group⟩ has the same effect as -Werror=... for each warning flag in the group (for example, -Werror=compat will turn every warning in the -Wcompat group into a fatal error).

-Wwarn

Warnings are treated only as warnings, not as errors. This is the default, but can be useful to negate a -Werror flag.

-Wwarn=⟨wflag⟩

Causes a specific warning to be treated as normal warning, not fatal error.

Note that it doesn’t fully negate the effects of -Werror=⟨wflag⟩ - the warning will still be enabled.

-Wwarn=⟨group⟩ has the same effect as -Wwarn=... for each warning flag in the group (for example, -Wwarn=compat will mark every warning in the -Wcompat group as non-fatal).

-Wno-error=⟨wflag⟩

Alternative spelling for -Wwarn=⟨wflag⟩.

5.2.3. Individual warning options

The full set of warning options is described below. To turn off any warning, simply give the corresponding -Wno-... option on the command line. For backwards compatibility with GHC versions prior to 8.0, all these warnings can still be controlled with -f(no-)warn-* instead of -W(no-)*.

-Wunrecognised-warning-flags
Since:8.0
Default:on

Enables warnings when the compiler encounters a -W... flag that is not recognised.

-Wcompat-unqualified-imports
Since:8.10

Warns on unqualified imports of core library modules which are subject to change in future GHC releases. Currently the following modules are covered by this warning:

  • Data.List due to the future addition of Data.List.singleton and specialisation of exports to the [] type. See the mailing list for details.

This warning can be addressed by either adding an explicit import list or using a qualified import.

-Wprepositive-qualified-module
Since:8.10

Normally, imports are qualified prepositively: import qualified M. By using ImportQualifiedPost, the qualified keyword can be used after the module name. Like so: import M qualified. This will warn when the first, prepositive syntax is used.

-Wtyped-holes
Since:7.8
Default:on

Determines whether the compiler reports typed holes warnings. Has no effect unless typed holes errors are deferred until runtime. See Typed Holes and Deferring type errors to runtime.

-Wdeferred-type-errors
Since:8.0
Default:on

Causes a warning to be reported when a type error is deferred until runtime. See Deferring type errors to runtime.

-Wdeferred-out-of-scope-variables
Since:8.0

Warn when a deferred out-of-scope variable is encountered. See Deferring type errors to runtime.

-Wpartial-type-signatures
Since:7.10
Default:on

Determines whether the compiler reports holes in partial type signatures as warnings. Has no effect unless PartialTypeSignatures is enabled, which controls whether errors should be generated for holes in types or not. See Partial Type Signatures.

-fhelpful-errors
Since:7.4
Default:on

When a name or package is not found in scope, make suggestions for the name or package you might have meant instead.

-Wunrecognised-pragmas
Since:6.10
Default:on

Causes a warning to be emitted when a pragma that GHC doesn’t recognise is used. As well as pragmas that GHC itself uses, GHC also recognises pragmas known to be used by other tools, e.g. OPTIONS_HUGS and DERIVE.

-Wmisplaced-pragmas
Since:9.4
Default:on

Warn when a pragma that should only appear in the header of a module, such as a LANGUAGE or OPTIONS_GHC pragma, appears in the body of the module instead.

-Wmissed-specialisations
Since:8.0
Default:off

Emits a warning if GHC cannot specialise an overloaded function, usually because the function needs an INLINABLE pragma. Reports when the situation arises during specialisation of an imported function.

This form is intended to catch cases where an imported function that is marked as INLINABLE (presumably to enable specialisation) cannot be specialised as it calls other functions that are themselves not specialised.

Note that this warning will not throw errors if used with -Werror.

-Wmissed-specializations

Alias for -Wmissed-specialisations

-Wall-missed-specialisations
Since:8.0
Default:off

Emits a warning if GHC cannot specialise an overloaded function, usually because the function needs an INLINABLE pragma. Reports all such situations.

Note that this warning will not throw errors if used with -Werror.

-Wall-missed-specializations

Alias for -Wall-missed-specialisations

-Wextended-warnings
Since:9.8.1
Default:on

Causes a warning to be emitted when a module, function or type with a WARNING or DEPRECATED pragma is used, regardless of the category which may be associated with the pragma. See WARNING and DEPRECATED pragmas for more details on the pragmas. This implies -Wdeprecations and all -Wx-⟨category⟩ flags.

-Wx-⟨category⟩
Since:9.8.1
Default:on

Causes a warning to be emitted when a module, function or type with a WARNING in "x-⟨category⟩" pragma is used. See WARNING and DEPRECATED pragmas for more details on the pragmas.

-Wdeprecations
Default:on

Causes a warning to be emitted when a module, function or type with DEPRECATED pragma, or a WARNING pragma with the deprecated category, is used. See WARNING and DEPRECATED pragmas for more details on the pragmas.

-Wwarnings-deprecations
Since:6.10
Default:on

Causes a warning to be emitted when a module, function or type with DEPRECATED pragma, or a WARNING pragma with the deprecated category, is used. See WARNING and DEPRECATED pragmas for more details on the pragmas. An alias for -Wdeprecations.

-Wnoncanonical-monad-instances
Since:8.0
Default:on

Warn if noncanonical Applicative or Monad instances declarations are detected.

When this warning is enabled, the following conditions are verified:

In Monad instances declarations warn if any of the following conditions does not hold:

  • If return is defined it must be canonical (i.e. return = pure).
  • If (>>) is defined it must be canonical (i.e. (>>) = (*>)).

Moreover, in Applicative instance declarations:

  • Warn if pure is defined backwards (i.e. pure = return).
  • Warn if (*>) is defined backwards (i.e. (*>) = (>>)).

This warning is part of the -Wcompat option group.

-Wnoncanonical-monadfail-instances
Since:8.0

This warning is deprecated. It no longer has any effect since GHC 8.8. It was used during the transition period of the MonadFail proposal, to detect when an instance of the Monad class was not defined via MonadFail, or when a MonadFail instance was defined backwards, using the method in Monad.

-Wnoncanonical-monoid-instances
Since:8.0
Default:on

Warn if noncanonical Semigroup or Monoid instances declarations are detected.

When this warning is enabled, the following conditions are verified:

In Monoid instances declarations warn if any of the following conditions does not hold:

  • If mappend is defined it must be canonical (i.e. mappend = (Data.Semigroup.<>)).

Moreover, in Semigroup instance declarations:

  • Warn if (<>) is defined backwards (i.e. (<>) = mappend).

This warning is part of the -Wcompat option group.

-Wmissing-monadfail-instances
Since:8.0

This warning is deprecated. It no longer has any effect since GHC 8.8. It was used during the transition period of the MonadFail proposal, to warn when a failable pattern is used in a do-block that does not have a MonadFail instance.

-Wsemigroup
Since:8.0

Warn when definitions are in conflict with the future inclusion of Semigroup into the standard typeclasses.

  1. Instances of Monoid should also be instances of Semigroup
  2. The Semigroup operator (<>) will be in Prelude, which clashes with custom local definitions of such an operator

Being part of the -Wcompat option group, this warning is off by default, but will be switched on in a future GHC release.

-Wdeprecated-flags
Since:6.10
Default:on

Causes a warning to be emitted when a deprecated command-line flag is used.

-Wunsupported-calling-conventions
Since:7.6

Causes a warning to be emitted for foreign declarations that use unsupported calling conventions. In particular, if the stdcall calling convention is used on an architecture other than i386 then it will be treated as ccall.

-Wdodgy-foreign-imports
Since:6.10

Causes a warning to be emitted for foreign imports of the following form:

foreign import "f" f :: FunPtr t

on the grounds that it probably should be

foreign import "&f" f :: FunPtr t

The first form declares that f is a (pure) C function that takes no arguments and returns a pointer to a C function with type t, whereas the second form declares that f itself is a C function with type t. The first declaration is usually a mistake, and one that is hard to debug because it results in a crash, hence this warning.

-Wdodgy-exports
Since:6.12

Causes a warning to be emitted when a datatype T is exported with all constructors, i.e. T(..), but is it just a type synonym.

Also causes a warning to be emitted when a module is re-exported, but that module exports nothing.

-Wdodgy-imports
Since:6.8

Causes a warning to be emitted in the following cases:

  • When a datatype T is imported with all constructors, i.e. T(..), but has been exported abstractly, i.e. T.
  • When an import statement hides an entity that is not exported.
-Woverflowed-literals
Since:7.8

Causes a warning to be emitted if a literal will overflow, e.g. 300 :: Word8.

-Wempty-enumerations
Since:7.8

Causes a warning to be emitted if an enumeration is empty, e.g. [5 .. 3].

-Wderiving-defaults
Since:8.10

Causes a warning when both DeriveAnyClass and GeneralizedNewtypeDeriving are enabled and no explicit deriving strategy is in use. For example, this would result a warning:

class C a
newtype T a = MkT a deriving C
-Wduplicate-constraints
Since:7.8

Have the compiler warn about duplicate constraints in a type signature. For example

f :: (Eq a, Show a, Eq a) => a -> a

The warning will indicate the duplicated Eq a constraint.

This option is now deprecated in favour of -Wredundant-constraints.

-Wredundant-constraints
Since:8.0

Have the compiler warn about redundant constraints in a type signature. In particular:

  • A redundant constraint within the type signature itself:

    f :: (Eq a, Ord a) => a -> a
    

    The warning will indicate the redundant Eq a constraint: it is subsumed by the Ord a constraint.

  • A constraint in the type signature is not used in the code it covers:

    f :: Eq a => a -> a -> Bool
    f x y = True
    

    The warning will indicate the redundant Eq a constraint: : it is not used by the definition of f.)

Similar warnings are given for a redundant constraint in an instance declaration.

When turning on, you can suppress it on a per-module basis with -Wno-redundant-constraints. Occasionally you may specifically want a function to have a more constrained signature than necessary, perhaps to leave yourself wiggle-room for changing the implementation without changing the API. In that case, you can suppress the warning on a per-function basis, using a call in a dead binding. For example:

f :: Eq a => a -> a -> Bool
f x y = True
where
    _ = x == x  -- Suppress the redundant-constraint warning for (Eq a)

Here the call to (==) makes GHC think that the (Eq a) constraint is needed, so no warning is issued.

-Wduplicate-exports
Since:at least 5.04
Default:on

Have the compiler warn about duplicate entries in export lists. This is useful information if you maintain large export lists, and want to avoid the continued export of a definition after you’ve deleted (one) mention of it in the export list.

-Whi-shadowing
Since:at least 5.04, deprecated

Causes the compiler to emit a warning when a module or interface file in the current directory is shadowing one with the same module name in a library or other directory.

This flag was not implemented correctly and is now deprecated. It will be removed in a later version of GHC.

-Widentities
Since:7.2

Causes the compiler to emit a warning when a Prelude numeric conversion converts a type T to the same type T; such calls are probably no-ops and can be omitted. The functions checked for are: toInteger, toRational, fromIntegral, and realToFrac.

-Wimplicit-kind-vars
Since:8.6

This warning is deprecated. It no longer has any effect since GHC 8.10. It was used to detect if a kind variable is not explicitly quantified over. For instance, the following would produce a warning:

f :: forall (a :: k). Proxy a

This is now an error and can be fixed by explicitly quantifying over k:

f :: forall k (a :: k). Proxy a

or

f :: forall {k} (a :: k). Proxy a
-Wimplicit-lift
Since:9.2

Template Haskell quotes referring to local variables bound outside of the quote are implicitly converted to use lift. For example, f x = [| reverse x |] becomes f x = [| reverse $(lift x) |]). This flag issues a warning for every such implicit addition of lift. This can be useful when debugging more complex staged programs, where an implicit lift can accidentally conceal a variable used at a wrong stage.

-Wimplicit-prelude
Since:6.8
Default:off

Have the compiler warn if the Prelude is implicitly imported. This happens unless either the Prelude module is explicitly imported with an import ... Prelude ... line, or this implicit import is disabled (either by NoImplicitPrelude or a LANGUAGE NoImplicitPrelude pragma).

Note that no warning is given for syntax that implicitly refers to the Prelude, even if NoImplicitPrelude would change whether it refers to the Prelude. For example, no warning is given when 368 means Prelude.fromInteger (368::Prelude.Integer) (where Prelude refers to the actual Prelude module, regardless of the imports of the module being compiled).

-Wincomplete-patterns
Since:5.04

The option -Wincomplete-patterns warns about places where a pattern-match might fail at runtime. The function g below will fail when applied to non-empty lists, so the compiler will emit a warning about this when -Wincomplete-patterns is enabled.

g [] = 2

This option isn’t enabled by default because it can be a bit noisy, and it doesn’t always indicate a bug in the program. However, it’s generally considered good practice to cover all the cases in your functions, and it is switched on by -W.

-Wincomplete-uni-patterns
Since:7.2

The flag -Wincomplete-uni-patterns is similar to -Wincomplete-patterns, except that it applies only to lambda-expressions and pattern bindings, constructs that only allow a single pattern:

h = \[] -> 2
Just k = f y

Furthermore, this flag also applies to lazy patterns, since they are syntactic sugar for pattern bindings. For example, f ~(Just x) = (x,x) is equivalent to f y = let Just x = y in (x,x).

-fmax-pmcheck-models=⟨n⟩
Since:8.10
Default:30

The pattern match checker works by assigning symbolic values to each pattern. We call each such assignment a ‘model’. Now, each pattern match clause leads to potentially multiple splits of that model, encoding different ways for the pattern match to fail. For example, when matching x against Just 4, we split each incoming matching model into two uncovered sub-models: One where x is Nothing and one where x is Just y but y is not 4.

This can be exponential in the arity of the pattern and in the number of guards in some cases. The -fmax-pmcheck-models=⟨n⟩ limit makes sure we scale polynomially in the number of patterns, by forgetting refined information gained from a partially successful match. For the above example, if we had a limit of 1, we would continue checking the next clause with the original, unrefined model.

-Wincomplete-record-updates
Since:6.4

The function f below will fail when applied to Bar, so the compiler will emit a warning about this when -Wincomplete-record-updates is enabled.

data Foo = Foo { x :: Int }
         | Bar

f :: Foo -> Foo
f foo = foo { x = 6 }

This option isn’t enabled by default because it can be very noisy, and it often doesn’t indicate a bug in the program.

-Wmissing-deriving-strategies
Since:8.8.1
Default:off

The datatype below derives the Eq typeclass, but doesn’t specify a strategy. When -Wmissing-deriving-strategies is enabled, the compiler will emit a warning about this.

data Foo a = Foo a
  deriving (Eq)

The compiler will warn here that the deriving clause doesn’t specify a strategy. If the warning is enabled, but DerivingStrategies is not enabled, the compiler will suggest turning on the DerivingStrategies extension.

-Wmissing-fields
Since:at least 5.04

This option is on by default, and warns you whenever the construction of a labelled field constructor isn’t complete, missing initialisers for one or more fields. While not an error (the missing fields are initialised with bottoms), it is often an indication of a programmer error.

-Wmissing-export-lists
Since:8.4

This flag warns if you declare a module without declaring an explicit export list. For example

module M where

  p x = x

The -Wmissing-export-lists flag will warn that M does not declare an export list. Declaring an explicit export list for M enables GHC dead code analysis, prevents accidental export of names and can ease optimizations like inlining.

-Wmissing-import-lists
Since:7.0

This flag warns if you use an unqualified import declaration that does not explicitly list the entities brought into scope. For example

module M where
  import X( f )
  import Y
  import qualified Z
  p x = f x x

The -Wmissing-import-lists flag will warn about the import of Y but not X If module Y is later changed to export (say) f, then the reference to f in M will become ambiguous. No warning is produced for the import of Z because extending Z’s exports would be unlikely to produce ambiguity in M.

-Wmissing-methods
Since:at least 5.04
Default:on

This option warns you whenever an instance declaration is missing one or more methods, and the corresponding class declaration has no default declaration for them.

The MINIMAL pragma can be used to change which combination of methods will be required for instances of a particular class. See MINIMAL pragma.

-Wmissing-signatures
Since:at least 5.04
Default:off

If you would like GHC to check that every top-level function/value has a type signature, use the -Wmissing-signatures option. As part of the warning GHC also reports the inferred type.

-Wmissing-exported-sigs
Since:7.10

This option is now deprecated in favour of -Wmissing-exported-signatures.

-Wmissing-exported-signatures
Since:8.0
Default:off

If you would like GHC to check that every exported top-level function/value has a type signature, but not check unexported values, use the -Wmissing-exported-signatures option. If this option is used in conjunction with -Wmissing-signatures then every top-level function/value must have a type signature. As part of the warning GHC also reports the inferred type.

-Wmissing-local-sigs
Since:7.0

This option is now deprecated in favour of -Wmissing-local-signatures.

-Wmissing-local-signatures
Since:8.0

If you use the -Wmissing-local-signatures flag GHC will warn you about any polymorphic local bindings. As part of the warning GHC also reports the inferred type. The option is off by default.

-Wmissing-pattern-synonym-signatures
Since:8.0
Default:off

If you would like GHC to check that every pattern synonym has a type signature, use the -Wmissing-pattern-synonym-signatures option. If this option is used in conjunction with -Wmissing-exported-signatures then only exported pattern synonyms must have a type signature. GHC also reports the inferred type.

-Wmissing-kind-signatures
Since:9.2
Default:off

If you would like GHC to check that every data, type family, type-class definition has a standalone kind signature or a CUSK, use the -Wmissing-kind-signatures option. You can specify the kind via StandaloneKindSignatures or CUSKs.

Note that -Wmissing-kind-signatures does not warn about associated type families, as GHC considers an associated type family declaration to have a CUSK if its enclosing class has a CUSK. (See Complete user-supplied kind signatures and polymorphic recursion for more on this point.) Therefore, giving the parent class a standalone kind signature or CUSK is sufficient to fix the warning for the class’s associated type families as well.

-Wmissing-poly-kind-signatures
Since:9.8
Default:off

This is a restricted version of -Wmissing-kind-signatures.

It warns when a declaration defines a type constructor that lacks a standalone kind signature and whose inferred kind is polymorphic (which happens with -PolyKinds. For example

data T a = MkT (a -> Int)    -- T :: Type -> Type
                             -- Not polymorphic, hence no warning
data W f a = MkW (f a)       -- W :: forall k. (k->Type) -> k -> Type
                             -- Polymorphic, hence warning!

It is useful to catch accidentally polykinded types, or to make that polymorphism explicit, without requiring a kind signature for every type.

-Wmissing-exported-pattern-synonym-signatures
Default:off

If you would like GHC to check that every exported pattern synonym has a type signature, but not check unexported pattern synonyms, use the -Wmissing-exported-pattern-synonym-signatures option. If this option is used in conjunction with -Wmissing-pattern-synonym-signatures then every pattern synonym must have a type signature. As part of the warning GHC also reports the inferred type.

-Wname-shadowing
Since:at least 5.04

This option causes a warning to be emitted whenever an inner-scope value has the same name as an outer-scope value, i.e. the inner value shadows the outer one. This can catch typographical errors that turn into hard-to-find bugs, e.g., in the inadvertent capture of what would be a recursive call in f = ... let f = id in ... f ....

The warning is suppressed for names beginning with an underscore. For example

f x = do { _ignore <- this; _ignore <- that; return (the other) }
-Worphans
Since:6.4

These flags cause a warning to be emitted whenever the module contains an “orphan” instance declaration or rewrite rule. An instance declaration is an orphan if it appears in a module in which neither the class nor the type being instanced are declared in the same module. A rule is an orphan if it is a rule for a function declared in another module. A module containing any orphans is called an orphan module.

The trouble with orphans is that GHC must pro-actively read the interface files for all orphan modules, just in case their instances or rules play a role, whether or not the module’s interface would otherwise be of any use. See Orphan modules and instance declarations for details.

The flag -Worphans warns about user-written orphan rules or instances.

-Woverlapping-patterns
Since:at least 5.04

By default, the compiler will warn you if a set of patterns are overlapping, e.g.,

f :: String -> Int
f []     = 0
f (_:xs) = 1
f "2"    = 2

where the last pattern match in f won’t ever be reached, as the second pattern overlaps it. More often than not, redundant patterns is a programmer mistake/error, so this option is enabled by default.

If the programmer is dead set on keeping a redundant clause, for example to prevent bitrot, they can make use of a guard scrutinising GHC.Exts.considerAccessible to prevent the checker from flagging the parent clause as redundant:

g :: String -> Int
g []                       = 0
g (_:xs)                   = 1
g "2" | considerAccessible = 2 -- No warning!

Note that considerAccessible should come as the last statement of the guard in order not to impact the results of the checker. E.g., if you write

h :: Bool -> Int
h x = case (x, x) of
  (True,  True)  -> 1
  (False, False) -> 2
  (True,  False) | considerAccessible, False <- x -> 3

The pattern-match checker takes you by your word, will conclude that False <- x might fail and warn that the pattern-match is inexhaustive. Put considerAccessible last to avoid such confusions.

Note that due to technical limitations, considerAccessible will not suppress -Winaccessible-code warnings.

-Winaccessible-code
Since:8.6

By default, the compiler will warn you if types make a branch inaccessible. This generally requires GADTs or similar extensions.

Take, for example, the following program

{-# LANGUAGE GADTs #-}

data Foo a where
 Foo1 :: Foo Char
 Foo2 :: Foo Int

data TyEquality a b where
        Refl :: TyEquality a a

checkTEQ :: Foo t -> Foo u -> Maybe (TyEquality t u)
checkTEQ x y = error "unimportant"

step2 :: Bool
step2 = case checkTEQ Foo1 Foo2 of
         Just Refl -> True -- Inaccessible code
         Nothing -> False

The Just Refl case in step2 is inaccessible, because in order for checkTEQ to be able to produce a Just, t ~ u must hold, but since we’re passing Foo1 and Foo2 here, it follows that t ~ Char, and u ~ Int, and thus t ~ u cannot hold.

-Wstar-is-type
Since:8.6

The use of * to denote the kind of inhabited types relies on the StarIsType extension, which in a future release will be turned off by default and then possibly removed. The reasons for this and the deprecation schedule are described in GHC proposal #143.

This warning allows to detect such uses of * before the actual breaking change takes place. The recommended fix is to replace * with Type imported from Data.Kind.

-Wstar-binder
Since:8.6

Under StarIsType, a * in types is not an operator nor even a name, it is special syntax that stands for Data.Kind.Type. This means that an expression like Either * Char is parsed as Either (*) Char and not (*) Either Char.

In binding positions, we have similar parsing rules. Consider the following example

{-# LANGUAGE TypeOperators, TypeFamilies, StarIsType #-}

type family a + b
type family a * b

While a + b is parsed as (+) a b and becomes a binding position for the (+) type operator, a * b is parsed as a (*) b and is rejected.

As a workaround, we allow to bind (*) in prefix form:

type family (*) a b

This is a rather fragile arrangement, as generally a programmer expects (*) a b to be equivalent to a * b. With -Wstar-binder we warn when this special treatment of (*) takes place.

-Wsimplifiable-class-constraints
Since:8.2
Default:on

Warn about class constraints in a type signature that can be simplified using a top-level instance declaration. For example:

f :: Eq [a] => a -> a

Here the Eq [a] in the signature overlaps with the top-level instance for Eq [a]. GHC goes to some efforts to use the former, but if it should use the latter, it would then have an insoluble Eq a constraint. Best avoided by instead writing:

f :: Eq a => a -> a
-Wtabs
Since:6.8

Have the compiler warn if there are tabs in your source file.

-Wtype-defaults
Since:at least 5.04
Default:off

Have the compiler warn/inform you where in your source the Haskell defaulting mechanism for numeric types kicks in. This is useful information when converting code from a context that assumed one default into one with another, e.g., the ‘default default’ for Haskell 1.4 caused the otherwise unconstrained value 1 to be given the type Int, whereas Haskell 98 and later defaults it to Integer. This may lead to differences in performance and behaviour, hence the usefulness of being non-silent about this.

-Wmonomorphism-restriction
Since:6.8
Default:off

Have the compiler warn/inform you where in your source the Haskell Monomorphism Restriction is applied. If applied silently the MR can give rise to unexpected behaviour, so it can be helpful to have an explicit warning that it is being applied.

-Wunsupported-llvm-version
Since:7.8

Warn when using -fllvm with an unsupported version of LLVM.

-Wmissed-extra-shared-lib
Since:8.8

Warn when GHCi can’t load a shared lib it deduced it should load when loading a package and analyzing the extra-libraries stanza of the target package description.

-Wunticked-promoted-constructors
Since:7.10

Warn if a promoted data constructor is used without a tick preceding its name.

For example:

data Nat = Succ Nat | Zero

data Vec n s where
  Nil  :: Vec Zero a
  Cons :: a -> Vec n a -> Vec (Succ n) a

Will raise two warnings because Zero and Succ are not written as 'Zero and 'Succ.

This also applies to list literals since 9.4. For example:

type L = [Int, Char, Bool]

will raise a warning, because [Int, Char, Bool] is a promoted list which lacks a tick.

-Wunused-binds
Since:at least 5.04

Report any function definitions (and local bindings) which are unused. An alias for

-Wunused-top-binds
Since:8.0

Report any function definitions which are unused.

More precisely, warn if a binding brings into scope a variable that is not used, except if the variable’s name starts with an underscore. The “starts-with-underscore” condition provides a way to selectively disable the warning.

A variable is regarded as “used” if

  • It is exported, or
  • It appears in the right hand side of a binding that binds at least one used variable that is used

For example:

module A (f) where
f = let (p,q) = rhs1 in t p  -- No warning: q is unused, but is locally bound
t = rhs3                     -- No warning: f is used, and hence so is t
g = h x                      -- Warning: g unused
h = rhs2                     -- Warning: h is only used in the
                             -- right-hand side of another unused binding
_w = True                    -- No warning: _w starts with an underscore
-Wunused-local-binds
Since:8.0

Report any local definitions which are unused. For example:

module A (f) where
f = let (p,q) = rhs1 in t p  -- Warning: q is unused
g = h x                      -- No warning: g is unused, but is a top-level binding
-Wunused-pattern-binds
Since:8.0

Warn if a pattern binding binds no variables at all, unless it is a lone wild-card pattern, or a banged pattern. For example:

Just _ = rhs3    -- Warning: unused pattern binding
(_, _) = rhs4    -- Warning: unused pattern binding
_  = rhs3        -- No warning: lone wild-card pattern
!() = rhs4       -- No warning: banged pattern; behaves like seq

In general a lazy pattern binding p = e is a no-op if p does not bind any variables. The motivation for allowing lone wild-card patterns is they are not very different from _v = rhs3, which elicits no warning; and they can be useful to add a type constraint, e.g. _ = x::Int. A banged pattern (see Bang patterns and Strict Haskell) is not a no-op, because it forces evaluation, and is useful as an alternative to seq.

-Wunused-imports
Since:at least 5.04

Report any modules that are explicitly imported but never used. However, the form import M() is never reported as an unused import, because it is a useful idiom for importing instance declarations, which are anonymous in Haskell.

-Wunused-matches
Since:at least 5.04

Report all unused variables which arise from term-level pattern matches, including patterns consisting of a single variable. For instance f x y = [] would report x and y as unused. The warning is suppressed if the variable name begins with an underscore, thus:

f _x = True

Note that -Wunused-matches does not warn about variables which arise from type-level patterns, as found in type family and data family instances. This must be enabled separately through the -Wunused-type-patterns flag.

-Wunused-do-bind
Since:6.12

Report expressions occurring in do and mdo blocks that appear to silently throw information away. For instance do { mapM popInt xs ; return 10 } would report the first statement in the do block as suspicious, as it has the type StackM [Int] and not StackM (), but that [Int] value is not bound to anything. The warning is suppressed by explicitly mentioning in the source code that your program is throwing something away:

do { _ <- mapM popInt xs ; return 10 }

Of course, in this particular situation you can do even better:

do { mapM_ popInt xs ; return 10 }
-Wunused-type-patterns
Since:8.0

Report all unused implicitly bound type variables which arise from patterns in type family and data family instances. For instance:

type instance F x y = []

would report x and y as unused on the right hand side. The warning is suppressed if the type variable name begins with an underscore, like so:

type instance F _x _y = []

When ExplicitForAll is enabled, explicitly quantified type variables may also be identified as unused. For instance:

type instance forall x y. F x y = []

would still report x and y as unused on the right hand side

Unlike -Wunused-matches, -Wunused-type-patterns is not implied by -Wall. The rationale for this decision is that unlike term-level pattern names, type names are often chosen expressly for documentation purposes, so using underscores in type names can make the documentation harder to read.

-Wunused-foralls
Since:8.0

Report all unused type variables which arise from explicit, user-written forall statements. For instance:

g :: forall a b c. (b -> b)

would report a and c as unused.

-Wunused-record-wildcards
Since:8.10

Report all record wildcards where none of the variables bound implicitly are used. For instance:

data P = P { x :: Int, y :: Int }

f1 :: P -> Int
f1 P{..} = 1 + 3

would report that the P{..} match is unused.

-Wredundant-bang-patterns
Since:9.2

Report dead bang patterns, where dead bangs are bang patterns that under no circumstances can force a thunk that wasn’t already forced. Dead bangs are a form of redundant bangs. The new check is performed in pattern-match coverage checker along with other checks (namely, redundant and inaccessible RHSs). Given

f :: Bool -> Int
f True = 1
f !x   = 2

The bang pattern on !x is dead. By the time the x in the second equation is reached, x will already have been forced due to the first equation (f True = 1). Moreover, there is no way to reach the second equation without going through the first one.

Note that -Wredundant-bang-patterns will not warn about dead bangs that appear on a redundant clause. That is because in that case, it is recommended to delete the clause wholly, including its leading pattern match.

Dead bang patterns are redundant. But there are bang patterns which are redundant that aren’t dead, for example:

f !() = 0

the bang still forces the argument, before we attempt to match on (). But it is redundant with the forcing done by the () match. Currently such redundant bangs are not considered dead, and -Wredundant-bang-patterns will not warn about them.

-Wredundant-record-wildcards
Since:8.10

Report all record wildcards where the wild card match binds no patterns. For instance:

data P = P { x :: Int, y :: Int }

f1 :: P -> Int
f1 P{x,y,..} = x + y

would report that the P{x, y, ..} match has a redundant use of ...

-Wredundant-strictness-flags
Since:9.4

Report strictness flags applied to unlifted types. An unlifted type is always strict, and applying a strictness flag has no effect.

For example:

data T = T !Int#
-Wwrong-do-bind
Since:6.12

Report expressions occurring in do and mdo blocks that appear to lack a binding. For instance do { return (popInt 10) ; return 10 } would report the first statement in the do block as suspicious, as it has the type StackM (StackM Int) (which consists of two nested applications of the same monad constructor), but which is not then “unpacked” by binding the result. The warning is suppressed by explicitly mentioning in the source code that your program is throwing something away:

do { _ <- return (popInt 10) ; return 10 }

For almost all sensible programs this will indicate a bug, and you probably intended to write:

do { popInt 10 ; return 10 }
-Winline-rule-shadowing
Since:7.8

Warn if a rewrite RULE might fail to fire because the function might be inlined before the rule has a chance to fire. See How rules interact with INLINE/NOINLINE pragmas.

-Wcpp-undef
Since:8.2

This flag passes -Wundef to the C pre-processor (if its being used) which causes the pre-processor to warn on uses of the #if directive on undefined identifiers.

-Wunbanged-strict-patterns
Since:8.2

This flag warns whenever you write a pattern that binds a variable whose type is unlifted, and yet the pattern is not a bang pattern nor a bare variable. See Unboxed types for information about unlifted types.

-Wmissing-home-modules
Since:8.2

When a module provided by the package currently being compiled (i.e. the “home” package) is imported, but not explicitly listed in command line as a target. Useful for Cabal to ensure GHC won’t pick up modules, not listed neither in exposed-modules, nor in other-modules.

-Wpartial-fields
Since:8.4

The option -Wpartial-fields warns about a record field f that is defined in some, but not all, the contructors of a data type, because f’s record selector function may fail. For exampe, the record selector function f, defined in the Foo constructor record below, will fail when applied to Bar, so the compiler will emit a warning at its definition when -Wpartial-fields is enabled.

The warning is suppressed if the field name begins with an underscore.

data Foo = Foo { f :: Int } | Bar
-Wunused-packages
Since:8.10

The option -Wunused-packages warns about packages, specified on command line via -package ⟨pkg⟩ or -package-id ⟨unit-id⟩, but were not needed during compilation. If the warning fires it means the specified package wasn’t needed for compilation.

This warning interacts poorly with GHCi because most invocations will pass a large number of -package arguments on the initial load. Therefore if you modify the targets using :load or :cd then the warning will be silently disabled if it’s enabled (see #21110).

-Winvalid-haddock
Since:9.0

When the -haddock option is enabled, GHC collects documentation comments and associates them with declarations, function arguments, data constructors, and other syntactic elements. Documentation comments in invalid positions are discarded:

myValue =
  -- | Invalid (discarded) comment in an expression
  2 + 2

This warning informs you about discarded documentation comments. It has no effect when -haddock is disabled.

-Woperator-whitespace-ext-conflict
Since:9.2

When TemplateHaskell is enabled, f $x is parsed as f applied to an untyped splice. But when the extension is disabled, the expression is parsed as a use of the $ infix operator.

To make it easy to read f $x without checking the enabled extensions, one could rewrite it as f $ x, which is what this warning suggests.

Currently, it detects the following cases:

It only covers extensions that currently exist. If you want to enforce a stricter policy and always require whitespace around all infix operators, use -Woperator-whitespace.

-Woperator-whitespace
Since:9.2

There are four types of infix operator occurrences, as defined by GHC Proposal #229:

a ! b   -- a loose infix occurrence
a!b     -- a tight infix occurrence
a !b    -- a prefix occurrence
a! b    -- a suffix occurrence

A loose infix occurrence of any operator is always parsed as an infix operator, but other occurrence types may be assigned a special meaning. For example, a prefix ! denotes a bang pattern, and a prefix $ denotes a TemplateHaskell splice.

This warning encourages the use of loose infix occurrences of all infix operators, to prevent possible conflicts with future language extensions.

-Wauto-orphans
Since:7.4

Does nothing.

-Wmissing-space-after-bang
Since:8.8

Does nothing.

-Wderiving-typeable
Since:7.10

This flag warns when Typeable is listed in a deriving clause or derived with StandaloneDeriving.

Since GHC 7.10, Typeable is automatically derived for all types. Thus, deriving Typeable yourself is redundant.

-Wambiguous-fields
Since:9.2

When DuplicateRecordFields is enabled, the option -Wambiguous-fields warns about occurrences of fields in selectors or updates that depend on the deprecated mechanism for type-directed disambiguation. This mechanism will be removed in a future GHC release, at which point these occurrences will be rejected as ambiguous. See the proposal DuplicateRecordFields without ambiguous field access and the documentation on DuplicateRecordFields for further details.

This warning has no effect when DuplicateRecordFields is disabled.

-Wforall-identifier
Since:9.4

In a future GHC release, forall will become a keyword regardless of enabled extensions. This will make definitions such as the following illegal:

-- from constraints-0.13
forall :: forall p. (forall a. Dict (p a)) -> Dict (Forall p)
forall d = ...

Library authors are advised to use a different identifier, such as forAll, forall_, or for_all:

forall_ :: forall p. (forall a. Dict (p a)) -> Dict (Forall p)
forall_ d = ...

The warning is only triggered at definition sites where it can be addressed by using a different name.

Users of a library that exports forall as an identifier cannot address the issue themselves, so the warning is not reported at use sites.

-Wunicode-bidirectional-format-characters
Since:9.0.2

Explicit unicode bidirectional formatting characters can cause source code to be rendered misleadingly in many viewers. We warn if any such character is present in the source.

Specifically, the characters disallowed by this warning are those which are a part of the ‘Explicit Formatting` category of the Unicode Bidirectional Character Type Listing

-Wgadt-mono-local-binds
Since:9.4.1

This warning is triggered on pattern matching involving GADTs, if MonoLocalBinds is disabled. Type inference can be fragile in this case.

See the OutsideIn(X) paper (section 4.2) and Let-generalisation for more details.

To resolve this warning, you can enable MonoLocalBinds or an extension implying it (GADTs or TypeFamilies).

The warning is also triggered when matching on GADT-like pattern synonyms (i.e. pattern synonyms containing equalities in provided constraints).

In previous versions of GHC (9.2 and below), it was an error to pattern match on a GADT if neither GADTs nor TypeFamilies were enabled.

-Wtype-equality-out-of-scope
Since:9.4.1

In accordance with GHC Proposal #371, the type equality syntax a ~ b is no longer built-in. Instead, ~ is a regular type operator that can be imported from Data.Type.Equality or Prelude.

To minimize breakage, a compatibility fallback is provided: whenever ~ is used but is not in scope, the compiler assumes that it stands for a type equality constraint. The warning is triggered by any code that relies on this fallback. It can be addressed by bringing ~ into scope explicitly.

The likely culprit is that you use NoImplicitPrelude and a custom Prelude. In this case, consider updating your custom Prelude to re-export ~ from Data.Type.Equality.

Being part of the -Wcompat option group, this warning is off by default, but will be switched on in a future GHC release.

-Wtype-equality-requires-operators
Since:9.4.1

In accordance with GHC Proposal #371, the type equality syntax a ~ b is no longer built-in. Instead, ~ is a regular type operator that requires the TypeOperators extension.

To minimize breakage, ~ specifically (unlike other type operators) can be used even when TypeOperators is disabled. The warning is triggered whenever this happens, and can be addressed by enabling the extension.

-Wloopy-superclass-solve
Since:9.6.1

As explained in Undecidable instances and loopy superclasses, when using UndecidableInstances it is possible for GHC to construct non-terminating evidence for certain superclass constraints.

This behaviour is scheduled to be removed in a future GHC version. In the meantime, GHC emits this warning to inform users of potential non-termination. Users can manually add the required constraint to the context to avoid the problem (thus silencing the warning).

-Wterm-variable-capture
Since:9.8.1

In accordance with GHC Proposal #281, a new extension RequiredTypeArguments will be introduced in a future GHC release.

Under RequiredTypeArguments, implicit quantification of type variables does not take place if there is a term variable of the same name in scope.

For example:

a = 15
f :: a -> a    -- Does ‘a’ refer to the term-level binding
               -- or is it implicitly quantified?

When -Wterm-variable-capture is enabled, GHC warns against implicit quantification that would stop working under RequiredTypeArguments.

-Wmissing-role-annotations
Since:9.8
Default:off

If you would like GHC to check that every data type definition has a role annotation, use the -Wmissing-role-annotations option. You can specify the role via RoleAnnotations.

GHC will not warn about type class definitions with missing role annotations, as their default roles are the strictest: all nominal. In other words the type-class role cannot be accidentally left representational or phantom, which could affected the code correctness.

-Wimplicit-rhs-quantification
Since:9.8
Default:off

In accordance with GHC Proposal #425, GHC will stop implicitly quantifying over type variables that occur free on the right-hand side of a type synonym but are not mentioned on the left-hand side. Type synonym declarations that rely on this form of quantification should be rewritten with invisible binders.

For example:

type T1 :: forall a . Maybe a
type T1    = 'Nothing :: Maybe a      -- old
type T1 @a = 'Nothing :: Maybe a      -- new

This warning detects code that will be affected by this breaking change.

If you’re feeling really paranoid, the -dcore-lint option is a good choice. It turns on heavyweight intra-pass sanity-checking within GHC. (It checks GHC’s sanity, not yours.)

-Wincomplete-export-warnings
Since:9.8.1

Ino accordance with GHC Proposal #134, it is now possible to deprecate certain exports of a name without deprecating the name itself.

As explained in WARNING and DEPRECATED pragmas, when a name is exported in several ways in the same module, but only some of those ways have a warning, it will not end up deprecated when imported in another module.

For example:

   module A (x) where

   x :: Int
   x = 2

   module M (
       {-# WARNING x "deprecated" #-} x
       module A
     )
   import A

When :ghc-flag:`-Wincomplete-export-warnings` is enabled, GHC warns about exports
that are not deprecating a name that is deprecated with another export in that module.