Provides elm-review
rules to disallow single-pattern case expressions.
- 🔧
NoSinglePatternCase
- Reportscase...of
expressions that match only a single pattern. Such case expressions typically are either unnecessary or may be written more concisely.
module ReviewConfig exposing (config)
import NoSinglePatternCase
import Review.Rule exposing (Rule)
config : List Rule
config =
[ NoSinglePatternCase.rule
NoSinglePatternCase.fixInArgument
]
You can try the example configuration above out by running the following command:
elm-review --template SiriusStarr/elm-review-no-single-pattern-case/example/fix-in-argument --rules NoSinglePatternCase
-
2.1.2
- ⚡️ Improve performance when dealing with ignored files (~10% in one real-world test case).
- 🐛 Bump
elm-review
to v2.12.1 for upstream bugfix.
-
2.1.1
-- Bumpelm-review
to v2.11.1 and mark rule as providing fixes. -
2.1.0
- New features:
-
✨ Non-wrapped types are now ignored by default. By default, only constructors whose names are identical to their type are reported, along with types imported from dependencies (since those aren't expected to be iterated on).
-- import the constructor `OutsideConstructor` from some other package import SomeOutsidePackage exposing (OutsideType(..)) type Date = -- Constructor has same name as type Date Int type Msg = -- Constructor has different name than type ThingieClicked update1 : Date -> Int -> Int update1 date i = case date of -- THIS CASE IS ALWAYS FLAGGED Date j -> i + j update2 : Msg -> Int -> Int update2 msg i = case msg of -- THIS CASE IS NOT FLAGGED BY DEFAULT, IS FLAGGED WITH `reportAllCustomTypes` ThingieClicked -> i + 1 update3 : OutsideType -> Int -> Int update3 oType i = case oType of -- THIS CASE IS ALWAYS FLAGGED OutsideConstructor j -> i + j
To revert to version
2.0.2
behavior and before, usereportAllCustomTypes
-
✨
replaceUnusedBindings
now replaces bindings with more specific patterns. For example,type Date = Date Int type CreateThing = CreateThing foo : ( Date, CreateThing ) -> Bool foo x = case x of ( Date i, CreateThing ) -> True
will be fixed to
type Date = Date Int type CreateThing = CreateThing foo : ( Date, CreateThing ) -> Bool foo ( Date _, CreateThing ) = True
Since some IDEs do not recognize
pointless CreateNewLet = foo
as valid Elm (even though it is), you can use
replaceUnusedBindingsWithWildcard
to revert to version2.0.2
behavior and before.
-
- New features:
-
2.0.2
- New features:
-
✨ The rule is now capable of reducing the expressions and patterns of a single-pattern case to find the original binding sites to reduce expressions at. The following improvements have been added:
-
Reduce parenthesized expressions:
unpack o = case (o) of Opaque i -> i --> unpack (Opaque i) = i
-
Reduce constructed tupled expressions
add a b = case ( a, b ) of ( Opaque i1, Opaque i2 ) -> Opaque <| i1 + i2 --> add (Opaque i1) (Opaque i2) = Opaque <| i1 + i2
-
Reduce constructed record expressions
pointless a b c = case { d = a, e = b, f = c } of { d, e, f } -> e --> pointless _ e _ = e
-
Reduce unnecessary
as
patternsunpack o = case o of ((Opaque i) as a) -> i --> unpack (Opaque i) = i
-
Reduce function application/named patterns
multWrap a b = case (Tuple a b) of Tuple i1 i2 -> Opaque <| i1 * i2 --> multWrap i1 i2 = Opaque <| i1 * i2
-
-
✨ With
replaceUnusedBindings
, unit patterns are now fixed to()
instead of_
to be more explicit. (There is anelm-review
rule to enforce this in your code, should you so desire.) -
🚸 The error range reported by the rule now flags just the pattern, instead of the entire case, for less cluttered errors (especially with tooling). No more red-underlining large sections of code!
-
- Bugfixes:
-
🐛 Fixed a bug where name clashes were only checked for expressions, not patterns, when generating fixes, leading to the generation of uncompilable code. This probably rarely happened in practice, but for instance this incorrect fix used to be generated:
unpack o = let foo i = 0 in case o of Opaque i -> i --> unpack (Opaque i) = let foo i = -- THIS IS A NAME CLASH FOR `i` 0 in i
-
🐛 Fixed a bug where both subexpressions of a single-pattern case were not actually checked by this rule. This means that nested single-pattern cases are now detected properly (previously, only the outermost one would be flagged). This applies to both those nested following the pattern or within the
case...of
.
-
- Documentation changes:
- 📝 Added a recommendation to add
NoUnused.Patterns
to one's config, as the fixes from this rule can sometimes generate nestedas
patterns, which will be detected by that rule. - 🔥 Removed some references to old v1.0 configs
- 📝 Added a recommendation to add
- New features:
-
2.0.1
-- Bumpelm-syntax
to v7.2.8, fixing issued with lambda ranges. Unlikely to affect this rule in practice, however. -
2.0.0
-- Add fixes, with option to destructure in the argument or in alet
-
1.0.1
-- Fix minor documentation issue -
1.0.0
-- Initial release