b091dea · elk · git.iko.soy

git.iko.soy/elk

Fix #1120, stop trying to guess how exactly the pattern came to be unhanded

Probably the case where someone adds a new tag is going to be less common than forgetting one in the first place. The new message covers both scenarios well, but without ever getting a “did you read my mind?” response. Which is a shame, but also net better.

commitb091dea75ab3a156d3f845f6f9892cbcdc326a3f authorEvan Czaplicki <info@elm-lang.org> dateTue 27 Oct 2015, 21:34 (11 years ago) parent0206049

1 file changed+9−5

compare with master

unifiedsplit

src/Reporting/Error/Pattern.hs+9−5

src/Reporting/Error/Pattern.hs+9−5

@@ -47,7 +47,7 @@ toReport _localizer err =

4747 "MISSING PATTERNS"

4848 Nothing

4949 "This `case` does not have branches for all possibilities."

50− (text (unhandledError unhandled newTagsMessage))

50+ (text (unhandledError unhandled (missingBranchesMessage (length unhandled))))

5151

5252 Redundant ->

5353 Report.report

@@ -74,7 +74,7 @@ unhandledError unhandledPatterns relevantMessage =

7474 ++ "\n"

7575 ++ "If you are seeing this error for the first time, check out these hints:\n"

7676 ++ Help.hintLink "missing-patterns" ++ "\n"

77− ++ "The recommendations about `Debug.crash` and wildcard patterns are important!"

77+ ++ "The recommendations about wildcard patterns and `Debug.crash` are important!"

7878

7979

8080toCaseMessage :: String

@@ -82,7 +82,11 @@ toCaseMessage =

8282 "Switch to a `case` expression to handle all possible patterns."

8383

8484

85−newTagsMessage :: String

86−newTagsMessage =

87− "Did you add a new tag to a union type? Add a branch for it here!"

85+missingBranchesMessage :: Int -> String

86+missingBranchesMessage numUnhandled =

87+ if numUnhandled == 1 then

88+ "Add a branch to cover this pattern!"

89+

90+ else

91+ "Add branches to cover each of these patterns!"

8892