[Git][ghc/ghc][master] Pattern guards BindStmt always use multiplicity Many

[Marge Bot]

 Marge Bot pushed to branch master at Glasgow Haskell Compiler / GHC


Commits:
2707c4ea by Arnaud Spiwack at 2020-09-25T21:13:58-04:00
Pattern guards BindStmt always use multiplicity Many

Fixes #18439 .

The rhs of the pattern guard was consumed with multiplicity one, while
the pattern assumed it was Many. We use Many everywhere instead.

This is behaviour consistent with that of `case` expression. See #18738.

- - - - -


4 changed files:

- compiler/GHC/Tc/Gen/Match.hs
- + testsuite/tests/linear/should_fail/LinearPatternGuardWildcard.hs
- + testsuite/tests/linear/should_fail/LinearPatternGuardWildcard.stderr
- testsuite/tests/linear/should_fail/all.T


Changes:

=====================================
compiler/GHC/Tc/Gen/Match.hs
=====================================
@@ -397,7 +397,14 @@ tcGuardStmt _ (BodyStmt _ guard _ _) res_ty thing_inside
         ; return (BodyStmt boolTy guard' noSyntaxExpr noSyntaxExpr, thing) }
 
 tcGuardStmt ctxt (BindStmt _ pat rhs) res_ty thing_inside
-  = do  { (rhs', rhs_ty) <- tcInferRhoNC rhs
+  = do  { -- The Many on the next line and the unrestricted on the line after
+          -- are linked. These must be the same multiplicity. Consider
+          --   x <- rhs -> u
+          --
+          -- The multiplicity of x in u must be the same as the multiplicity at
+          -- which the rhs has been consumed. When solving #18738, we want these
+          -- two multiplicity to still be the same.
+          (rhs', rhs_ty) <- tcScalingUsage Many $ tcInferRhoNC rhs
                                    -- Stmt has a context already
         ; (pat', thing)  <- tcCheckPat_O (StmtCtxt ctxt) (lexprCtOrigin rhs)
                                          pat (unrestricted rhs_ty) $


=====================================
testsuite/tests/linear/should_fail/LinearPatternGuardWildcard.hs
=====================================
@@ -0,0 +1,7 @@
+{-# LANGUAGE LinearTypes #-}
+module LinearPatternGuardWildcard where
+
+-- See #18439
+
+unsafeConsume :: a #-> ()
+unsafeConsume x | _ <- x = ()


=====================================
testsuite/tests/linear/should_fail/LinearPatternGuardWildcard.stderr
=====================================
@@ -0,0 +1,5 @@
+
+LinearPatternGuardWildcard.hs:7:15: error:
+    • Couldn't match type ‘'Many’ with ‘'One’
+        arising from multiplicity of ‘x’
+    • In an equation for ‘unsafeConsume’: unsafeConsume x | _ <- x = ()


=====================================
testsuite/tests/linear/should_fail/all.T
=====================================
@@ -27,3 +27,4 @@ test('LinearPolyType', expect_broken([436, broken_multiplicity_syntax]), compile
 test('LinearBottomMult', normal, compile_fail, [''])
 test('LinearSequenceExpr', normal, compile_fail, [''])
 test('LinearIf', normal, compile_fail, [''])
+test('LinearPatternGuardWildcard', normal, compile_fail, [''])



View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/2707c4eae4cf99e6da2709e128f560d91e468357

-- 
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/2707c4eae4cf99e6da2709e128f560d91e468357
You're receiving this email because of your account on gitlab.haskell.org.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-commits/attachments/20200925/cb7bdb45/attachment-0001.html>