danielmarjamaki added a comment. In https://reviews.llvm.org/D21134#508524, @jroelofs wrote:
> I think the replacement is wrong for something like: > > int *arr; int offs1, offs2; > offs1[arr + offs2] = 42; > > > which I think would give: > > int *arr; int offs1, offs2; > arr + offs2[offs1] = 42; > > > If the precedence of the "indexing" argument is less than that of the > subscript operator, then you need to insert parens: > > int *arr; int offs1, offs2; > (arr + offs2)[offs1] = 42; > I don't think so. The matcher says: hasRHS(ignoringParenImpCasts( anyOf(stringLiteral(), declRefExpr(hasType(pointsTo(qualType()))), memberExpr(hasType(pointsTo(qualType())) I think it's hard to know what the replaced code should be. In some cases it might be better with: arr[offs1 + offs2] = 42; https://reviews.llvm.org/D21134 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits