[PATCH] D22391: [Sema] Add warning for implicitly casting a null constant to a non null pointer type

[Jordan Rose via Phabricator via cfe-commits]

jordan_rose added inline comments.


================
Comment at: test/Sema/conditional-expr.c:20
   vp = 0 ? (double *)0 : (void *)0;
-  ip = 0 ? (double *)0 : (void *)0; // expected-warning {{incompatible pointer types assigning to 'int *' from 'double *'}}
+  ip = 0 ? (double *)0 : (void *)0; // expected-warning {{incompatible pointer types assigning to 'int *' from 'double * _Nullable'}}
 
----------------
ahatanak wrote:
> ahatanak wrote:
> > jordan_rose wrote:
> > > This seems like an unfortunate change to make, since most people do not bother with nullability.
> > Yes, this is unfortunate, but I'm not sure what's the right way to avoid printing nullability specifiers in the diagnostic message. Do you have any suggestions?
> It looks like I can use PrintingPolicy to print the nullability specifier only when needed.
> 
> I think it's also possible to add a flag to Expr that indicates the Expr is possibly null. For example, when an Expr is an IntegerLiteral of value 0, or a CastExpr or a ConditionalOperator has a subexpression whose flag is set. This could be a better solution than the current solution in this patch.
Whew. I hadn't had the chance to look at PrintingPolicy and was afraid you'd have to add a new flag to diagnostics or something to specify whether nullability was relevant.

An additional flag on Expr seems like overkill to me, given that `isNullPointerConstant` already exists. But I don't work in Clang these days, so maybe I'm wrong and it is something worth caching. 


Repository:
  rC Clang

https://reviews.llvm.org/D22391