[PATCH] D38773: [Sema] Add support for flexible array members in Obj-C.

[Volodymyr Sapsai via Phabricator via cfe-commits]

vsapsai added inline comments.


================
Comment at: clang/include/clang/Basic/DiagnosticSemaKinds.td:5226
+def err_objc_variable_sized_type_not_at_end : Error<
+  "field %0 with variable sized type %1 is not at the end of class">;
+def note_next_field_declaration : Note<
----------------
rjmccall wrote:
> "Variable sized type" is a bit too close to the C99 variably-sized array type extension.  Maybe "unbounded array type" if you're trying to cover both "int x[];" and "int x[0];"?
> 
> Well, I guess there's some precedent for using this terminology, but ugh.
I took "variable sized type" entirely from

```
def ext_variable_sized_type_in_struct : ExtWarn<
  "field %0 with variable sized type %1 not at the end of a struct or class is"
  " a GNU extension">, InGroup<GNUVariableSizedTypeNotAtEnd>;
```

I'm not covering `int x[0];`. All the changes are for `int x[];` and `struct { int x[]; }`


================
Comment at: clang/lib/Sema/SemaDecl.cpp:15055
           }
+          // If it is the last field is checked elsewhere.
         }
----------------
rjmccall wrote:
> "Whether" rather than "If", please.  You should also leave a comment about *why* we can't check this here — I assume because you also want to complain about the last explicit ivar if there are synthesized ivars?  I think we could at least still check this for `@interface` ivars.
Will change s/If/Whether/

Main reason for checking elsewhere is to check after ivars are synthesized, you are right. At some point I had this check done here but for detecting ivar-after-flexible-array on interface/extension, interface/implementation border I am relying on chained ObjCIvarDecl. But here I have `ArrayRef<Decl *> Fields` so implementation will be different. I decided that it would be cleaner to perform the check only in DiagnoseVariableSizedIvars.


https://reviews.llvm.org/D38773