[PATCH] D128449: [clang] Introduce -Warray-parameter

[serge via Phabricator via cfe-commits]

serge-sans-paille added inline comments.


================
Comment at: clang/test/Sema/array-parameter.c:2
+// RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -fsyntax-only -Warray-parameter -verify %s
+
+void f0(int a[]);
----------------
aaron.ballman wrote:
> aaron.ballman wrote:
> > I'd like to see some additional test cases:
> > ```
> > void func(int *p);
> > void func(int a[2]); // Diagnose this one, I presume?
> > void func(int a[]); // But also diagnose this one as well, yes?
> > void func(int a[2]) {} // Do we then diagnose this as well, or is this matching a previous declaration and thus fine?
> > 
> > void other(int n, int m, int a[n]);
> > void other(int n, int m, int a[m]); // Hopefully we diagnose this set!
> > 
> > void another(int n, int array[n]);
> > void another(int n, int array[*]); // I *think* this should be warned, but I'm still a bit on the fence about it
> > ```
> > 
> > Also, if this is expected to work in C++ as well, we should probably have cases like:
> > ```
> > template <int N>
> > void func(int i[10]);
> > 
> > template <int N>
> > void func(int i[N]); // Should we diagnose this before instantiation or wait until we see the instantiation and only diagnose if differs?
> > 
> > static constexpr int Extent = 10;
> > void other(int i[10]);
> > void other(int i[Extent]); // Should not be diagnosed
> > ```
> Do you plan to add the C++ tests?
> template <int N>
> void func(int i[N]); // Should we diagnose this before instantiation or wait until we see the instantiation and only diagnose if differs?


GCC never warns about this. Current patch always warn and this looks fine to me.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D128449/new/

https://reviews.llvm.org/D128449