On 04/14/2010 10:02 AM, Jim Meyering wrote:
From: Jim Meyering <meyering@redhat.com>
Among some here, there is a strong aversion to use of "assert", yet some others think it is essential (when applied judiciously) even -- perhaps "especially" -- at the heart of libraries and core hypervisor- related code. Here is a compromise that lets us make assertions about the code (e.g., to tell static analyzers about invariants) without even a hint of risk of an abort. * src/internal.h [STATIC_ANALYSIS]: Include <assert.h>. (sa_assert): Define. A no-op most of the time, but equivalent to classical assert when STATIC_ANALYSIS is nonzero.
Personally, I like this compromise. But I'll let others who have been more vocal against assert() give the actual ACK.
--- src/internal.h | 7 +++++++ 1 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/src/internal.h b/src/internal.h index 2e73210..4be17d8 100644 --- a/src/internal.h +++ b/src/internal.h @@ -9,6 +9,13 @@ # include <limits.h> # include <verify.h>
+# if STATIC_ANALYSIS
I think we should add a line here: # undef NDEBUG
+# include <assert.h> +# define sa_assert(expr) assert (expr) +# else +# define sa_assert(expr) /* empty */ +# endif
to guarantee that the STATIC_ANALYSIS always gets the real assert(), rather than the no-op variant required by POSIX when NDEBUG is present. -- Eric Blake eblake@redhat.com +1-801-349-2682 Libvirt virtualization library http://libvirt.org