Re: [Libreoffice] [PATCH] Refactor to make cppcheck happy

Hi Justin,

On Wed, 2010-09-29 at 11:48 -0700, Justin Malcolm wrote:
> From the 'easy hacks' page it seemed that getting rid of warnings and
> flags by tools like cppcheck was a goal.

        Sure ! of course :-)

> Attached is a small patch that does nothing except eliminate one
> cppcheck warning.  It is contributed under the LGPL3+.

        What is the warning ? :-)

> As I said, it does nothing different and, depending on the compiler,
> it could even be less efficient.  It does however make cppcheck happy.
> Is this kind of thing useful?

        Certainly; clearly it'd be more re-assuring if it was compiled & ran
afterwards.

 xrealloc(void *ptr, size_t size)
 {
-    ptr = realloc(ptr, size);
+    void *newptr = realloc(ptr, size);
     
-    if ( !ptr )
+    if (!newptr)
+       ptr = newptr;
+    else
         err_quit("out of memory");
 
     return ptr;

        So for size > 0 - realloc will return a non-NULL (perhaps different)
pointer. So - if (!newptr) [ ie. if (newptr == NULL) ] will never
succeed - surely ?

        HTH,

                Michael.

-- 
 michael.meeks@novell.com  <><, Pseudo Engineer, itinerant idiot