[Guy Harris <guy@xxxxxxxxxxxx>]

On Oct 28, 2003, at 5:59 AM, Biot Olivier wrote:

> I figured out that there was no function yet that only frees the value 
> part
> alone (indeed fvalue_free() frees the entire structure, not only the 
> value
> contained within it), so I added this function to 
> epan/ftypes/ftypes.[ch] as
> fvalue_free_value().

Is that the right answer, or is the right answer to have "fvalue_set" 
functions for values that are allocated free up the old value if there 
is one?  You can do an "fvalue_set" on values that aren't allocated, 
such as integral values, even if an "fvalue_set" has already on that 
fvalue_t; why shouldn't that be allowed for values that are allocated?