Re: master b944e88 5/7: emacsclient.c: use C99 better

Re: master b944e88 5/7: emacsclient.c: use C99 better

[Robert Pluim]

eggert@cs.ucla.edu (Paul Eggert) writes:

>  
> -  size_t dlen;
> -
> -  if (!data)
> -    return;
> -
> -  dlen = strlen (data);

I convinced myself that data could never be NULL here, but perhaps a
comment to that effect? (or restore the NULL check).

Robert

Re: master b944e88 5/7: emacsclient.c: use C99 better

[Paul Eggert]

On 11/20/18 3:19 AM, Robert Pluim wrote:
>>   
>> -  size_t dlen;
>> -
>> -  if (!data)
>> -    return;
>> -
>> -  dlen = strlen (data);
> I convinced myself that data could never be NULL here, but perhaps a
> comment to that effect? (or restore the NULL check).

Any such check should be a compile-time check, by declaring the 
containing function with the ARG_NONNULL () attribute. But why stop 
there? ARG_NONNULL () could be applied to every function in emacsclient.c.

For Emacs, ARG_NONNULL () is typically overkill, as it doesn't help the 
user or the compiler enough to be worth the overhead of using it. In 
this particular case we have a static function that calls strlen on its 
argument so it is obviously expecting the argument to be a nonnull 
pointer and ARG_NONNULL () isn't needed. That being said, a comment 
might not hurt, and I'll look into doing that in my next revision (I'm 
currently looking into fixing some problems with emacsclient.c anyway....).