Re: PATCH: Remove unnecessary zero-initializations

[Daniel Jacobowitz <drow at mvista dot com>]

On Mon, Nov 11, 2002 at 02:48:33AM -0800, Felix Lee wrote:
> Daniel Jacobowitz <drow@mvista.com>:
> > Currently, thirteen files which provide a target_ops explicitly initialize
> > members they don't support to NULL.
> 
> this is a style and readability issue.  if every set of
> initializations is complete and mentions all members, even
> when "unnecessary", then it's easier to quickly check that
> an implementation correctly matches the specification.  a
> statement like
>     foo.bar = 0;
> indicates that the programmer was aware that foo.bar exists,
> thought about it, and decided that 0 is a correct value.
> 
> if the statement is missing, you have to spend time deciding
> if the programmer omitted it accidentally or not.
> 
> a different technique is to put a note in a comment instead
> of as a statement:
>     // foo.bar can be 0
> but this is a pointless micro-optimization, it's an
> optimization a compiler could be taught how to do, and you
> lose the opportunity to have an automated tool check for you
> that initializations are complete.

Certainly it's a style issue.  However, it's an awkward style issue and
anyone implementing a target should be looking over the complete list
of methods anyway.  I stand by my patch... although perhaps not as
"obvious" after two objections.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer