Re: [PATCH] checkpatch: prefer = {} initializations to = {0}

[Julia Lawall <julia.lawall@inria.fr>]

On Thu, 5 Aug 2021, Joe Perches wrote:

> On Thu, 2021-08-05 at 20:17 +0200, Julia Lawall wrote:
> > On Thu, 5 Aug 2021, Joe Perches wrote:
> > > On Thu, 2021-08-05 at 05:27 -0700, Joe Perches wrote:
> > > > On Thu, 2021-08-05 at 13:43 +0300, Dan Carpenter wrote:
> > > > > The "= {};" style empty struct initializer is preferred over = {0}.
> > > > > It avoids the situation where the first struct member is a pointer and
> > > > > that generates a Sparse warning about assigning using zero instead of
> > > > > NULL.  Also it's just nicer to look at.
> > >
> > > Perhaps a cocci script like the below could help too:
> > >
> > > $ cat zero_init_struct.cocci
> > > @@
> > > identifier name;
> > > identifier t;
> > > @@
> > >
> > > 	struct name t = {
> > > -	       0
> > > 	};
> > >
> > > @@
> > > identifier name;
> > > identifier t;
> > > identifier member;
> > > @@
> > >
> > > 	struct name t = {
> > > 	       ...,
> > > 		.member = {
> > > -		0
> > > 		},
> > > 		...,
> > > 	};
> >
> > My test turns up over 1900 occurrences.  There is the question of whether
> > {} or { } is preferred.  The above semantic patch replaces {0} by {} and
> > ( 0 } by { }.
>
> I saw that and I don't recall how to force one style or another
> to be output.

If you remove something and put it back, then Coccinelle takes care of
pretty printing it.  So the following produces {} everywhere.  Fortunately
Dan seems to prefer that...

@@
identifier name;
identifier t;
@@

        struct name t =
-              {0}
+              {}
        ;

@@
identifier name;
identifier t;
identifier member;
@@

        struct name t = {
               ...,
                .member =
-               {0}
+               {}
                ,
                ...,
        };

julia