Re: CVS commit: src/sys/arch

[Valery Ushakov <uwe%stderr.spb.ru@localhost>]

On Mon, Jan 25, 2021 at 20:28:52 +0300, Valery Ushakov wrote:

> On Mon, Jan 25, 2021 at 08:19:44 -0800, Jason Thorpe wrote:
> 
> > > On Jan 25, 2021, at 6:22 AM, Kamil Rytarowski <kamil%netbsd.org@localhost> wrote:
> > > 
> > > I have no problem with this change but I am curious why should we use "{
> > > }"? It's a C GNU extension and C++ syntax.
> > 
> > Using { 0 } makes an assumption about the first member of the
> > structure which is not guaranteed to remain true.
> 
> The commit message says:
> 
> | Since we're using designated initialisers for compat data, we should
> | use a completely empty initializer for the sentinel. 
> 
> but that "should" is not true.  The code that checks that sentinel
> uses some particular member to access it, so, pedantically speaking,
> the initialization must designate that member in the sentinel
> initialization.  Yes, this is verbose and doesn't look as pretty, but
> that is what "should" happen here.  Using non-standard {} extension
> makes it look nicer, but is not a "should" kind of necessity.

PS: Forgot to add that C++ doesn't have designated initializers (or at
least it didn't last time I looked), so they are in a different
situation here and need an empty initializer list.  In C the only
difference it makes is, as far as I can tell, exactly this kind of an
array with a sentinel at the end and the difference is cosmetic.

-uwe