Hi,
On Mon, Jun 27, 2011 at 03:35:41PM +0530, Jaswinder Singh wrote:
> On 27 June 2011 14:05, Felipe Balbi <balbi@xxxxxx> wrote:
> >> >> > +static struct regulator_consumer_supply sdp4430_vusb_supply =
> >> >> > + REGULATOR_SUPPLY("hsusb0", "ehci-omap.0");
> >> >
> >> > this should be an array.
> >> Ok, I can make it an array of _one_ element.
> >> Though I am not sure why is that a good thing, or are we to use another
> >> possible VUSB supply on Panda/SDP boards ? Please suggest so that
> >> I can add that too.
> >
> > same comment I gave before to another patch:
> >
> > it makes the diff a lot easier to understand should anyone modify this
> > later. It's also a matter of consistency.
> >
> A quick grep showed otherwise though ...
>
> In arch/arm/mach-omap2/
> Total regulators defined = 71
> Regulators with exactly 1 supply = 58
> Single element non-array definitions = 46/58
> Single element array definitions = 12/58
>
> Even if we consider 20% to be norm for consistency, I am not sure it's
> a good one.
the patch which converted all non-array, to array seems to have been
taken yet, then.
> Still many samsung guys piously enclose magic value defines in parenthesis,
> just to maintain 'consistency' !
that's just plain stupidity.
> And, I don't understand how does diff become any easier beyond 2
> elements in the array.
http://marc.info/?l=linux-omap&m=130738044715490&w=2
> Sorry for being bitchy, but I am unable to buy any reason other than
> having more than
> one element to use array.
I also seem to recall someone (either Russell or Linus) once explained
why we should never mistake one-element arrays with pointers.
Unfortunately, I fail to find the thread, it's quite old.
--
balbi
Attachment:
signature.asc
Description: Digital signature
