Re: [PATCH] checkpatch: remove unneeded or ("|")

[Joe Perches <joe@perches.com>]

On Tue, 2015-03-24 at 12:54 +0100, Christian Borntraeger wrote:
> Am 12.03.2015 um 15:56 schrieb Andy Whitcroft:
> > On Thu, Mar 12, 2015 at 07:13:35AM -0700, Joe Perches wrote:
> >> On Thu, 2015-03-12 at 15:07 +0100, Christian Borntraeger wrote:
> >>> while porting commit 89a883530fe7 ("checkpatch: ## is not a valid
> >>> modifier") to QEMU, Peter Maydell noticed that the | at the end of
> >>> the list is not necessary.
> >>>
> >>> Lets get rid of it in kernel checkpatch.pl
> >>
> >> Andy?  Was that meant to allow nothing?
> >>
> >>> diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
> >> []
> >>> @@ -1550,7 +1550,7 @@ sub possible {
> >>>  			asm|__asm__|
> >>>  			do|
> >>>  			\#|
> >>> -			\#\#|
> >>> +			\#\#
> >>>  		)(?:\s|$)|
> >>>  		^(?:typedef|struct|enum)\b
> >>>  	    )}x;
> > 
> > Blimey you tax my memory indeed.
> > 
> > The two places it is used we are saying that the strings which match are
> > not modifiers.  So it seems sane that we would want to reject the empty
> > string in that case.  That said, it does not appear any of the callers
> > would call with a blank string.  I would suggest we had it like that as
> > a safety feature, _though_ if we had I would have expected it to have a
> > nice shiney comment to say just how smart we being using that trailing
> > or and so likely as not is is unintentional.
> > 
> > tl;dr I think it is safe to elide it with the current callers, it being
> > there seems safe, but if it stays damn it should have a comment to say
> > its a safety net and not just hide out.
> > 
> > -apw
> > 
> 
> So how to proceed?
> Take my patch, drop my patch or rework my patch?

If Andy acks it, I suggest sending it again
along with Andy's ack, to Andrew Morton (cc'd)