Re: [PATCH] checkpatch: improve error message for p1-check

[Joe Perches <joe@perches.com>]

On Thu, 2011-10-27 at 13:11 -0700, David Rientjes wrote:
> On Wed, 26 Oct 2011, Joe Perches wrote:
> > > I mean it only makes sense if both prefixes exist (otherwise patch and 
> > > git-apply will assume it's not a -p0 patch).
> > I think we should not care about the prefixes at all,
> > only whether or not the patched file exists.
> Nack,

Hi David.

It might be better if you would submit patches
to checkpatch before you nack others.

How about you track the --- and +++ lines and
submit a suggested patch yourself?

> there's nothing wrong with storing original files that you're 
> modifying in a subdirectory with a name of your choice in the kernel tree.  

Just as there's nothing wrong with storing original
and modified versions of subdirectories too.

> It doesn't imply a -p0 patch unless both prefixes appear and that's the 
> best indication that it appears in both the patch author and patch 
> applier's tree whereas the file being modified is ambiguous.

There's no single perfect test and it's just a silly
warning anyway.

I think the most common case is the direct editing of
a single file and production of a diff to submit as
a patch.
 
$ emacs <file>
# make changes, save original as ~ backup
$ diff -urN <file>~ <file> > ./foo.diff
$ make mrproper; make allyesconfig ; make
# deletes backup files
$ ./scripts/checkpatch.pl foo.diff

I didn't bother even finding out why the message
was emitted for me even though I had a b temp
directory in my tree.  I just ignored it.

cheers, Joe