Re: [Qemu-devel] [PATCH] checkpatch: allow tabs in linux-headers

[Cornelia Huck]

On Wed, 10 Aug 2016 15:55:28 +0200
Radim Krčmář <address@hidden> wrote:

> 2016-08-10 09:09+0200, Cornelia Huck:
> > On Tue, 9 Aug 2016 12:14:14 -0400 (EDT)
> > Paolo Bonzini <address@hidden> wrote:
> >> > Make scripts/checkpatch.pl accept tabs in linux-headers/, instead of
> >> > changing scripts/update-linux-headers.sh to expand tabs when importing.
> >> > 
> >> > Signed-off-by: Radim Krčmář <address@hidden>
> >> > ---
> >> > diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
> >> > index 929708721299..38232d4b25c3 100755
> >> > --- a/scripts/checkpatch.pl
> >> > +++ b/scripts/checkpatch.pl
> >> > @@ -1355,7 +1355,7 @@ sub process {
> >> >                  next if ($realfile !~ /\.(h|c|cpp|pl)$/);
> >> >  
> >> >  # in QEMU, no tabs are allowed
> >> > -                if ($rawline =~ /^\+.*\t/) {
> >> > +                if ($rawline =~ /^\+.*\t/ && $realfile !~ 
> >> > /^linux-headers\//) {
> >> >                          my $herevet = "$here\n" . cat_vet($rawline) . 
> >> > "\n";
> >> >                          ERROR("code indent should never use tabs\n" . 
> >> > $herevet);
> >> >                          $rpt_cleaners = 1;
> >> > 
> >> 
> >> Could you do the same for standard-headers/ too?
> > 
> > I think it would be better to not apply any qemu coding style checks to
> > a headers update. Something like 'check if this contains header updates
> > _only_' would make more sense, but that is beyond my nonexisting perl
> > skills...
> 
> I have posted another vesion that does not check for any code style in
> hunks that modify linux-headers and include/standard-headers,
> http://lists.nongnu.org/archive/html/qemu-devel/2016-08/msg01824.html
> 
> We still want to check header-only updates in other headers ...
> Your condition would draw attention to linux header updates that also
> touch other files, but I think that a diffstat is enough.
> 
> The script would need some preprocessing to know that only headers are
> modified or buffering of errors until the script knows that only headers
> were modified; neither is hard, but the added complexity is not
> compensated by usefulness, IMO.
> 

If there's no quick way to check, it's not worth spending too much time
on it, I agree.