Re: weird diff output?

From: Stefan Beller <sbeller@google.com>
Date: Mar 30, 2016, 19:40 UTC
Message-ID: <CAGZ79kbk5T5SdSzfZ8Q6TQmXgiG+ZSUYc5E7_95KtariDU8MHQ@mail.gmail.com>
In-Reply-To: <CA+P7+xrbNQqGhR_EoVe7zou_g6oVFGN_v+q+tyHguv1BCMcimQ@mail.gmail.com>

On Wed, Mar 30, 2016 at 12:31 PM, Jacob Keller <jacob.keller@gmail.com> wrote:

Show 26 quoted lines

>
>           If unsure, say Y.
> +
> +config RMI4_I2C
> +       tristate "RMI4 I2C Support"
> +       depends on RMI4_CORE && I2C
> +       help
> +         Say Y here if you want to support RMI4 devices connected to an I2C
> +         bus.
> +
> +         If unsure, say Y.
>
> after:
>
>           required for all RMI4 device support.
>
> +         If unsure, say Y.
> +
> +config RMI4_I2C
> +       tristate "RMI4 I2C Support"
> +       depends on RMI4_CORE && I2C
> +       help
> +         Say Y here if you want to support RMI4 devices connected to an I2C
> +         bus.
> +
>           If unsure, say Y.

The optimum would be:

  >
  >           If unsure, say Y.
  >
  > +config RMI4_I2C
  > +       tristate "RMI4 I2C Support"
  > +       depends on RMI4_CORE && I2C
  > +       help
  > +         Say Y here if you want to support RMI4 devices connected to an I2C
  > +         bus.
  > +
  > +         If unsure, say Y.
  > +
  >  config BLA_I2C

The overlapping lines:
  > +
  > +         If unsure, say Y.
  > +

However that broke the lines at the first empty line, not the last as Jeff claimed it. (Could there be a problem in the perl script when empty lines are at the first or last overlapping line?)

Thanks for going through examples! (I would, too. But fixing a submodule regression is more important now; I only develop new features when there are no known regressions caused by me)

Thanks, Stefan

Show 11 quoted lines

>
> So in this particular instance which has multiple blank lines and is a
> similar issue as with Stefan's note above, this is where the heuristic
> falls apart. At least for C code this is basically vanishingly small
> compared to the number of comment header fix ups.
>
> I think it may be that Stefan's suggestions above may be on the right
> track to resolve that too.
>
> Regards,
> Jake