Re: [PATCH v7 00/12] iio: adc: Add support for AD4170 series of ADCs

[Marcelo Schmitt <marcelo.schmitt1@gmail.com>]

On 07/03, Andy Shevchenko wrote:
> On Wed, Jul 02, 2025 at 03:59:12PM -0300, Marcelo Schmitt wrote:
> > On 07/02, Andy Shevchenko wrote:
> > > On Wed, Jul 02, 2025 at 09:00:42AM -0300, Marcelo Schmitt wrote:
> > > > On 07/02, Andy Shevchenko wrote:
> > > > > On Mon, Jun 30, 2025 at 10:57:32AM -0300, Marcelo Schmitt wrote:
> 
> ...
> 
> > > > > >  6 files changed, 3601 insertions(+)
> > > > > 
> > > > > This is weird. At least patches 11 & 12 have '-' lines...
> > > > > 
> > > > Yeah, sorry about that. These ADCs are fancy such that the base driver is about
> > > > 1500 LoCs due to channel setup handling and support for multiple combinations of
> > > > voltage references and channel setups.
> > > > 
> > > > About the '-' lines, I will rework ad4170_parse_channel_node() on earlier
> > > > patches to avoid 3 line removals in patch 11. Patch 12 is only makes sense
> > > > after patch 7 and I think it would lead to '-' lines if coming before patch 10
> > > > since both increment the number of IIO channels. Anyway, I'll see how to further
> > > > reduce the number of lines being removed.
> > > 
> > > My point is that the above statistics is mangled and I don't know how I can
> > > trust the contents of this series if it already lied about that.
> > 
> > Looks like git format-patch summarizes the changes from all patches when
> > printing the statistics to the cover letter. Also, git format-patch doc [1]
> > says the 'changes' dirstat option (default behavior) doesn't count
> > rearranged lines as much as other changes.
> 
> TIL. Thanks for pointing that out.
> 
> > There are cover letters of other
> > patch sets where the number of '-' lines don't match the sum of lines
> > removed by each patch. [2] and [3] are examples of that.
> 
> That's different I believe due to the diff algorithm in use
> (btw, do you use histogramm?).

Nope, I was using the default diff algorithm to generate the patches.
I tried the --histogram option today but didn't notice any difference. Maybe
default and histogram output the same result for the ad4170 set.
I'll send v8 using histogram alg. Nevertheless, is there any preferred alg for
generating patches (so I can use what's best on next patch sets)? 

Thanks,
Marcelo