Re: [PATCH v4] devtools: add .clang-format file

[Tyler Retzlaff <roretzla@linux.microsoft.com>]

On Tue, May 14, 2024 at 08:56:49AM +0100, Bruce Richardson wrote:
> On Mon, May 13, 2024 at 09:11:32PM +0200, Morten Brørup wrote:
> > > From: Stephen Hemminger [mailto:stephen@networkplumber.org]
> > > Sent: Monday, 13 May 2024 17.55
> > > 
> > > On Mon, 13 May 2024 14:08:07 +0100
> > > Ferruh Yigit <ferruh.yigit@amd.com> wrote:
> > > 
> > > > 2. Double tab indentation vs parenthesis align
> > > >          if (iter->bus != NULL &&
> > > >  -                       /* not in middle of rte_eth_dev iteration, */
> > > >  -                       iter->class_device == NULL) {
> > > >  +           /* not in middle of rte_eth_dev iteration, */
> > > >  +           iter->class_device == NULL) {
> > > >
> > > > DPDK coding guide suggests double tab, but also accepts alignment by
> > > > spaces. But as far as I can see most of code has double tab.
> > > > Majority of the diff caused because of this rule.
> > > 
> > > 
> > > I personally am more used the aligned style, and most tools support
> > > that.
> > > The DPDK one is unique (not done by most other projects). So can we just
> > > keep the kernel (what is this clang-format) version.
> > 
> > I personally prefer the double tab.
> > It also works with editors showing tab as 4 space indentation.
> > 
> > Mixing tabs and spaces only works if the editor shows tabs as 8 space indentation.
> > 
> > Double tab works with both editor configurations.
> > 
> > And there is no confusion if the following block happens to be aligned with the following parameters. E.g.:
> > 
> > if fool(x,
> >         y)
> >         myfn();
> > 
> > vs.
> > 
> > if fool(x,
> >                 y)
> >         myfn();
> > 
> 
> +1, I also prefer the double tab too for this reason. The other
> consideration is that double tab leads to smaller diffs on refactor - with
> aligning brackets if something on the first line changes it could cause
> whitespace changes to be needed on all subsequent lines 
> 
> Overall, ignoring our individual preferences, since we already have a mix
> in DPDK, I think it's infeasible to try and enforce a single standard now.
> :-(

-1 to anything that aligns with parameters. it's the most mechanically
inconsistent convention ever, hostile to refactoring and causes a
requirement to endlessly shift your eyes horizontally when reading
code sequentially.