Re: Clang-format configuration discussion - pt 1

[George Dunlap <george.dunlap@xxxxxxxxx>]

On Wed, Nov 8, 2023 at 9:53 AM Luca Fancellu <Luca.Fancellu@xxxxxxx> wrote:
>
> Hi all,
>
> Let’s kick off the discussion about clang-format configuration, with this 
> part 1 I would like to discuss some configurable
> that I feel are not controversial.
>
> You can find the serie introducing clang-format here:
> https://patchwork.kernel.org/project/xen-devel/cover/20231031132304.2573924-1-luca.fancellu@xxxxxxx/
> and there is also a patch linked to my gitlab account where you can find the 
> output for the hypervisor code.
>
> For a full list of configurables and to find the possible values for them, 
> please refer to this page:
> https://clang.llvm.org/docs/ClangFormatStyleOptions.html

Luca,

Thank you so much for the work that you've done here.

The approach in your v2 series looks plausible, as does a brief
overview of the items in this list.

One problem I have is how to really evaluate the proposed changes.  I
spent a few minutes skimming through the "megadiff" [1] output from
the v2 series, and while everything looked fine, that is a HUGE patch
to skim through.  I don't really have any way to know if there's some
rule introduced that I don't really agree with.

On the other hand, I want to avoid busy make-work and the invitation
to interminable bike-shedding discussions.

Is it possible, for instance, to start with a diff which will enforce
*just these settings* (column width, indentation, and so on)?  And
then add on new coding style changes one (or a few) at a time, in a
way that would make it easier to understand what effect each change is
having?  If so, do you think that's a reasonable approach?

If not, how do you propose to proceed?

Thanks,
 -George

[1] 
https://gitlab.com/luca.fancellu/xen-clang/-/commit/8938bf2196be66b05693a48752ebbdf363e8d8e1.patch