Re: [PATCH] .clang-format: update column limit

[Miguel Ojeda]

On Thu, Jun 11, 2020 at 12:36 PM Joe Perches <joe@xxxxxxxxxxx> wrote:
>
> Exactly.  So don't set a new hard limit of 100.
>
> This would _always_ wrap lines to 100 columns when
> 80 columns is still preferred.

Why is 80 "still preferred" to begin with? The patch you sent for
`coding-style.rst` was picked up by Linus, yes, but the wording seems
too strong still, considering it is for newcomers.

The point is that 80 *isn't* a limit, so I don't see why it is
mentioned, much less "preferred". Rather, I would have worded it like
[*]. What do you think?

> Imagine using a 100 column limit where a statement still
> fits on 2 lines.  Now imagine the same statement wrapped
> at 80 columns still fitting on 2 lines.
>
> Which would you prefer and why?

The former. While sometimes it may be more aesthetically pleasing to
have 2 lines of similar lengths rather than a long one and a short
one, having a deterministic approach allows us to use automatic
formatters. Which in turn makes code more regular since breaks are
always done the same way (modulo heuristic differences between
clang-format versions etc.).

In other words, I prefer automatic breaks vs. discussing every break :-)

Cheers,
Miguel

[*] (please excuse any word-wrap)