Re: clang-format

"Kevin J. McCarthy" <[email protected]> Sun, 15 Mar 2026 10:49:16 +0800

Newsgroups	gmane.mail.mutt.devel
Message-ID	<abYeLNLCBDiN-X4F@qinghai>

On Sat, Mar 14, 2026 at 06:07:51PM +0100, Rene Kita wrote:
>I don't like formatters. From my experience they work most of the time,
>but are a massive PITA the other times. Especially when you don't
>control the machine the formatter is running on. Different versions will
>behave differently or will not accept some options because the version
>is too old. Even worse when the formatter is part of the CI pipeline.
>Then I will have to fiddle manually with code to make the version
>running in CI happy if my local version outputs something different.

I understand, especially after noting the config vars and expected 
values seem to change quite a bit over versions.

As I mentioned in my reply to Alex, perhaps we could run this once, as a 
general clean up, leave the config as a reference, but not hook it up in 
the pipeline?

>I would expect volunteer contributors to be motivated enough to submit
>readable code - in contrast to some corporate environment. Is/Was bad
>formatting really a problem that warrants the extra effort?

Mmm... generally, but not always.  The biggest problem is the space
before parens, but see my reply to Alex.

Another issue are the various differences between source files.  For 
example, some files align '*' next to vars, some next to type:

   int* foo;
   int *foo;

And little quirks like that.  It's not a huge deal for me either, but 
when you go to work in a function that's different, it's hard to know 
whether to fix up the whole function, go with the flow in the function, 
or do it the "right way" in just your change, which contributes to the 
feeling of "random formatting".

-- 
Kevin J. McCarthy
GPG Fingerprint: 8975 A9B3 3AA3 7910 385C  5308 ADEF 7684 8031 6BDA

signature.asc (application/pgp-signature, 833 B)