Re: clang-format

Alejandro Colomar via Mutt-dev <[email protected]> Wed, 18 Mar 2026 20:55:53 +0100
Newsgroups	gmane.mail.mutt.devel
Message-ID	<absCpD42YlnY6WKj@devuan>
On 2026-03-18T18:14:02+0100, Alejandro Colomar wrote:
> Hi Will,
> 
> On 2026-03-18T09:54:38-0700, Will Yardley wrote:
> > On Wed, Mar 18, 2026 at 11:21:28AM +0800, Kevin J. McCarthy wrote:
> > > On Wed, Mar 18, 2026 at 01:11:14AM +0100, Alejandro Colomar via Mutt-dev wrote:
> > > > Okay, so I've tried today clang-format(1) for the first time, after
> > > > someone suggested it in a project I co-maintain, and now have a well
> > > > formed opinion about it:
> > 
> > > > It's pure crap.  Please don't use it.
> > [...]
> > > Before I drop the proposal, though, are there any Mutt devs/contributors who
> > > would like to play devil's advocate?
> > 
> > In more or less the same boat as Robin on this.
> > 
> > I am not a C programmer, and can't speak to clang-format specifically. I
> > will say broadly that, since starting to use them, I have really
> > appreciated autoformatters generally (ruff / black for Python, prettier
> > for js, etc.), especially when working with a team of people. It may
> > well turn out that there isn't a good tool available for C, but I do
> > think that if it is possible to find a formatter that works and can be
> > enforced, there is a lot to be said for both the resulting consistency,
> > and kind of taking arguments / bikeshedding about code style and
> > formatting out of the equation.
> > 
> > I'm not sure if the concerns Alejandro mentioned are about the
> > functionality of the program, its default configuration, or both. It
> > does seem like the program supports formatting based on different style
> > guides, and configuring certain parameters
> > (https://clang.llvm.org/docs/ClangFormatStyleOptions.html), so maybe
> > some objections could be overcome by either disabling certain types of
> > formatting, or choosing different config settings?
> 
> The problem is the fundamental functionality of the program.  It doesn't
> have any way to fix the problems with configuration.

BTW, clang-format(1) programmers are aware of this limitation (people
have reported it in the past), and they claim it will never be
implemented, because the design of clang-format makes it almost
impossible.

See also:
<https://stackoverflow.com/questions/30763869/how-can-i-apply-only-one-clang-format-action>
<https://github.com/llvm/llvm-project/issues/53924>


Cheers,
Alex

> 
> > Or, maybe come up with the style guide first, then work backwards from
> > that to see if a formatter / formatter config can be created that will
> > work?
> 
> I don't know if there are any other formatters that would be good.
> And I don't think writing a formatter from scratch just for mutt(1)
> would be reasonable.  But what is clear is that clang-format(1) is
> really bad.  If anyone wants to work on a new formatter, or knows of
> another formatter that is better, I'm fine with that.
> 
> 
> Have a lovely day!
> Alex
> 
> -- 
> <https://www.alejandro-colomar.es>



-- 
<https://www.alejandro-colomar.es>
signature.asc (application/pgp-signature, 833 B)