Re: [RFC PATCH v2 3/5] rust: add PL011 device model

[Manos Pitsidianakis]

Hey Zhao,

On Thu, 11 Jul 2024 at 07:05, Zhao Liu <zhao1.liu@intel.com> wrote:
>
> Hi Manos and all,
>
> On Tue, Jun 11, 2024 at 01:33:32PM +0300, Manos Pitsidianakis wrote:
> > diff --git a/rust/rustfmt.toml b/rust/rustfmt.toml
> > new file mode 100644
> > index 0000000000..ebecb99fe0
> > --- /dev/null
> > +++ b/rust/rustfmt.toml
> > @@ -0,0 +1,7 @@
> > +edition = "2021"
> > +format_generated_files = false
> > +format_code_in_doc_comments = true
> > +format_strings = true
> > +imports_granularity = "Crate"
> > +group_imports = "StdExternalCrate"
> > +wrap_comments = true
> >
>
> I find it's stiil necessary to strictly limit the width of the lines by
> "error_on_line_overflow = true" [1].
>
> Currently rustfmt defaults the width limitation with "max_width = 100",
> but it has bugs and doesn't always work. For example, the line of
> rust/qemu-api/src/device_class.rs:108 comes to 157 characters and is
> ignored by rustfmt, which doesn't even fit in one line of my screen!
>
> Of course I think it's feasible to manually review and fix similar cases,
> but it's definitely better to have readily available tool that can help
> us rigorously formatted...
>
> [1]: 
> https://github.com/rust-lang/rustfmt/blob/master/Configurations.md#error_on_line_overflow
>
> -Zhao

Thank you for pointing that out, I hadn't noticed it! The problem is
that macro definitions are also macros and macros aren't formatted
because they have to be expanded, IIUC. I agree that a hard error on
an overflow is necessary for readable code.

By the way, my usual go-to workaround for this bug is to format the
macro body outside the macro_rules! { } scope, (rustfmt works even if
the code does not compile) and then put it back in.

Manos