style question: breaking inline assembly lines

[Chris Johns]

On 14/6/2023 10:20 am, Joel Sherrill wrote:
> 
> 
> On Tue, Jun 13, 2023, 4:43 PM Chris Johns <chrisj at rtems.org
> <mailto:chrisj at rtems.org>> wrote:
> 
>     On 14/6/2023 5:47 am, Joel Sherrill wrote:
>     >
>     >
>     > On Tue, Jun 13, 2023 at 9:51 AM Sebastian Huber
>     > <sebastian.huber at embedded-brains.de
>     <mailto:sebastian.huber at embedded-brains.de>
>     <mailto:sebastian.huber at embedded-brains.de
>     <mailto:sebastian.huber at embedded-brains.de>>>
>     > wrote:
>     >
>     >     On 13.06.23 00:04, Gedare Bloom wrote:
>     >     >       "b _ARM_Exception_default\n"
>     >     >       :
>     >     > -    : [cpufsz] "i" (sizeof(CPU_Exception_frame)),
>     >     > -      [cpuspoff] "i" (offsetof(CPU_Exception_frame, register_sp)),
>     >     > -      [v7mlroff] "i" (offsetof(ARMV7M_Exception_frame, register_lr)),
>     >     > -      [cpuvecoff] "J" (offsetof(CPU_Exception_frame, vector)),
>     >     > -      [cpuvfpoff] "i" (ARM_EXCEPTION_FRAME_VFP_CONTEXT_OFFSET),
>     >     > -      [cpacr] "i" (ARMV7M_CPACR),
>     >     > -      [vfpsz] "i" (ARM_VFP_CONTEXT_SIZE)
>     >     > +    : [cpufsz] "i"( sizeof( CPU_Exception_frame ) ),
>     >
>     >     If we place operators (e.g. &&, ||, ...) at the end of a broken line,
>     >     then we should do this for the : as well.
>     >
>     >
>     > OK.
>     >  
>     >
>     >
>     >     My current preference would be to format all non-third-party sources
>     >     with a standard clang-format selection. I guess in the long run, this
>     >     will be the easiest approach to maintain. If we use exotic options, then
>     >     we may up ending as clang-format maintainers.
>     >
>     >
>     > I think this is the thing we have to keep in mind and I even said I would
>     > go along with compromises when we started. Get as close as you can
>     > and we will have to accept that -- for now. We should definitely file tickets 
>     > with clang-format and ourselves to track things we don't like. If we get an 
>     > option in the future whether we or someone else implements it, we can 
>     > use it and reformat again. Those hopefully are not that invasive.
> 
>     Sounds like a plan.
> 
>     I am a little concerned about the version of clang-format I need as some
>     machines I work on are no at current OS versions.
> 
> 
> I am finding myself experimenting with it for a customer's C++ project. So far,
> I have not matched the style. But I am now on Rocky 9 since I built it from
> source and needed cmake 3 at least. So you can't be on an old distribution and
> build from source.

I use clang-format on a couple projects with something close the standard
format. It works well.

> And Doxygen has and three different major versions between RHEL 7, 8, and 9.
> Using a configuration file from one on another seems to give you warnings about
> unknown options or obsolete options. 
> 
> Mixing new and old systems is a pain.

It is but we have to for reasons specific to local sites.

Chris