On Mon, Jan 30, 2023 at 10:12:19PM +0100, Rasmus Villemoes wrote:
> On 30/01/2023 16.54, Tom Rini wrote:
> > On Sun, Jan 29, 2023 at 10:57:28PM +0100, Rasmus Villemoes wrote:
> >> On 29/01/2023 01.57, Simon Glass wrote:
> >>> CONFIG options must not use lower-case letter. 
> >>
> >> Why?
> > 
> > So, kconfiglib complains about these.
> 
> Which IMO would be a bug in kconfiglib. Can you point me at where that
> warning is in kconfiglib.py and how it looks and when one would
> encounter it?
> 
> > However, I can't find a formal
> > language definition and the kernel documentation doesn't specify, merely
> > imply that it should always be all uppercase.
> 
> Well, yes, mostly, but since the de facto specification (namely, the
> kernel's implementation) doesn't complain and the kernel's Kconfig files
> do contain several examples of config symbols with lowercase characters,
> why deviate? In particular, since we share a lot of code, if some piece
> of kernel code has an IS_ENABLED(CONFIG_FOO876xx), why make it harder to
> import and keep that in sync?
> 
> Perhaps we can get Masahiro to tell us whether lowercase characters are
> allowed in kconfig symbols or not.
> 
> For reference, another kconfig-using project decided to fix their own
> infrastructure around kconfig instead of enforcing uppercase symbols:
> 
> https://github.com/zephyrproject-rtos/zephyr/issues/40420

I think it would be very good if the kconfig language itself declared
what does and does not make up a valid symbol. Note that the only thing
complaining in U-Boot today is kconfiglib using tools.

-- 
Tom

Attachment: signature.asc
Description: PGP signature

Reply via email to