lkml.org 
[lkml]   [2021]   [Feb]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: code style: Re: [PATCH v4] printk: Userspace format enumeration support
On Wed 2021-02-17 15:56:38, Chris Down wrote:
> Petr Mladek writes:
> > > > How about config PRINTK_INDEX?
> > >
> > > Ah yes, I also like that. PRINTK_INDEX is fine from my perspective and is
> > > more straightforward than "enumeration", thanks.
> >
> > It is better than enumeration. But there is still the same
> > problem. The word "index" is used neither in the code
> > nor in the debugfs interface. It is like enabling cars and
> > seeing apples.
> >
> > What about CONFIG_PRINTK_DEBUGFS?
> >
> > It seems that various subsystems use CONFIG_<SUBSYSTEM>_DEBUGFS
> > pattern when they expose some internals in debugfs.
>
> The thing I don't like about that is that it describes a largely
> inconsequential implementation detail rather than the semantic intent of the
> config change, which is what the person deciding what to include in their
> config is likely to care about. Often when I see "XXX debug interface" when
> doing `make oldconfig` I think to myself "yes, but what does the debugfs
> interface _do_?".

I see.

> If someone else was writing this patch, and I saw "CONFIG_PRINTK_DEBUGFS"
> appear in my prod kernel, I'd probably say N, because I don't need printk
> debugging information. On the other hand, if I saw "CONFIG_PRINTK_INDEX", I'd
> immediately understand that it's probably applicable to me.
>
> I'm happy to rename the debugfs structure as <debugfs>/printk/fmt_index if it
> helps, but personally I really feel CONFIG_PRINTK_{INDEX,ENUMERATION,CATALOGUE}
> is a lot more descriptive than just saying "it has a debugfs interface" in the
> config name for that reason.

PRINTK_INDEX sounds the best to me. Keep in mind that I am not a
native speaker.

And my concern will be gone when we use it also in the API and debugfs
hierarchy as suggested by Johannes.

Another compromise might be to have CONFIG_PRINTK_FORMATS_INDEX.
Then the prefix printk_format_, pf_ would still match the option.
Or we could use printk_format_index_m, pfi_ indexes.

Best Regards,
Petr

PS: I feel that I have enough bike-shading. I think that I will be
fine with anything that you choose ;-)

\
 
 \ /
  Last update: 2021-02-18 13:48    [W:0.052 / U:1.760 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site