Re: [PATCH] CodingStyle: add typedefs chapter

[Johannes Stezenbach <js@linuxtv.org>]

On Tue, May 02, 2006, David Woodhouse wrote:
> On Tue, 2006-05-02 at 02:37 +0200, Johannes Stezenbach wrote:
> > IMHO u32 etc. are the well established data types used
> > everywhere in kernel source. Your wording suggests that
> > the use of C99 types would be better, and while I respect
> > your personal opinion, I think it is wrong to put that in the
> > kernel CodingStyle document.
> 
> Perhaps the word 'gratuitous' should be removed, then, if you object to
> being able to infer my opinion.
> 
> The point remains that the peculiarity should definitely be documented,
> along with an explanation of the reasoning (or lack of such) behind it.
> 
> > c.f. http://lkml.org/lkml/2004/12/14/127 
> 
> That's about types used for _export_. It's accepted that __uXX types are
> necessary in stuff which is visible by userspace. That was point (e).¹
> 
> The only bit in that mail which is relevant to my point (d) is the
> penultimate (and final) paragraphs. And those are a complete non
> sequitur and make just as much sense if you swap over 'u32' and
> 'uint32_t' and also 'kernel' and 'C language'...
> 
> "In other words, uint8_t/uint16_t/uint32_t/uint64_t (and the signed
> versions: int8_t and friends) _are_ the standard names in the C
> language, and the __u8/__u16/etc versions have always existed alongside
> them for things like header files that have namespace issues.
> 
> "So forget about that stupid abortion called "u32" already. It buys
> you absolutely nothing."

;-)

Maybe I got it wrong, but my impression so far was that
u8 etc. are preferred for kernel code, and C99 types
are merely tolerated. (Mostly for consistency reasons,
I guess, since most old code uses u8 etc.)

However, personally I don't care either way, I just
want to make sure that code written acording to
Documentation/CodingStyle also meets Linus' expectations.


Johannes