Re: [Qemu-devel] [PATCH] Drop braces around single statement rule

[Blue Swirl]

On Mon, Aug 2, 2010 at 3:20 PM, Anthony Liguori <address@hidden> wrote:
> On 07/31/2010 06:49 PM, Aurelien Jarno wrote:
>>
>> On Sat, Jul 31, 2010 at 08:23:55PM +0000, Blue Swirl wrote:
>>
>>>
>>> On Sat, Jul 31, 2010 at 4:23 PM, malc<address@hidden>  wrote:
>>>
>>>>
>>>> History has shown that this particular rule is unenforcable.
>>>>
>>>> Signed-off-by: malc<address@hidden>
>>>> ---
>>>>  CODING_STYLE |   11 ++++++-----
>>>>  1 files changed, 6 insertions(+), 5 deletions(-)
>>>>
>>>
>>> Not again:
>>> http://lists.nongnu.org/archive/html/qemu-devel/2009-12/msg00484.html
>>>
>>> There are plenty of ways to make the rule enforceable, for example we
>>> could agree to start to revert commits which introduce new
>>> CODING_STYLE violations.
>>>
>>>
>>
>> It seems to be possible to add a pre-applypatch script to the git hook
>> directory, that will verify the commit and reject it if it doesn't
>> comply with the coding rules. Of course it's possible to commit a patch
>> anyway by using --no-verify.
>>
>
> There are certain aspects of CODING_STYLE that I think are pretty important.
>  For instance, space vs. tabs can really screw things up for people that
> have non-standard tabs.  This is something that enforcing at patch
> submission time seems to be really important.
>
> Type naming seems important too because it's often not isolated.  IOW, a
> poor choice in one file can end up polluting other files quickly that
> require interacting.  The result is a mess of naming styles.
>
> But something like braces around an if doesn't seem like it creates a big
> problem.  Most C programmers are used to seeing braces in some statements
> and not other.  Therefore, it's hard to argue that the code gets really
> unreadable if this isn't strictly enforced.
>
> So really, I think the problem is that we're enforcing the words of
> CODING_STYLE instead of the intent.  The intent of CODING_STYLE is to
> improve the readability of the code.  I think it requires a certain amount
> of taste to be applied.

The problem with that approach is that CODING_STYLE is not written
like that at all. All four requirements equally state that they
describe QEMU coding style. There is also no room left for
interpretation: "Every indented statement is braced; even if the block
contains just one
statement."