Re: [PATCH] variants: avoid type punning issue.

[Akim Demaille]

Le 31 janv. 2013 à 21:51, Théophile Ranquet <address@hidden> a écrit :

> 2013/1/31 Akim Demaille <address@hidden>:
>>> diff --git a/data/variant.hh b/data/variant.hh
>>> index e2a537a..78aa876 100644
>>> --- a/data/variant.hh
>>> +++ b/data/variant.hh
>>> @@ -143,7 +143,7 @@ m4_define([b4_variant_define],
>>>     YYASSERT (tname == typeid (T).name ());
>>>     YYASSERT (sizeof (T) <= S);])[
>>>     {
>>> -        void *dummy = buffer.raw;
>>> +        void* dummy = buffer.raw;
>> 
>> Actually it was right in the first place.  This is the style
>> promoted by the GNU Coding Standards.
> 
> Oh? Well, I am of course aware of the GNU Coding Standards, but I saw
> these lines a bit further down in the file, so I assumed we were not
> following the standard there:
> 
>    /// Whether the content is built: if defined, the name of the stored type.
>    const char* tname;])[
> 
> Shall I address the style discrepancy here?

Yes please.