[FFmpeg-devel] [PATCH]Simplify #if defined(__GCC__) && ...

[Benoit Fouet]

On 12/22/2008 01:52 PM, Michael Niedermayer wrote:
> On Mon, Dec 22, 2008 at 01:37:04PM +0100, Benoit Fouet wrote:
>   
>> On 12/22/2008 01:22 PM, Michael Niedermayer wrote:
>>     
>>> On Mon, Dec 22, 2008 at 08:30:02AM +0100, Benoit Fouet wrote:
>>>   
>>>       
>>>> On 12/20/2008 06:00 PM, Michael Niedermayer wrote:
>>>>     
>>>>         
>>>>> On Fri, Dec 19, 2008 at 08:35:43PM +0100, Carl Eugen Hoyos wrote:
>>>>>   
>>>>>       
>>>>>           
>>>>>> Hi!
>>>>>>
>>>>>> Attached patch should simplify the #if GCC_VERSION> condition that occurs 
>>>>>> several times in libavutil.
>>>>>>     
>>>>>>         
>>>>>>             
>>>>> [...]
>>>>>   
>>>>>       
>>>>>           
>>>>>> Index: libavutil/common.h
>>>>>> ===================================================================
>>>>>> --- libavutil/common.h	(revision 16223)
>>>>>> +++ libavutil/common.h	(working copy)
>>>>>> @@ -41,8 +41,10 @@
>>>>>>  #    include <math.h>
>>>>>>  #endif /* HAVE_AV_CONFIG_H */
>>>>>>  
>>>>>> +#define AV_GCC_VERSION_GREATER(x,y) (defined(__GNUC__) && (__GNUC__ > x || __GNUC__ == x && __GNUC_MINOR__ > y))
>>>>>> +
>>>>>>     
>>>>>>         
>>>>>>             
>>>>> i just wanted to say looks ok but then i realized that a
>>>>>
>>>>> AV_GCC_VERSION_AT_LEAST() would be more natural as it would contain the
>>>>> version of the first gcc that supported something not the last that did
>>>>> not
>>>>>
>>>>>   
>>>>>       
>>>>>           
>>>> well... it is the number of the last revision that does not support the
>>>> feature, no ?
>>>> we check for the minor to be strictly greater than the one which is
>>>> specified as an argument to the macro...
>>>> (not that I care much about the name, though)
>>>>     
>>>>         
>>> to me it feels more natural to list the first version that does support it.
>>>   
>>>       
>> to me too...
>>
>>     
>>> I mean if you look in some manual or press release or anything, they always
>>> list the first version that supports something not the last that did not.
>>> As a sideeffect it also makes it easier to use as one doesnt have to
>>> subtract 1 from what is listed somewhere in a manual to add a AV_GCC...
>>>
>>>   
>>>       
>> ... but when we check 'if version > XX.YY', we check for our version to
>> be (strictly) above the last one that did not support the feature, or am
>> I missing something ?
>>     
>
> we are checking ver >= XX.YY
> #define AV_GCC_VERSION_AT_LEAST(x,y) (defined(__GNUC__) && (__GNUC__ > x || __GNUC__ == x && __GNUC_MINOR__ >= y))
>
>   

ok, I obviously missed something then... sorry

>   
>> (once again, I am just wondering, this is not really important)
>>     
>
> if it really is buggy this of course woudl be important ...
>
>   

*I* am the buggy one... I hope this is not too important ;)

Ben