[FFmpeg-devel] [PATCH 1/2] Replace FFMIN/FFMAX by type specific macros

[Andreas Rheinhardt]

Michael Niedermayer:
> Hi
> 
> On Wed, Jun 04, 2025 at 03:05:12AM +0200, Michael Niedermayer wrote:
> [...]
>>>>  #define FFSWAP(type,a,b) do{type SWAP_tmp= b; b= a; a= SWAP_tmp;}while(0)
>>>
>>> 1. I don't like that you change the common case (namely the integer one)
>>> and thereby making the name longer (and making this patch huge).
>>
>> I can just rename FFMINI/MAXI to FFMIN/MAX if preferred
> 
> are there any objections to FFMIND / FFMINF / FFMAXD / FFMAXF ?
> if not i intend to apply these in the next days (unless i forget)
> 

I don't want them to be public.

>>
>>
>>> 2. This is supposed to be used to provide optimized platform-specific
>>> floating-point min and max functions. How is this supposed to be done
>>> with your approach? Would macros.h get internal plattform-specific
>>> headers like intreadwrite.h that are only included when building FFmpeg
>>> to override the ordinary macros?
>>
>> I just intended to do the work that noone else wanted to do,
>> which was replacing the macros with type specific ones.
>>
>> These can then be moved to whatever is the best place when optimizations
>> are introduced or replaced by a static inline function where thats possible
> 
> thx
> 
> [...]
>