[vlc-devel] vlc_math.h ?

[Rémi Denis-Courmont]

Le 2015-12-11 19:13, Francois Cartegnie a écrit :
> Le 11/12/2015 11:22, Rémi Denis-Courmont a écrit :
>> Le 2015-12-11 00:59, Francois Cartegnie a écrit :
>>> I need log2 and log2_ceil which are already duplicated in some 
>>> modules:
>>
>> log2 is a standard function so you can use it, so I fail to see why 
>> you
>> would need to wrap that.
>> log2_ceil is best implemented with clz family of functions, rather 
>> than
>> floating point maths.
>>
>
> moving
>
> static inline uint8_t vlc_log2( uint32_t val )
> {
>     return 31 - clz(val);
> }

You make it sound like a replacement of log2(), which this is 
definitely *not*. Also the principle of least surprise is defeated by 
failure to handle larger than 32-bits values. All in all I don't really 
see the point in adding such a trivial function.

> static inline uint8_t vlc_ceil_log2( uint32_t val )
> {
>     uint8_t n = vlc_log2( val );
>     if ((1U << n) != val)
>         n++;
>     return n;
> }

And this seems even more trivial.

-- 
Rémi Denis-Courmont
http://www.remlab.net/