> On 26 Jan 2016, at 12:17, Jacob Erlbeck <[email protected]> wrote:
>
> On 26.01.2016 11:22, Holger Freyther wrote:
>>
>>> On 22 Jan 2016, at 16:46, [email protected] wrote:
>
>>> +/* we assume that x have at least 1 non-b bit */
>>> +static inline unsigned _leading_bits(uint8_t x, bool b)
>>> +{
>>> + if (b) {
>>> + if (x < 0x80) return 0;
>>> + if (x < 0xC0) return 1;
>>> + if (x < 0xE0) return 2;
>>> + if (x < 0xF0) return 3;
>>> + if (x < 0xF8) return 4;
>>> + if (x < 0xFC) return 5;
>>> + if (x < 0xFE) return 6;
>>> + } else {
>>> + if (x > 0x7F) return 0;
>>> + if (x > 0x3F) return 1;
>>> + if (x > 0x1F) return 2;
>>> + if (x > 0xF) return 3;
>>> + if (x > 7) return 4;
>>> + if (x > 3) return 5;
>>> + if (x > 1) return 6;
>>> + }
>>
>> same coding style thing
>
> In such cases, the readability would suffer when putting the returns in
> the next line. I'd rather right align the numbers after the '>', but YMMV.
sure or somemthing like
case 0xff..0x80:
return 0;
case 0x7F..0xC0:
return 1;
anyway, the above is okay and something we can argue with but if it is more
than a "look-up table" then we should break the line.
