[Lldb-commits] [lldb] r183716 - <rdar://problem/12783351>

[Greg Clayton]

On Jun 10, 2013, at 8:48 PM, Chris Lattner <clattner at apple.com> wrote:

> 
> On Jun 10, 2013, at 5:18 PM, Enrico Granata <egranata at apple.com> wrote:
> 
>> +++ lldb/trunk/source/Core/DataExtractor.cpp Mon Jun 10 19:18:18 2013
>> @@ -1302,6 +1302,22 @@ DumpAPInt (Stream *s, const DataExtracto
>>    return offset;
>> }
>> 
>> +static float half2float (uint16_t half)
>> +{
>> +    union{ float       f; uint32_t    u;}u;
>> +    int32_t v = (int16_t) half;
>> +    
>> +    if( 0 == (v & 0x7c00))
>> +    {
>> +        u.u = v & 0x80007FFFU;
>> +        return u.f * 0x1.0p125f;
>> +    }
>> +    
>> +    v <<= 13;
>> +    u.u = v | 0x70000000U;
>> +    return u.f * 0x1.0p-112f;
>> +}
>> +
> 
> Hi Enrico,
> 
> Where did you get this algorithm?  Did you consider using llvm::APFloat to do this for you?

The reason we didn't use lldb::APFloat when we first created the code was that it didn't support emitting human readable strings, it only emitted the hex compiler string format. We should switch _all_ float display over to using llvm::APFloat now that the class supports emitting human readable strings (at least I think it does).

Greg