[Numpy-discussion] Unexpected float96 precision loss

[Christopher Barker]

Charles R Harris wrote:
>>> So if you write "float96(0.0001)", the result is not the float96 number
>>> closest to 0.0001, but the 96-bit representation of the 64-bit number
>>> closest to 0.0001.
...

>> but wouldn't it be better to exactly handle strings since those can be
>> converted exactly, which is what the user wants/expects?
>>
> Well, yes. But then we would need to write our own routines for the
> conversions...

The mechanism used in fromstring() and fromfile() (when used for text 
parsing) looks like it's designed to support this -- i.e. the actual 
string parsing is done be a function provided by the datatype.

I haven't looked to see if there is anything special there for float96.

Does the clib for a compiler that provides a float64 also provide an 
atof() function that supports it? Its seems that it should.

-Chris



-- 
Christopher Barker, Ph.D.
Oceanographer

Emergency Response Division
NOAA/NOS/OR&R            (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chris.Barker at noaa.gov