Re: exp sets the errno to ERANGE error .

[Szabolcs Nagy <Szabolcs dot Nagy at arm dot com>]

On 07/05/2019 10:43, Umesh Kalappa wrote:
> Thank you Nagy,
> 
> Standard states that "errno acquires the value ERANGE is
> implementation-defined" .
> 
> So ,its not mandate the setting the errno for uflow..

implementation-defined == implementation has to document
the behaviour, which glibc does, and it intends to set
ERANGE (whether that's a good decision may be debatable)
but currently we should set errno if possible.

> 
> ~Umesh
> 
> On Tue, May 7, 2019 at 3:07 PM Szabolcs Nagy <Szabolcs.Nagy@arm.com> wrote:
>>
>> On 07/05/2019 08:51, Umesh Kalappa wrote:
>>> Hi everyone ,
>>>
>>> we have situations ,where the following program sets the errno to
>>> 34(range error) in the version 2.28
>>>
>>> $cat test.c
>>> int main() {
>>> printf("errno %d\n", errno);
>>> printf("%f\n", exp(-1421.0));
>>> printf("errno %d\n", errno);
>>> }
>>>
>>> according the C11 standard the exp should not set the errno to ERANGE
>>> when it underflow to zero .
>>>
>>> so we think that  glibc (atleast 2.28) is not consistent with the C
>>> standard and any comments regrading the same .
>>
>> the c standard allows ERANGE for uflow.
>>
>> "If the result underflows, the function returns an
>>  implementation-defined value whose magnitude is no
>>  greater than the smallest normalized positive number
>>  in the specified type; if the integer expression
>>  math_errhandling & MATH_ERRNO is nonzero, whether
>>  errno acquires the value ERANGE is implementation-defined;"
>>
>> http://port70.net/~nsz/c/c11/n1570.html#7.12.1p6
>>
>> glibc documents its behaviour as
>>
>> "When underflow occurs, the underflow exception is raised,
>>  and zero (appropriately signed) or a subnormal value, as
>>  appropriate for the mathematical result of the function and
>>  the rounding mode, is returned. errno may be set to ERANGE,
>>  but this is not guaranteed; it is intended that the GNU C
>>  Library should set it when the underflow is to an appropriately
>>  signed zero, but not necessarily for other underflows."
>>
>> https://www.gnu.org/software/libc/manual/html_node/Math-Error-Reporting.html
>>
>>>
>>> Thank you  and looking forward for any suggestions / comments.
>>> ~Umesh
>>>
>>