Re: [Bug-apl] Bug in 8○ function?

[Blake McBride]

Good to know, and I agree.

Thanks.

Blake

On Sat, Apr 18, 2015 at 7:12 AM, Juergen Sauermann <address@hidden> wrote:

Hi Elias,

correct. Unfortunately there seems to be a mismatch between the ISO standard  on one side,
and the IBM reference manual and the IBM APL2 interpreter on the other.

For example, the formula for 8○ is (¯1-B*2)*.5 in the ISO standard on page 92 but
-(¯1-B*2)*.5 in the IBM reference manual on page 80.

In the past I have resolved such differences in favour of the IBM implementation because few people
really follow the standard and following the standard would decrease the compatibility of GNU APL with
IBM APL2.

/// Jürgen

On 04/15/2015 04:42 PM, Elias Mårtenson wrote:

According to my documentation that I copied from the ISO spec, 8○ should be equivalent to: (¯1-R⋆2)⋆0.5 and ¯8○ should be equal to -(¯1-R×2)⋆0.5.

In GNU APL the opposite happens. I.e. ¯8○ returns what 8○ is supposed to return and vice versa.

Is my analysis correct?

Regards,

Elias