[J3] Problem with IEEE_ARITHMETIC examples

Bill Long longb at cray.com
Fri Jan 18 15:00:14 EST 2019


> On Jan 18, 2019, at 12:16 PM, Steidel, Jon L via J3 <j3 at mailman.j3-fortran.org> wrote:
> 
> The examples in 17.11 for IEEE_MAX_NUM, IEEE_MIN_NUM, IEEE_QUIET_EQ, IEEE_QUIET_GE, IEEE_QUIET_GT, IEEE_QUIET_LE, IEEE_QUIET_LT, IEEE_QUIET_NE, IEEE_SIGNALING_EQ, IEEE_SIGNALING_GE, IEEE_SIGNALING_GT, IEEE_SIGNALING_LE, IEEE_SIGNALING_LT, and IEEE_SIGNALING_NE are all flawed.  The examples all contain a call to IEEE_VALUE (IEEE_INVALID).  The definition of IEEE_VALUE has two arguments, IEEE_VALUE (X, CLASS); X is not OPTIONAL.  The simple fix would be to change IEEE_VALUE (IEEE_INVALID) to IEEE_VALUE (0.0, IEEE_INVALID) in all 14 places.

Arg.  Thie mildly counts as an internal inconsistency in the standard, but not at the usual level of causing confusion or a logical contradiction. 

>  
> There is also a problem in the Introduction, page xiii,  p2 halfway thru the bullet “Data usage and computation”.  “The standard intrinsic operations . . .; the = and /= operations …” should be “The standard intrinsic operations . . .; the == and /= operations . . .”
>  

This looks like a simple typo. 


> I am not sure if these can be editorial changes, or if they require an interpretation.  I can prepare a paper for the next meeting either way.  Please advise as to the proper way proceed.

I would submit an ordinary paper to be assigned to /Edit.  If Malcolm thinks an Interp is right (especially for the IEEE problem) he will convert (part of) the paper into an interp. 

The main difference with making it an interp is that the Standard is modified with the repairs before the next revision is published.  And we have to vote on these on the next Interp Ballot.  (OTOH, a no-brainer ballot vote would be nice for a change.)

Cheers,
Bill


>  
> -jon

Bill Long                                                                       longb at cray.com
Principal Engineer, Fortran Technical Support &   voice:  651-605-9024
Bioinformatics Software Development                      fax:  651-605-9143
Cray Inc./ 2131 Lindau Lane/  Suite 1000/  Bloomington, MN  55425




More information about the J3 mailing list