[Octave-bug-tracker] [bug #65291] [octave forge] (mapping) degrees2dms produces incorrect results

[anonymous]

Follow-up Comment #11, bug #65291 (group octave):

Somehow I didn't realize the Octave mapping package had angl2str.  Thanks for
pointing that out.

> > Does the `elseif (ischar (deg))` branch prevent some incorrect result, or
was it just added for clarity or to save unnecessary computation?
> Again, degrees2dms.m is not my code.
Based on your comment #2 I thought you added that part.  Maybe I didn't
understand what you meant.

> Then, you mention a d2dms function you wrote. Would you like to swap the
current mapping function's code with your code? Or amend it with your code?
> I'm all open to suggestions.
I am offering to modify the current degrees2dms function to resolve the
off-by-one-minute issue, improve tests, and correct some issues in the
documentation.  To calculate the magnitudes of the degrees-minutes-seconds
components, I propose to use one of the following two methods:
* The method used in the d2dms function I provided in comment #6.
        * Accuracy is reasonable.  Recomputing degrees from the output as 
`s/3600 +
m/60 + d` seems to be within 1 ulp or so of the original input.
        * It is quite simple.
        * Hopefully it is reasonably easy to become convinced that it has no 
major
issues.
        * Performance is good.
        * Based on comment #7 it seems to MATLAB-compatible for single.  I would
guess it is for double as well.
* A more accurate method I developed.
        * Output angle is as near as possible to the input angle.
        * It is roughly four times more complicated.
        * It was not very easy to convince myself that it's correct.  I still
wouldn't put my confidence over 99%.
        * Performance on my machine is roughly two times worse.
        * Due to the better accuracy, it is not MATLAB-compatible.

I am seeking input as to which approach would be preferred.

As for the handling of negative inputs, I am hopping someone will be able to
provide MATLAB output for the test in comment #9.



    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?65291>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/