[Octave-patch-tracker] [patch #9566] mapping package: ecc2n, n2ecc, majaxis, minaxis

[Philip Nienhuis]

Follow-up Comment #4, patch #9566 (project octave):

Anonymous, I'm not angry at you, not at all. Apologies if my tone was felt too
sharp :-)

But I really wonder if you did run the functions in Octave? and did you test
them (e.g., "test majaxis")? that would have shown a few obvious things to
you.

* ischar assumes the only wrong input could be char. But unwary users, or
unwary scripts, might inadvertently also input cell arrays, structs, objects
or in far-fetched cases maybe even graphics handles. I also tried with
complex, which doesn't make much sense in the output.
So the best is to not check for a specific wrong type of input, but do the
other way round: reject all types that don't make sense and accept just the
one type that does.
I didn't (yet) exclude complex; should we simply discard the imaginary part?
Orientation (row or column vector): proper orientation should be checked for.

* All if conditions should be between ()

* I like to check the math a last time before proceeding.

* I have little time the next weeks so expect slow progress.

You'll probably see by now, contributing a simple function looks easy, but to
make it good quality it needs much more attention and code for dull stuff like
input validation and error checking.

As to how further:
No idea yet. With my adaptations it isn't that much work anymore to make them
publishable. I might do it, you might do it, see my comments above.


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/patch/?9566>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/