[Octave-bug-tracker] [bug #65116] classdef display override populates ans with class when using inputname

[Rik]

Update of bug#65116 (group octave):

                  Status:                    None => Patch Reviewed         

    _______________________________________________________

Follow-up Comment #6:

I reviewed the patch and it works for the issue of this bug report, and it
passes all of the existing inputname() tests.

One thing I note is that the m-file had a FIXME note with three test cases.


## FIXME: Actually, it probably *isn't* worth fixing, but there are small
## differences between Octave and Matlab.
##
## 1) When called from the top-level or a script, Matlab throws an error
##
##   inputname (1)   % at command prompt
##   % Octave returns "", Matlab throws an error
##
## 2) cell or struct indexing causes all further names to be returned as ""
##
##   c = {'a', 'b'}
##   y = 1; z = 2;
##   fcn (c, y, z)
##   % inputname() would return 'c', 'y', 'z' for the inputs.
##   fcn (c{1}, y, z)
##   % inputname() would return '', '', '' for the inputs.
##
## 3) If inputname is not called from a function, Matlab walks up the stack
##    until it finds some valid code and then works from there.  This could
##    be relevant for mex files or anonymous functions.
##
##   f = @(x) inputname (x);
##   a = 1:4;
##   arrayfun (fn, a, 'uniformoutput', false)
##   % output is {'fn', 'a', '', ''}


I verified that Matlab does the same thing as Octave for case #2 so that
difference is no longer true.

For item #1, I think it would be better to throw an error at the first
opportunity when a programmer is using a function incorrectly.  The
alternative is that an error is eventually found downstream of the actual
source of the problem and it takes a long time to trace it back to the source.
 This seems like it might be easier to implement now that the code is in C++. 
Can't we know whether we are in a function context, or not, and thus whether
to emit an error?  I'm thinking of the function is_user_fcn_frame() for
example.

I'm less concerned about item #3.  The inputname() function seems more useful
in true functions rather than anonymous function handles.  Still, maybe there
is a way to simply implement this?  If not, I think we should still leave a
note in the code about the difference between Octave and Matlab. 





    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?65116>

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