[Octave-bug-tracker] [bug #42528] textscan: complex numbers are incorrectly processed when format is empty

[Philip Nienhuis]

Follow-up Comment #3, bug #42528 (project octave):

I had a brief look:

The culprit is a matrix transpose operator that also invokes the complex
transpose (as per the definition of the transpose operator BTW, see the
docs).

Quick & dirty fix:  replace L. 332:
C(1) = reshape (C{1}, ncols, numel (C{1}) / ncols)';
by
C(1) = conj (reshape (C{1}, ncols, numel (C{1}) / ncols))';

textscan passes all tests then, but I'm not sure yet whether this is a
reliable fix. More investigation is needed for other data types - but I have
not time for that now.

Note that Matlab doesn't accept "textscan ('1i')" (it complains about a
missing format string).


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?42528>

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