[Octave-bug-tracker] [bug #42061] dot() fails to diagnose non-conformable arguments

[Glenn Golden]

URL:
  <http://savannah.gnu.org/bugs/?42061>

                 Summary: dot() fails to diagnose non-conformable arguments
                 Project: GNU Octave
            Submitted by: grepfor
            Submitted on: Sun 06 Apr 2014 09:25:22 PM GMT
                Category: Interpreter
                Severity: 3 - Normal
                Priority: 5 - Normal
              Item Group: Incorrect Result
                  Status: None
             Assigned to: None
         Originator Name: Glenn Golden
        Originator Email: 
             Open/Closed: Open
         Discussion Lock: Any
                 Release: 3.8.0
        Operating System: GNU/Linux

    _______________________________________________________

Details:

The dot() function fails to complain if the operands are non-conformable(!).



    x = [1 2 3];
    y = [4 5];

    dot(x,y)
    ans =  14


This seems pretty gross, and one must ask (without meaning any disrespect):
How in the world could a bug like this have escaped testing?  Can it actually
be considered correct operation in some sense?

The doc claims that dot(x,y) is supposed to be equivalent to


    sum(conj(x).*y, DIM)



which makes sense, but clearly this cannot be the case when x and y are
non-conformable. Based on the results obtained, the behavior seems like the
shorter operand is being extended with zeros to the length of the longer one.

Can this actually be intended in some sense?




    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?42061>

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