inconsistent row <-> column sub-vector behavior

[John W. Eaton]

On 22-Aug-2008, Sergei Steshenko wrote:

| Intuitively it's a bug; LHS type (row or column) is known even before
| the OP=, so the necessary type conversion routine can be called easily
| and early enough by the interpreter.
| 
| At all, in a type-loose language if I deal with vectors, why should I care
| about its column or row nature ? 

Neither Octave nor Matlab really have "vectors".  All objects have at
least two dimensions, so orientation (almost) always matters.  There
are some special cases where Matlab plays fast and loose with
orientation and lets slide some things that arguably should be
errors.  Octave generally tries to do the compatible thing for these
cases to be nice to users who want to run Matlab code in Octave.

| But, anyway, the deciding factor should be Matlab compatibility, shouldn't
| it ?
| 
| Again, could anyone please check how Matlab behaves ?

Matlab does not have any OP= operators.

jwe