kicad-developers team mailing list archive

[Lorenzo Marcantonio <l.marcantonio@xxxxxxxxxxxx>]

On Fri, Mar 06, 2015 at 11:23:44AM +0100, Simon Richter wrote:
> Which is not really worse than the current state -- the derived class's
> method then becomes a non-virtual member function, and people call the
> base version.

AFAIK the suggested best practice is to add the virtual even on the
derived classes (not held by some developer here; IIRC there was
a commit just to *remove* the virtual on the subclasses member).

But the real issue is that overloaded and virtual are orthogonal so if
you overload don't have to virtualize. And many other things that make
C++ more painful to use than it should be :P:P

I think that at the end we are stuck to cope with bad language design.

Personally I prefer virtual on the derived classes (in fact I'd have
made it the default choice, with a nonvirtual keyword to specify static
dispatching:P)

-- 
Lorenzo Marcantonio
Logos Srl

Attachment: smime.p7s
Description: S/MIME cryptographic signature