Re: [Gnucap-devel] problem with clone_instance

[Felix Salfelder]

On Sun, Nov 17, 2013 at 02:33:43PM -0500, al davis wrote:
> Immediately, my suggestion followed in that same email.

i didn't get it, and i still fail to see the connection to "pulse".
the example in d_logic looks more instructive.

> An assert in the code is documentation of some condition that is 
> assumed to be true.  An assert fail is not a bug in itself, but 
> rather a hint at where the bug really is.o

i'm now considerably successful with something like

/*--------------------------------------------------------------------------*/
class NOTHING : public COMPONENT { //
public:
  NOTHING():COMPONENT(){ }
  CARD* clone() const { return new(NOTHING);}
private:
  std::string value_name()const { return "error";}
  int max_nodes()const {return 0;}
  std::string dev_type()const {return "error";}
  std::string port_name(int)const {untested(); return "error";}
  bool print_type_in_spice()const {return 0;}
};
/*--------------------------------------------------------------------------*/
static NOTHING p0;
static MODEL_SEMI_RESISTOR  p1(&p0); // <= fix the corresponding constructors
static MODEL_SEMI_CAPACITOR p2(&p0); // <= fix the corresponding constructors
static DISPATCHER<MODEL_CARD>::INSTALL
  d1(&model_dispatcher, "r|res", &p1),
  d2(&model_dispatcher, "c|cap", &p2);
/*--------------------------------------------------------------------------*/

in bmm_semi.cc. but it segfaults because of what i consider a bug in
COMPONENT::set_port_by_index(..). the comparison "num <= max_nodes()" seems to
be off by one.

try
gnucap>verilog
gnucap-verilog>resistor r(1,2);
gnucap-verilog>resistor r(1,2,3);
gnucap-verilog>resistor r(1,2,3,4);
resistor r(1,2,3,4);
                 ^ ? too many: requested=3 max=2

it should read "num < max_nodes()" in COMPONENT::set_port_*, no?

and, while i'm at it. in e_compon.h i found
[..]
public:  // obsolete -- do not use in new code
  virtual bool print_type_in_spice()const = 0;
[..]
which should read
  virtual bool print_type_in_spice()const{return some_default;},
if that member is really obsolete.

regards
felix