a problem with poliporphism

**KPB** · Jul 22 '05, 11:10 PM

Re: a problem with poliporphism

Nafai wrote:[color=blue]
> I am going to give a more detailed example:
>
> // I don't type the constructors.
>
> class Event {
> private:
> int time;
> EventType kind;
> public:
> enum EventType { Event1, Event2,...,Emer gency,Crime};
> int when() {return time; }
> EventType what() { return kind; }
> }
>
> class Emergency : public Event {
> private:
> int area;
> public:
> int where() {return area;}
> };
>
> class Crime : public Event {
> private:
> string person;
> int aCrime;.
> public:
> string who() { return person; }
> int whatHeDid() { return aCrime; }
> };
>
> void doSomethingWith Events(Event* pE)
> {
> switch(pE->kind) {
> case Event::Emergenc y : callAmbulance(p E->where(),pE->when());
> break;
> case Event::Crime: blame(pE->who(),pE->whatHeDid(), pE->when());
> break;
> case Event::Event1 : doSomethingRela tedToEvent1(pE->when()); break;
> ...
> }
> }
>
> int main() {
> list<Event*> lst;
> .... // insert somehow Events in lst
> Event* pEvt = lst.first();
> lst.pop_first() ;
> doSomethingWith Event(pEvt);
> ....
> }
>
>
> My question is: which is the best way to do that? Is it OK like before?[/color]

In case it's not an oversight on your part, I just want to point out
that you should add a virtual destructor to your Event class. Even if
it's an empty *noop* you should still add it.

KPB

**Nafai** · Jul 22 '05, 11:10 PM

Re: a problem with poliporphism

"KPB" <k@w.net> escribió en el mensaje
news:vNSAd.2577 3$EH1.11348@fe1 0.lga...[color=blue]
> Nafai wrote:[color=green]
>> I am going to give a more detailed example:
>>
>> // I don't type the constructors.
>>
>> class Event {
>> private:
>> int time;
>> EventType kind;
>> public:
>> enum EventType { Event1, Event2,...,Emer gency,Crime};
>> int when() {return time; }
>> EventType what() { return kind; }
>> }
>>
>> class Emergency : public Event {
>> private:
>> int area;
>> public:
>> int where() {return area;}
>> };
>>
>> class Crime : public Event {
>> private:
>> string person;
>> int aCrime;.
>> public:
>> string who() { return person; }
>> int whatHeDid() { return aCrime; }
>> };
>>
>> void doSomethingWith Events(Event* pE)
>> {
>> switch(pE->kind) {
>> case Event::Emergenc y : callAmbulance(p E->where(),pE->when());
>> break;
>> case Event::Crime: blame(pE->who(),pE->whatHeDid(), pE->when());
>> break;
>> case Event::Event1 : doSomethingRela tedToEvent1(pE->when());
>> break;
>> ...
>> }
>> }
>>
>> int main() {
>> list<Event*> lst;
>> .... // insert somehow Events in lst
>> Event* pEvt = lst.first();
>> lst.pop_first() ;
>> doSomethingWith Event(pEvt);
>> ....
>> }
>>
>>
>> My question is: which is the best way to do that? Is it OK like before?[/color]
>
> In case it's not an oversight on your part, I just want to point out that
> you should add a virtual destructor to your Event class. Even if it's an
> empty *noop* you should still add it.
>
> KPB
>[/color]

OK.

But do you think that is a good desing? How could it be done better?

**Jonathan Mcdougall** · Jul 22 '05, 11:15 PM

Re: a problem with poliporphism

Nafai wrote:[color=blue]
> I am going to give a more detailed example:
>
> // I don't type the constructors.
>
> class Event {
> private:
> int time;
> EventType kind;
> public:
> enum EventType { Event1, Event2,...,Emer gency,Crime};
> int when() {return time; }
> EventType what() { return kind; }[/color]

Event is meant to be used as a base class and does
not have a virtual destructor nor virtual
functions. What kind of base class is that?
[color=blue]
> }
>
> class Emergency : public Event {
> private:
> int area;
> public:
> int where() {return area;}
> };
>
> class Crime : public Event {
> private:
> string person;
> int aCrime;.
> public:
> string who() { return person; }
> int whatHeDid() { return aCrime; }
> };
>
>
> void doSomethingWith Events(Event* pE)
> {
> switch(pE->kind) {
> case Event::Emergenc y : callAmbulance(p E->where(),pE->when());
> break;
> case Event::Crime: blame(pE->who(),pE->whatHeDid(), pE->when());
> break;
> case Event::Event1 : doSomethingRela tedToEvent1(pE->when()); break;
> ...
> }
> }[/color]

That is not recommended. You seem not to
understand the use of polymorphism and public
inheritance.

Public inheritance has many uses, but in your
case, it is a simple case of interface<->behavior.
The base class specifies an interface (with
[pure] virtual functions) and derived classes
specify the behavior.

class Event
{
public:
virtual ~Event();

virtual void process() = 0; // that's the
// interface
};

class Emergency : public Event
{
private:
void call_ambulance( );

public:
virtual void process() // that's a behavior
{
call_ambulance( );
}
};

class Crime : public Event
{
private:
void blame();

public:
virtual void process()
{
blame(); // that's another behavior
}
};

typedef std::sector<Eve nt*> Events;

void f(Events &events)
{
for (Events::iterat or itor=events.beg in();
itor!=events.en d();
++itor)
{
Event *e = *itor;
e->process(); // calls the correct
// process()
}
}

Jonathan

**Nafai** · Jul 22 '05, 11:15 PM

Re: a problem with poliporphism

KPB escribió:[color=blue]
> Nafai wrote:
>[color=green]
>> Do you think it is a good a idea to do this:
>> void doSomethingWith Events(Event* pE)
>> {
>> Emergency* pEmergency;
>> switch(pE->kind) {
>> case Event::Emergenc y :
>> pEmergency=dyna mic_cast<Emerge ncy*>(pE);
>> callAmbulance(p Emergency->where(), pEmergency->when());
>> break;
>> ... // and so on
>> }
>> }[/color]
>
>
> No. That leads me to another point on your classes. What's the point in
> having polimorphism if the specific kind of event is stored as some
> internal value of the Event class? Every time you add a new class, you
> have to update your Event class to deal with a new "kind" enumeration.
>
> So, with your current design, when you derive a class from Event, you
> also have to modify the Event class. Not good design.
>
> Ditch the enum list in Event.
>
> KPB
>[/color]

OK. So what desing would be the best in your opinion? How would you
solve that problem?

**KPB** · Jul 22 '05, 11:15 PM

Re: a problem with poliporphism

Nafai wrote:[color=blue]
> KPB escribió:
>[color=green]
>> Nafai wrote:
>>[color=darkred]
>>> Do you think it is a good a idea to do this:
>>> void doSomethingWith Events(Event* pE)
>>> {
>>> Emergency* pEmergency;
>>> switch(pE->kind) {
>>> case Event::Emergenc y :
>>> pEmergency=dyna mic_cast<Emerge ncy*>(pE);
>>> callAmbulance(p Emergency->where(), pEmergency->when());
>>> break;
>>> ... // and so on
>>> }
>>> }[/color]
>>
>>
>>
>> No. That leads me to another point on your classes. What's the point
>> in having polimorphism if the specific kind of event is stored as some
>> internal value of the Event class? Every time you add a new class, you
>> have to update your Event class to deal with a new "kind" enumeration.
>>
>> So, with your current design, when you derive a class from Event, you
>> also have to modify the Event class. Not good design.
>>
>> Ditch the enum list in Event.
>>
>> KPB
>>[/color]
>
> OK. So what desing would be the best in your opinion? How would you
> solve that problem?[/color]

I've told you quite a bit about what I'd do.

I'm sorry but I think you need to study the topic of inheritence a
little better.

KPB

**Nafai** · Jul 22 '05, 11:15 PM

Re: a problem with poliporphism

KPB escribió:[color=blue]
> Nafai wrote:
>[color=green]
>> KPB escribió:
>>[color=darkred]
>>> Nafai wrote:
>>>
>>>> Do you think it is a good a idea to do this:
>>>> void doSomethingWith Events(Event* pE)
>>>> {
>>>> Emergency* pEmergency;
>>>> switch(pE->kind) {
>>>> case Event::Emergenc y :
>>>> pEmergency=dyna mic_cast<Emerge ncy*>(pE);
>>>> callAmbulance(p Emergency->where(), pEmergency->when());
>>>> break;
>>>> ... // and so on
>>>> }
>>>> }
>>>
>>>
>>>
>>>
>>> No. That leads me to another point on your classes. What's the point
>>> in having polimorphism if the specific kind of event is stored as
>>> some internal value of the Event class? Every time you add a new
>>> class, you have to update your Event class to deal with a new "kind"
>>> enumeration.
>>>
>>> So, with your current design, when you derive a class from Event, you
>>> also have to modify the Event class. Not good design.
>>>
>>> Ditch the enum list in Event.
>>>
>>> KPB
>>>[/color]
>>
>> OK. So what desing would be the best in your opinion? How would you
>> solve that problem?[/color]
>
>
>
> I've told you quite a bit about what I'd do.
>
> I'm sorry but I think you need to study the topic of inheritence a
> little better.
>
> KPB[/color]

Just tell me an outline of your solution. I know what it is polymorphism
and what is inheritance. Please, just tell me your solution. If you have it.

**Jeff Flinn** · Jul 22 '05, 11:15 PM

Re: a problem with poliporphism

Nafai wrote:[color=blue]
> KPB escribió:[color=green]
>> Nafai wrote:[color=darkred]
>>> KPB escribió:
>>>> Nafai wrote:[/color][/color][/color]

....
[color=blue]
> Just tell me an outline of your solution. I know what it is
> polymorphism and what is inheritance. Please, just tell me your
> solution. If you have it.[/color]

You've been told a few times by a few people already. Give it your best
shot, post your try here and you'll get some help.

Jeff

**Nafai** · Jul 22 '05, 11:15 PM

Re: a problem with poliporphism

"Jonathan Mcdougall" <jonathanmcdoug all@DELyahoo.ca > escribió en el mensaje
news:UOWAd.33$a z2.83455@weber. videotron.net.. .[color=blue]
> Nafai wrote:[color=green]
>> I am going to give a more detailed example:
>>
>> // I don't type the constructors.
>>
>> class Event {
>> private:
>> int time;
>> EventType kind;
>> public:
>> enum EventType { Event1, Event2,...,Emer gency,Crime};
>> int when() {return time; }
>> EventType what() { return kind; }[/color]
>
> Event is meant to be used as a base class and does not have a virtual
> destructor nor virtual functions. What kind of base class is that?
>[color=green]
>> }
>>
>> class Emergency : public Event {
>> private:
>> int area;
>> public:
>> int where() {return area;}
>> };
>>
>> class Crime : public Event {
>> private:
>> string person;
>> int aCrime;.
>> public:
>> string who() { return person; }
>> int whatHeDid() { return aCrime; }
>> };
> >
> >
>> void doSomethingWith Events(Event* pE)
>> {
>> switch(pE->kind) {
>> case Event::Emergenc y : callAmbulance(p E->where(),pE->when());
>> break;
>> case Event::Crime: blame(pE->who(),pE->whatHeDid(), pE->when());
>> break;
>> case Event::Event1 : doSomethingRela tedToEvent1(pE->when());
>> break;
>> ...
>> }
>> }[/color]
>
> That is not recommended. You seem not to understand the use of
> polymorphism and public inheritance.
>
> Public inheritance has many uses, but in your case, it is a simple case of
> interface<->behavior. The base class specifies an interface (with [pure]
> virtual functions) and derived classes specify the behavior.
>
> class Event
> {
> public:
> virtual ~Event();
>
> virtual void process() = 0; // that's the
> // interface
> };
>
> class Emergency : public Event
> {
> private:
> void call_ambulance( );
>
> public:
> virtual void process() // that's a behavior
> {
> call_ambulance( );
> }
> };
>
> class Crime : public Event
> {
> private:
> void blame();
>
> public:
> virtual void process()
> {
> blame(); // that's another behavior
> }
> };
>
>
> typedef std::sector<Eve nt*> Events;
>
> void f(Events &events)
> {
> for (Events::iterat or itor=events.beg in();
> itor!=events.en d();
> ++itor)
> {
> Event *e = *itor;
> e->process(); // calls the correct
> // process()
> }
> }
>
>
> Jonathan[/color]

That's OK, but I can't do that. If I could of course I would have. The
"process()" is done by another class. That is: I DO NEED to take "area",
"person" etc with where(), who() etc. There's no way to change this.

**Jonathan Mcdougall** · Jul 22 '05, 11:15 PM

Re: a problem with poliporphism

Nafai wrote:[color=blue]
>
> That's OK, but I can't do that. If I could of course I would have. The
> "process()" is done by another class. That is: I DO NEED to take "area",
> "person" etc with where(), who() etc. There's no way to change this.[/color]

If you can't have virtual functions, you need to
emulate them either by using an id-switch (as you
did) or a type switch with dynamic_cast.

void f(Event *event)
{
if ( Emergency *e =
dynamic_cast<Em ergency*>(event ) )
{
}
else if ( Crime *c =
dynamic_cast<Cr ime*>(event) )
{
}
}

But that means f() must know about all the
subtypes and must test from down to top of the
hierarchy. That's a pain and it's ugly, but you
may have no choice.

Jonathan

**Mike Hewson** · Jul 22 '05, 11:15 PM

Re: a problem with poliporphism [OT]

Victor Bazarov wrote:[color=blue]
> "E. Robert Tisdale" <E.Robert.Tisda le@jpl.nasa.gov > wrote...
>[color=green]
>>Nafai wrote:
>>
>>[snip]
>>
>>What is poliporphism?[/color]
>
>
> Pinch your nose and try saying "polyMorphi sm". Whaddya get?
>
>[/color]
For me, alas, a fart. :-)
Seriously, it's a cross between a parrot widget and a dolphin thingy.

--

Cheers
--
Hewson::Mike
"This letter is longer than usual because I lack the time to make it
shorter" - Blaise Pascal

**Howard** · Jul 22 '05, 11:20 PM

Re: a problem with poliporphism

"Nafai" <nafai3000@yaho o.es> wrote in message
news:D81Bd.1003 23$oN1.169983@t elenews.telelin e.es...
[color=blue][color=green]
>> class Event
>> {
>> public:
>> virtual ~Event();
>>
>> virtual void process() = 0; // that's the
>> // interface
>> };
>>
>> class Emergency : public Event
>> {
>> private:
>> void call_ambulance( );
>>
>> public:
>> virtual void process() // that's a behavior
>> {
>> call_ambulance( );
>> }
>> };
>>
>> class Crime : public Event
>> {
>> private:
>> void blame();
>>
>> public:
>> virtual void process()
>> {
>> blame(); // that's another behavior
>> }
>> };
>>
>>
>> typedef std::sector<Eve nt*> Events;
>>
>> void f(Events &events)
>> {
>> for (Events::iterat or itor=events.beg in();
>> itor!=events.en d();
>> ++itor)
>> {
>> Event *e = *itor;
>> e->process(); // calls the correct
>> // process()
>> }
>> }
>>
>>
>> Jonathan[/color]
>
>
> That's OK, but I can't do that. If I could of course I would have. The
> "process()" is done by another class. That is: I DO NEED to take "area",
> "person" etc with where(), who() etc. There's no way to change this.[/color]

How about passing a pointer to the instance of the class that will handle
the process() call? Or, perhaps better, make the pointer a member of the
Event class, and initialize it when creating the Event objects? Like
this...

// forward declared, so pointer can be used here
class ProcessClass;

class Event
{
ProcessClass* pProcessor;

public:
Event( ProcessClass* theProcessor ) : pProcessor(theP rocessor);

virtual ~Event();

virtual void process() = 0; // that's the
// interface
};

class Emergency : public Event
{
public:
virtual void process() // that's a behavior
{
pProcessor->call_ambulance (where(),when() );
}
};

class Crime : public Event
{
public:
virtual void process()
{
pProcessor->blame(who(),wh atHeDid(),when( )); // that's
another behavior
}
};

typedef std::sector<Eve nt*> Events;

void f(Events &events)
{
for (Events::iterat or itor=events.beg in();
itor!=events.en d();
++itor)
{
Event *e = *itor;
e->process(); // calls the correct
// process()
}
}

I haven't compiled this, just modified what Jonathon wrote, and I haven't
added back in the functions who(), where(), etc., but you can do that
yourself. Does this concept make sense now?

-Howard

a problem with poliporphism

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment