Home » Community » Newbie corner » Questions about static casting Polymorphic Array Elements, iterator, Ptr and Pte
 Questions about static casting Polymorphic Array Elements, iterator, Ptr and Pte [message #38561] |
Mon, 24 December 2012 01:58  |
navi
Messages: 107
Registered: February 2012
Location: Sydney, Australia
|
Experienced Member |
|
|
struct shape{
int type;
}
struct circle : shape{
circle(){ type=1; }
int radius;
int x, y;
}
struct triangle : shape{
triangle(){ type=2; }
int x[3], y[3];
}
struct rectangle : shape{
rectangle(){ type=3; }
int x[4],y[4];
}
Array<shape> a;
a.Add(new circle);
a.Add(new triangle);
a.Add(new rectangle);
if(a[1].type==2){
triangle *m = static_cast<triangle *> (&a[1]);
}
in the above example, is this the correct Syntax & correct way to static casting Polymorphic Array Elements?triangle *m = static_cast<triangle *> (&a[1]);
How do I create an iterator and static cast the iterator instead? Is this a better way then the previous method?
What is the correct way to static cast an Element from Polymorphic Array?
As suggested in NTL Tutorial - Point.6 Polymorpic Array is great for the purpose of storing Polymorphic objects. no problem of object slicing. No need for memory management. So, the below question is purely for the curiosity sake.
Are there smart pointers in U++? "Stack Overflow" and other forums people suggesting to use STL vector of smart pointers to keep Polymorphic Elements. Is there any way to achieve that using U++ NTL? I have seen Ptr and Pte in the Manuel but confuse about what do they actually do? do they only assign null when object is destroyed or do they actually destroy pointing object (i.e. manages the object De-Allocation?) when they go out of scope?
Thanks & Regards
Navi
[Updated on: Mon, 24 December 2012 02:00] Report message to a moderator |
|
|
|
| Re: Questions about static casting Polymorphic Array Elements, iterator, Ptr and Pte [message #38562 is a reply to message #38561] |
Mon, 24 December 2012 09:50  |
 |
mirek
Messages: 13975
Registered: November 2005
|
Ultimate Member |
|
|
| navi wrote on Sun, 23 December 2012 19:58 |
struct shape{
int type;
}
struct circle : shape{
circle(){ type=1; }
int radius;
int x, y;
}
struct triangle : shape{
triangle(){ type=2; }
int x[3], y[3];
}
struct rectangle : shape{
rectangle(){ type=3; }
int x[4],y[4];
}
Array<shape> a;
a.Add(new circle);
a.Add(new triangle);
a.Add(new rectangle);
|
Event better (more "U++ish") is to use
circle& c = a.Create<circle>();
here.
| Quote: |
in the above example, is this the correct Syntax & correct way to static casting Polymorphic Array Elements?triangle *m = static_cast<triangle *> (&a[1]);
|
Yep. Although, IME, if you use the trick above, you in fact seldom need to cast later. I guess that you only need to know the final type to initialize it, rest can be taken care about with virtual methods.
| Quote: |
How do I create an iterator and static cast the iterator instead? Is this a better way then the previous method?
|
Well, U++ way is to prefer index based iteration over iterator. Anyway, if you insist, you can of course use iterator.
#include <Core/Core.h>
using namespace Upp;
struct Foo {
int foo;
};
struct Bar : Foo {
int bar;
};
CONSOLE_APP_MAIN
{
Array<Foo> x;
x.Create<Bar>().bar = 54321;
for(Array<Foo>::Iterator it = x.Begin(); it != x.End(); ++it)
DUMP(static_cast<Bar &>(*it).bar);
}
| Quote: |
Are there smart pointers in U++?
|
Well, there are some historical in 'non-canonical' parts of U++ that are not normally part of U++ releases, anyway, shared smart pointers are generally considered "BIG EVIL", something to avoid.
| Quote: |
I have seen Ptr and Pte in the Manuel but confuse about what do they actually do? do they only assign null when object is destroyed
|
Yes. Thing is, the U++ way nicely solves most issues about resource management, so that you never call 'delete' in high-level code, but the solution has somewhat weak spot that you have to be careful about dangling pointers. In some cases, Ptr is a good tool to deal with them...
| Quote: |
or do they actually destroy pointing object (i.e. manages the object De-Allocation?) when they go out of scope?
|
No.
Mirek
|
|
|
|
| Re: Questions about static casting Polymorphic Array Elements, iterator, Ptr and Pte [message #38563 is a reply to message #38562] |
Mon, 24 December 2012 10:57 |
navi
Messages: 107
Registered: February 2012
Location: Sydney, Australia
|
Experienced Member |
|
|
Dear Mirek,
First of all, Thank you for the quick response.
| mirek wrote on Mon, 24 December 2012 09:50 |
Event better (more "U++ish") is to use
circle& c = a.Create<circle>();
here.
| Quote: |
in the above example, is this the correct Syntax & correct way to static casting Polymorphic Array Elements?triangle *m = static_cast<triangle *> (&a[1]);
|
Yep. Although, IME, if you use the trick above, you in fact seldom need to cast later. I guess that you only need to know the final type to initialize it, rest can be taken care about with virtual methods.
|
I will use the U++ method Create<type>() when I first create and add element into the array.
I am casting element at a later time depending on the "type" variable which can be access via the base type pointer/referance returned by the Array's index operator, to access the derived part of the object. I am under impression that since the Array is of the base type, I cant access the derived part of the object using the operator[] without casting. As it might return a reference of the base type. does Array return a derived type reference if added using Create<type>()? does Array actually remembers type for individual elements?! if so, then its awesome! I don't need casting what so ever at later time.
P.S. I do not have any virtual method in my objects other then the virtual destructor. which is also empty at the point. I mainly need to retrieve and set various public variables exposed by the different kind of objects. for instance circle only has 3 separate int variable, where triangle has an int array of size 3. some objects might even have other different type of variable like String, Date and etc.
| mirek wrote on Mon, 24 December 2012 09:50 |
...anyway, shared smart pointers are generally considered "BIG EVIL", something to avoid....
...Yes. Thing is, the U++ way nicely solves most issues about resource management, so that you never call 'delete' in high-level code, but the solution has somewhat weak spot that you have to be careful about dangling pointers. In some cases, Ptr is a good tool to deal with them...
|
Understood. pte/ptr is NOT Equal to "shared smart pointers"
Thanks & Regards
Navi
|
|
|
|
| Re: Questions about static casting Polymorphic Array Elements, iterator, Ptr and Pte [message #38566 is a reply to message #38563] |
Mon, 24 December 2012 14:40 |
|
mirek
Messages: 13975
Registered: November 2005
|
Ultimate Member |
|
|
| navi wrote on Mon, 24 December 2012 04:57 |
I am casting element at a later time depending on the "type" variable which can be access via the base type pointer/referance returned by the Array's index operator, to access the derived part of the object. I am under impression that since the Array is of the base type, I cant access the derived part of the object using the operator[] without casting.
|
Correct, but virtual methods of that type CAN.
Of course, it is not an universal truth and sometimes casting you suggest is really needed, but quite often you can better handle issues by good design of base class interface, with the right set of virtual methods doing things.
All in all, the whole purpose of such design is to handle all subclasses the same way...
To illustrate, say that your classes are intended for some graphical designer and that you want a menu with a set of operations for each type. Then you can do something like
if(a[1].type==2){
triangle *m = static_cast<triangle *> (&a[1]);
CreateMenuForTriangle(m);
}
or you can just use
struct shape {
virtual void Menu(Bar& bar);
}
Mirek
|
|
|
|
|
|
Goto Forum:
Current Time: Sat Apr 27 22:30:49 CEST 2024
Total time taken to generate the page: 0.03683 seconds
|
Members
Pages
Help
Register
Login
Home
