iterator problem

[Chameleon]

All these happen in VC++ NET 2003, if you ask me about the compiler.
it._Myptr is a VC++ specific member of the iterator implementation.
But from this, we extract usefull informations about crash.
----------------------------
vector<int*> v;
vector<int*>::iterator it = v.begin(); // it._Myptr == 0 (what pointer
is this?)
v.insert(it, 5); // works
it++; // it._Myptr == 4
v.insert(it, 5); // crashes
----------------------------
and this.
----------------------------
vector<int*> v;
vector<int*>::iterator it = v.begin(); // it._Myptr == 0
v.push_back(5);
vector<int*>::iterator it = v.begin(); // it._Myptr == 0x00323b40 (the
pointer to first element)
----------------------------
All of these are compiler bugs on iterator implementation, or I miss
something?

Thanks

[Calum Grant]

Chameleon wrote:

All these happen in VC++ NET 2003, if you ask me about the compiler.
it._Myptr is a VC++ specific member of the iterator implementation.
But from this, we extract usefull informations about crash.
----------------------------
vector<int*> v;
vector<int*>::iterator it = v.begin(); // it._Myptr == 0 (what
pointer is this?)
The same as v.end(). No bug. You're not going to dereference it.
v.insert(it, 5); // works
That's okay, but invalidates "it".
it++; // it._Myptr == 4
Unfortunately, "it" is invalid.
v.insert(it, 5); // crashes
Yup.
----------------------------
and this.
----------------------------
vector<int*> v;
vector<int*>::iterator it = v.begin(); // it._Myptr == 0
v.push_back(5);
This compiles?
vector<int*>::iterator it = v.begin(); // it._Myptr == 0x00323b40
(the pointer to first element)
----------------------------
All of these are compiler bugs on iterator implementation, or I miss
something?

Ah, not every error is a compiler error!

Cheers, Calum

[Daniel T.]

In article <ds**********@volcano1.grnet.gr>,
Chameleon <ch******@hotmail.NOSPAM.com> wrote:

All of these are compiler bugs on iterator implementation, or I miss
something?

Until you know enough that you can build your own compiler, go ahead and
assume that the bug is in your code rather than the compiler's code...
Just a thought.
--
Magic depends on tradition and belief. It does not welcome observation,
nor does it profit by experiment. On the other hand, science is based
on experience; it is open to correction by observation and experiment.

[Chameleon]

Calum Grant wrote:

Chameleon wrote:

All these happen in VC++ NET 2003, if you ask me about the compiler.
it._Myptr is a VC++ specific member of the iterator implementation.
But from this, we extract usefull informations about crash.
----------------------------
vector<int*> v;
vector<int*>::iterator it = v.begin(); // it._Myptr == 0 (what
pointer is this?)

The same as v.end(). No bug. You're not going to dereference it.

v.insert(it, 5); // works

That's okay, but invalidates "it".

it++; // it._Myptr == 4

Unfortunately, "it" is invalid.

v.insert(it, 5); // crashes

Yup.

----------------------------
and this.
----------------------------
vector<int*> v;
vector<int*>::iterator it = v.begin(); // it._Myptr == 0
v.push_back(5);

This compiles?

sorry, vector<int> v;

vector<int*>::iterator it = v.begin(); // it._Myptr == 0x00323b40
(the pointer to first element)
----------------------------
All of these are compiler bugs on iterator implementation, or I miss
something?

Ah, not every error is a compiler error!

of-course!

my solution is this until now:

-------------------------------------
vector<int*> v;
vector<int*>::iterator it = v.begin();
it = v.insert(it, 5);
it++;
it = v.insert(it, 5);
-------------------------------------

[Luke Meyers]

Chameleon wrote:

vector<int*> v;
vector<int*>::iterator it = v.begin(); // it._Myptr == 0
v.push_back(5);
This compiles?

sorry, vector<int> v;

Okay, vector<int>. Good.
my solution is this until now:

-------------------------------------
vector<int*> v;
vector<int*>::iterator it = v.begin();
it = v.insert(it, 5);
it++;
it = v.insert(it, 5);
-------------------------------------

Back to vector<int*> now? Well, in any case... line 2 above serves no
purpose. I would change this snippet to:

std::vector<int> v;
v.push_back(5);
v.push_back(5);

See? No need for iterators at all in this case. Maybe you got
confused because you didn't know about push_back? You owe it to
yourself to learn the public interface of important classes like
std::vector. It is only wafer-thin.

Luke