ForumSections

Errors, bugs, questions - page 2296

New comment
A100:

There is no problem... there is an error. I also don't understand why you're trying to pull the explanation out of your head. The default Print has the following signature:

And it can also clash with other functions (if needed)

The Print signature is out of range of MQL5 language. If MQL5 prohibits Print overloading at all, it won't be a bug.

I hope this "bug" does not prevent you from creating TS and does not require writing crutches.

I measured the performance of the Tester. To do this, I opened and closed a position on each tick. I measured the time of execution of 100 ticks. I ran it through 100 000 ticks. I measured a total of 1000 ticks. I got this chart

There are spikes of slowing down at almost equal distances between them. I measured it in the Optimization mode. If I ignore those spikes the tester's performance varies by 25% instead of the constant value. Maybe it's Windows speed metering tricks again.


SZZ The same code I ran on MT4

On average it takes MT4 Tester 1.5-2 times less time to process one tick than MT5.

Files:
Tester.mq5  2 kb

Possibly surges through the use of a genetic optimisation algorithm.


... On every tick opened and closed a position. <br / translate="no">
On average it takes MT4-Tester 1.5-2 times less time to process one tick than MT5.
And this is pure manipulation and misleading.
Sergey Dzyublik:

Possibly surges through the use of a genetic optimisation algorithm.

No GA. Optimisation from two passes on one Agent.

And this is already pure manipulation and misleading.

The source code is in place.

In one case a warning, in the other an error

void f()
{
    for ( int i = 0;; )
    {
        Print( i );
        int    i = 5; //Warning: declaration of 'i' hides local declaration
        Print( i );
        break;
    }
}
void For( int i = 0 )
    {
        Print( i );
        int    i = 5; //Error: redefinition of formal parameter 'i'
        Print( i );
    }

What's the fundamental difference? In C++, for example, there is an error in both cases

I'm asking for help from knowledgeable people to understand the issue of pointers to class instances. I do not understand it.

Here is an example of the script:

class A
  {
public :
                     A() { Print("Start"); };
                    ~A() { 
                     Print("End"); 
                     Print(EnumToString(CheckPointer(GetPointer(this)))); 
                     if (CheckPointer(GetPointer(this))!=POINTER_DYNAMIC) 
                     delete GetPointer(this);};
  };
  
A a;

void OnStart()
  {
  }

When executing it we have, as expected:

2018.09.23 21:56:20.574 Test_CheckPointer (EURUSD,M1)   Start
2018.09.23 21:56:20.574 Test_CheckPointer (EURUSD,M1)   End
2018.09.23 21:56:20.574 Test_CheckPointer (EURUSD,M1)   POINTER_AUTOMATIC


if an instance of a class is declared as:

A *a= new A;

then on execution we have:

2018.09.23 21:46:42.960 Test_CheckPointer (EURUSD,M1)   Start
2018.09.23 21:46:42.961 Test_CheckPointer (EURUSD,M1)   1 undeleted objects left
2018.09.23 21:46:42.961 Test_CheckPointer (EURUSD,M1)   1 object of type A left
2018.09.23 21:46:42.961 Test_CheckPointer (EURUSD,M1)   32 bytes of leaked memory

i.e. the destructor is not even started and therefore memory is not freed.


But if an instance of a class is declared as:

A a= new A;

the constructor is launched twice, the destructor - once, but memory is not freed and we have thePOINTER_AUTOMATIC object pointer type, although it was intended to bePOINTER_DYNAMIC

2018.09.23 21:54:24.844 Test_CheckPointer (EURUSD,M1)   Start
2018.09.23 21:54:24.844 Test_CheckPointer (EURUSD,M1)   Start
2018.09.23 21:54:24.844 Test_CheckPointer (EURUSD,M1)   End
2018.09.23 21:54:24.844 Test_CheckPointer (EURUSD,M1)   POINTER_AUTOMATIC
2018.09.23 21:54:24.844 Test_CheckPointer (EURUSD,M1)   1 undeleted objects left
2018.09.23 21:54:24.845 Test_CheckPointer (EURUSD,M1)   1 object of type A left
2018.09.23 21:54:24.845 Test_CheckPointer (EURUSD,M1)   32 bytes of leaked memory


How to always execute the destructor and correctly execute delete

Как добиться всегда выполнения деструктора и правильного выполнения delete.


See from C++ the topic of smart pointers and adapt for MQL(https://habr.com/post/140222/).
Maybe there's something in kodobase...

Sergey Dzyublik:


See from C++ the topic of smart pointers and adapt for MQL(https://habr.com/post/140222/).
Maybe there's something in kodobase...

thanks, but didn't see any answers to my questions there.
I don't understand why the destructor is not called whenA *a= new A;

Nikolai Semko:

Thank you, but I didn't see any answers to my questions there.
I don't understand why the destructor is not called whenA *a= new A;

Then try it like this:
class A { public:
         A() { Print( 1, ":", EnumToString(CheckPointer(GetPointer(this)))); }
        ~A() { Print( 2, ":", EnumToString(CheckPointer(GetPointer(this)))); }
};
class B { public:
         B( void *b ) : b( b ) {}
        ~B() { delete b; }
        void *b;
};
A a1;
A *a2 = new A;
B b( a2 );
void OnStart() {}

Result:

1:POINTER_AUTOMATIC
1:POINTER_DYNAMIC
2:POINTER_DYNAMIC
2:POINTER_AUTOMATIC

Nikolai Semko:

Thank you, but I didn't see any answers to my questions there.
I don't understand why destructor is not called whenA *a= new A;

Create with new and delete with delete

1...228922902291229222932294229522962297229822992300230123022303...3193
New comment