Features of the mql5 language, subtleties and tricks - page 235

 
fxsaber #:

If it stopped working, it's good to know if it's the right thing to do.

If you make pointers instead of objects, the old version will work too.

Great point and thanks for the tip!

Yes, indeed, the situation is perfectly solved with a pointer:

class A;

class B
{
   public: A * a;
   public: int Val;
};

class A
{
   public: B * b;
   public: int Test() {return b.Val + 1;}
};

//+------------------------------------------------------------------+
//|                                                                                |
//+------------------------------------------------------------------+
void OnStart()
{
   B b;
   A a;
   
   b.a = GetPointer(a);
   b.a.b = GetPointer(b);
   b.Val = 1;
   
   Print(b.a.Test());
}
 

Fans of fast algorithms. Those who fight for nanoseconds :)


Task: Find the time of bar opening according to the given time and TF, when it is known that the bar exists at this time. For example, by the time of opening and closing positions.

Most programmers will use a combination of iTime and iBarShift. This will be the slowest implementation, especially such an implementation requires an up-to-date history of uploaded data or combed arrays. Moreover, this approach may generate errors if the required history is missing.

More advanced programmers will solve this problem through MqlDateTime structure and TimeToStruct() function. This is not a bad solution and fast enough.

But there is a third solution, which is more productive than the previous solution several times:

//+------------------------------------------------------------------+
// получает время открытия виртуального бара по входному времени и Таймфрейму, вне зависимости от того, существует реальный бар или нет.
// корректно считает только до 28.02.2100 !!!!
// не является заменой iBarShift!!! Не зависит от истории баров.  
datetime getStartTimeOfBarFast(ENUM_TIMEFRAMES tf, datetime t) {
   if (tf==0) tf=_Period;

   int ts=0;
   if (tf<PERIOD_MN1) {
      ushort i_tf= ushort(tf);
      uchar _i =uchar(i_tf>>14);
      int n = i_tf & 0x0FFF;
      ts = (_i==0)?n*60:(_i==1)?n*60*60:60*60*24*7;
   }
   if (tf<PERIOD_W1) return t-t%ts;
   if (tf==PERIOD_W1) return t-(t+4*24*60*60)%ts;
   else { // Period MN1
      static int dm[12] = {0,31,61,92,122,153,184, 214, 245, 275, 306, 337};
      static int last_days = 0;
      static datetime last_result = 0;
      int days = int(t/(24*60*60));
      if (last_days!=days) {
         last_days = days;
         int d1 = (days+306+365)%1461;
         int y = d1/365;
         datetime t1 = t - t%(24*60*60) - d1*24*60*60;
         int m = 0;
         if (d1==1460) {
            m=11;
            y--;
         };
         int d = d1-y*365+1;
         if (d!=31) if (d==276) m = 9;
            else m = int (d/30.68);
         if (m<0 || m>11) return -1;
         last_result = t1+y*365*24*60*60+dm[m]*24*60*60;
      }
      return last_result;
   }
}
//+------------------------------------------------------------------+

The main difficulty in this algorithm is calculating the time of the beginning of the month (highlighted in green) Please don't try to understand the workings of the algorithm. There is magic going on there, which is the result of going from simple to complex. The reverse path from complex to simple will be much harder to go through.

The performance gain is also provided by the algorithm of getting seconds in a bar from the TF instead of the standard PeriodSeconds() function - highlighted in yellow colour.


I attach a test script that calculates and compares the performance of all three methods:

2023.11.14 12:15:29.145 timeToStartMonth (EURUSD,M1)    =====LOOP=10000========STEPS=100000 seconds======PERIOD_MN1========
2023.11.14 12:15:29.146 timeToStartMonth (EURUSD,M1)    контрольная сумма - 11987239507200, время выполнения 1 иттерации = 21.20 наносекунд - Расчет через структуру MqlDateTime
2023.11.14 12:15:29.146 timeToStartMonth (EURUSD,M1)    контрольная сумма - 11987239507200, время выполнения 1 иттерации = 6.10 наносекунд - Быстрый расчет
2023.11.14 12:15:29.147 timeToStartMonth (EURUSD,M1)    контрольная сумма - 11987239507200, время выполнения 1 иттерации = 142.00 наносекунд - Расчет через iBarShift
2023.11.14 12:15:29.147 timeToStartMonth (EURUSD,M1)    ========================================================================
2023.11.14 12:15:34.226 timeToStartMonth (EURUSD,M1)    =====LOOP=10000========STEPS=100000 seconds======PERIOD_MN1========
2023.11.14 12:15:34.227 timeToStartMonth (EURUSD,M1)    контрольная сумма - 11987239507200, время выполнения 1 иттерации = 19.80 наносекунд - Расчет через структуру MqlDateTime
2023.11.14 12:15:34.227 timeToStartMonth (EURUSD,M1)    контрольная сумма - 11987239507200, время выполнения 1 иттерации = 6.70 наносекунд - Быстрый расчет
2023.11.14 12:15:34.228 timeToStartMonth (EURUSD,M1)    контрольная сумма - 11987239507200, время выполнения 1 иттерации = 127.50 наносекунд - Расчет через iBarShift
2023.11.14 12:15:34.228 timeToStartMonth (EURUSD,M1)    ========================================================================
2023.11.14 12:15:39.856 timeToStartMonth (EURUSD,M1)    =====LOOP=10000========STEPS=100000 seconds======PERIOD_W1========
2023.11.14 12:15:39.856 timeToStartMonth (EURUSD,M1)    контрольная сумма - 11997367833600, время выполнения 1 иттерации = 2.80 наносекунд - Расчет через структуру MqlDateTime
2023.11.14 12:15:39.856 timeToStartMonth (EURUSD,M1)    контрольная сумма - 11997367833600, время выполнения 1 иттерации = 1.50 наносекунд - Быстрый расчет
2023.11.14 12:15:39.857 timeToStartMonth (EURUSD,M1)    контрольная сумма - 11997367833600, время выполнения 1 иттерации = 98.30 наносекунд - Расчет через iBarShift
2023.11.14 12:15:39.857 timeToStartMonth (EURUSD,M1)    ========================================================================
2023.11.14 12:15:52.770 timeToStartMonth (EURUSD,M1)    =====LOOP=10000========STEPS=100000 seconds======PERIOD_H2========
2023.11.14 12:15:52.771 timeToStartMonth (EURUSD,M1)    контрольная сумма - 12000355999200, время выполнения 1 иттерации = 4.10 наносекунд - Расчет через структуру MqlDateTime
2023.11.14 12:15:52.771 timeToStartMonth (EURUSD,M1)    контрольная сумма - 12000355999200, время выполнения 1 иттерации = 1.50 наносекунд - Быстрый расчет
2023.11.14 12:15:54.255 timeToStartMonth (EURUSD,M1)    контрольная сумма - 0, время выполнения 1 иттерации = 148466.50 наносекунд - Расчет через iBarShift
2023.11.14 12:15:54.255 timeToStartMonth (EURUSD,M1)    ========================================================================
2023.11.14 12:15:58.759 timeToStartMonth (EURUSD,M1)    =====LOOP=10000========STEPS=100000 seconds======PERIOD_M4========
2023.11.14 12:15:58.759 timeToStartMonth (EURUSD,M1)    контрольная сумма - 12000391999920, время выполнения 1 иттерации = 3.60 наносекунд - Расчет через структуру MqlDateTime
2023.11.14 12:15:58.759 timeToStartMonth (EURUSD,M1)    контрольная сумма - 12000391999920, время выполнения 1 иттерации = 1.50 наносекунд - Быстрый расчет
2023.11.14 12:15:59.864 timeToStartMonth (EURUSD,M1)    контрольная сумма - 12000017286960, время выполнения 1 иттерации = 110555.70 наносекунд - Расчет через iBarShift
2023.11.14 12:15:59.864 timeToStartMonth (EURUSD,M1)    ========================================================================

The checksum with iBarShift will not match , because iBarShift works with real bars. The checksum will coincide only on MN1 and W1 timeframes, because there are no holes in the history of such bars.
Performance will be higher if you use a small time step in the loop (less than one day) when the algorithm starts working to save previous calculations:

2023.11.14 12:14:10.714 timeToStartMonth (EURUSD,M1)    =====LOOP=1000000========STEPS=1000 seconds======PERIOD_MN1========
2023.11.14 12:14:10.722 timeToStartMonth (EURUSD,M1)    контрольная сумма - 1198674131961600, время выполнения 1 иттерации = 8.03 наносекунд - Расчет через структуру MqlDateTime
2023.11.14 12:14:10.723 timeToStartMonth (EURUSD,M1)    контрольная сумма - 1198674131961600, время выполнения 1 иттерации = 1.18 наносекунд - Быстрый расчет
2023.11.14 12:14:10.860 timeToStartMonth (EURUSD,M1)    контрольная сумма - 1198674131961600, время выполнения 1 иттерации = 136.80 наносекунд - Расчет через iBarShift
2023.11.14 12:14:10.860 timeToStartMonth (EURUSD,M1)    ========================================================================
2023.11.14 12:14:17.502 timeToStartMonth (EURUSD,M1)    =====LOOP=1000000========STEPS=1000 seconds======PERIOD_MN1========
2023.11.14 12:14:17.510 timeToStartMonth (EURUSD,M1)    контрольная сумма - 1198674131961600, время выполнения 1 иттерации = 7.70 наносекунд - Расчет через структуру MqlDateTime
2023.11.14 12:14:17.511 timeToStartMonth (EURUSD,M1)    контрольная сумма - 1198674131961600, время выполнения 1 иттерации = 1.18 наносекунд - Быстрый расчет
2023.11.14 12:14:17.648 timeToStartMonth (EURUSD,M1)    контрольная сумма - 1198674131961600, время выполнения 1 иттерации = 137.54 наносекунд - Расчет через iBarShift
2023.11.14 12:14:17.648 timeToStartMonth (EURUSD,M1)    ========================================================================


Excessive values for the algorithm via iBarShift (highlighted in blue) are caused by the current lack of necessary history or TFs calculated by the array, which initiates their uploading.
After uploading the result will be as follows:

2023.11.14 12:47:06.158 timeToStartMonth (EURUSD,M1)    =====LOOP=10000========STEPS=100000 seconds======PERIOD_H2========
2023.11.14 12:47:06.158 timeToStartMonth (EURUSD,M1)    контрольная сумма - 12000379996800, время выполнения 1 иттерации = 4.60 наносекунд - Расчет через структуру MqlDateTime
2023.11.14 12:47:06.158 timeToStartMonth (EURUSD,M1)    контрольная сумма - 12000379996800, время выполнения 1 иттерации = 2.60 наносекунд - Быстрый расчет
2023.11.14 12:47:06.159 timeToStartMonth (EURUSD,M1)    контрольная сумма - 12000009103200, время выполнения 1 иттерации = 129.10 наносекунд - Расчет через iBarShift
2023.11.14 12:47:06.159 timeToStartMonth (EURUSD,M1)    ========================================================================
2023.11.14 12:47:13.899 timeToStartMonth (EURUSD,M1)    =====LOOP=10000========STEPS=100000 seconds======PERIOD_M4========
2023.11.14 12:47:13.899 timeToStartMonth (EURUSD,M1)    контрольная сумма - 12000411199920, время выполнения 1 иттерации = 2.80 наносекунд - Расчет через структуру MqlDateTime
2023.11.14 12:47:13.899 timeToStartMonth (EURUSD,M1)    контрольная сумма - 12000411199920, время выполнения 1 иттерации = 1.40 наносекунд - Быстрый расчет
2023.11.14 12:47:13.903 timeToStartMonth (EURUSD,M1)    контрольная сумма - 12000038634480, время выполнения 1 иттерации = 381.30 наносекунд - Расчет через iBarShift
2023.11.14 12:47:13.903 timeToStartMonth (EURUSD,M1)    ========================================================================

Files:
 
Nikolai Semko #:

Fans of fast algorithms. Those who fight for nanoseconds :)

...

😮😲😳🥴🤪

...

ah...

mmm....

oooh....

gkghm... I didn't think my simple question would come out like this.

Just like that. Oh.

 
Artyom Trishkin #:

😮😲😳🥴🤪

...

ah...

mmm.

ohhhh....

ahem. I didn't think my simple question would come out like this.

Oh.

Yes, Artem, you fooled me for a while.
It was a sporting interest.
I hope it will be useful to someone, including me. :))

 
Nikolai Semko #:

Yes, Artem, you have cheated me for a while.
I worked on sporting interest.
I hope it will be useful to someone, and me among others. :))

Of course it will. Great. Thank you again!

S.F. This amused me: "counts correctly only till 28.02.2100 !!!!".

What do we do after that?

 
Artyom Trishkin #:

Of course it'll come in handy. Great. Thank you again!

S.F. This amused me: "counts correctly only till 28.02.2100 !!!!".

What do we do after that?

haha.
I doubt that this algorithm will be in demand for 75 years. Quantum computers will probably rule the world already, with completely different programming.
To be honest, it was lazy to take into account the Gregorian calendar. 2000 was high-stakes, 2100 is not anymore.

 
Nikolai Semko #:

haha.
I doubt that this algorithm will be in demand for 75 years. Quantum computers will probably rule the world already, with completely different programming.
to be honest, it was lazy to take into account the Gregorian calendar completely. 2000 was high-stakes, 2100 is not anymore.

For MN you can use a pre-calculated array, there's almost nothing in there

Ln2(12 months * hundred years)...11 if`'s and comparisons in binary search, but without other calculations.

 
Maxim Kuznetsov #:

You can use a pre-calculated array for MN, there is hardly any data there.

Ln2(12 months * hundred years)...11 if`'s and comparisons in binary search, but without other calculations.

No difficulties. I know how to implement it.
I just really don't want to do something that I'm 100% sure will never be useful.
 
Maxim Kuznetsov #:

You can use a pre-calculated array for MN, there is hardly any data there.

Ln2(12 months * hundred years)...11 if`'s and comparisons in binary search, but without other calculations.

ah, I read it wrong at first.
No, you are wrong. Performance increase will not work. You will still be stuck in calculations. And access to array elements slows down the algorithm very much.

 
A faster implementation of the standard PeriodSeconds() function:
int PeriodSecondsFast(ENUM_TIMEFRAMES tf) {
   return (tf>>14==0)?(tf&0x0FFF)*60:(tf>>14==1)?(tf&0x0FFF)*60*60:(tf>>14==2)?60*60*24*7:2419200;
}