Errors, bugs, questions - page 200

 
Ashes:

Nope. The usual test chase of one of the (multi-currency) championship participants, a long history swap on the pairs involved and that's it.

By the way, I'm curious, how does the tester determine which pairs should be swapped? Generally speaking, it is impossible to know in advance without having started testing...

As far as I know, the first pair to be swapped is the one being tested (it is specified in the tester's parameters and automatically added to its "Market Watch").

Other pairs are swapped according to different conditions. Personally I think the most correct way is to select them (by filling in the "market review" of the tester) in the initialization block, and then check the synchronization of data (it is useful not only in the tester).

It's also worth noting that the tester synchronises with the terminal, which in turn synchronises with the server...

PS

If I understand correctly, the tester can still load the data by itself, but in real conditions it will have to do it all by itself.

sergeev:

Confirmed.

Easiest exp by MA. After initialisation straight into disconnect. But it's unstable and every once in a while.

I'll have to see, but I don't seem to have had such glitches. Although there was a problem with the timer (I think it will be solved in the next build)...
 
After initialisation straight to disconnect. But it's unstable and happens in different timesThis is
exactly
what
happens in the latest build (355) if OnTimer instead of OnTick is used in the tester.
 
Erm955:
This is exactly what happens in the latest build (355) if OnTimer is used in the tester instead of OnTick.
I fortunately have both of these handlers. Although, I really suffer from timer bug (as EA is a mule)... :(
 
Rosh:
Perhaps the function of removing the indicator from the chart by means of MQL5 will appear.
Oh, that would be good.
 

Attempted to save tester report in Open XML format. Minutes of waiting (Pentium 4, 3.0GHz), application not responding, terminal.exe process was running rampant in task manager (about 50% of CPU) eating up RAM (about 250Mb), Result - nothing. MT5 only responds, no quotes coming in (connection status indicator with grey spinning circle). A second try, it is the same. The window "Tools" has 2 journal entries "MemoryException 107878464 bytes not available", the time corresponding to the attempt to receive a report in OpenXML.
The HTML file was generated almost instantly (about 12 Mb). It takes a long time to open, but it's not MT5's fault...

 

The compiler gives a warning if a variable is not used anywhere, and that's fine.

But it doesn't see such rubbish (unused variables) in classes.

 

version 5 build 355

Calculation of the time difference between bars:

datetime DefTime;

string PrintDate;

DefTime=Time[0]-Time[i]; //i can be replaced by any integer in our case from 0 and up ...

PrintDate=TimeToString(DefTime,TIME_DATE|TIME_MINUTES);

Print (PrintDate);

Here we have:

!!! 2010.11.22 02:09:24 A-PS (EURUSD,M15) 1970.01.03 04:00
!!! 2010.11.22 02:09:22 A-PS (EURUSD,M15) 1970.01.03 04:45 !
!!! 2010.11.22 02:09:20 A-PS (EURUSD,M15) 1970.01.03 04:30 !
!!! 2010.11.22 02:09:19 A-PS (EURUSD,M15) 1970.01.03 03:15 !
!!! 2010.11.22 02:09:18 A-PS (EURUSD,M15) 1970.01.03 03:00 !
!!! 2010.11.22 02:09:17 A-PS (EURUSD,M15) 1970.01.03 02:45 !
!!! 2010.11.22 02:09:16 A-PS (EURUSD,M15) 1970.01.03 02:30 !
!!! 2010.11.22 02:09:14 A-PS (EURUSD,M15) 1970.01.03 02:15 !
2010.11.22 02:09:13 A-PS (EURUSD,M15) 1970.01.01 02:00
2010.11.22 02:09:12 A-PS (EURUSD,M15) 1970.01.01 01:45
2010.11.22 02:09:11 A-PS (EURUSD,M15) 1970.01.01:30
2010.11.22 02:09:11 A-PS (EURUSD,M15) 1970.01.01:15
2010.11.22 02:09:09 A-PS (EURUSD,M15) 1970.01.01:00
2010.11.22 02:09:08 A-PS (EURUSD,M15) 1970.01.01 00:45
2010.11.22 02:09:08 A-PS (EURUSD,M15) 1970.01.01 00:30
2010.11.22 02:09:07 A-PS (EURUSD,M15) 1970.01.01 00:15
2010.11.22 02:08:58 A-PS (EURUSD,M15) 1970.01.01 00:00

Where does the 3rd number come from? The same situation is observed when using MqlDateTime.

 
KffAlex:

Where does the 3rd number come from? The same picture is observed when using MqlDateTime.

That's right - you show the time delta between bars in seconds as a standard date.

The delta between Monday and Friday bars is exactly 3 days.

 
I frankly hadn't thought of that. I keep forgetting the lapse in bars between Friday and Monday. Thank you.
 

I want to make during the optimization advisor to choose whether to connect indicator or not, (eg TRIX) if TRIX=0 - then the indicator will not be used, if TRIX = 1 or more, the indicator will be included and the number that is used in the value - this is a parameter of the indicator.

But the problem is that many indicators have more than one parameter - but if TRIX_1 = 0, TRIX_2 = 4 for example set during the optimization, the parameters will contradict themselves, this problem I solved this way: in the Expert Advisor goes through the values of TRIX_1 and TRIX_2 possible.i.e., TRIX=0 is TRIX_1=0 and TRIX_2=0 in the EA; TRIX=1 is TRIX_1=1 and TRIX_2=1; TRIX=2 is TRIX_1=1 and TRIX_2=2, etc. But the question arises

will the genetic algorithm work correctly in the tester, if all indicator parameters have already been enumerated in the EA itself and only one value TRIX, which stores parameter variations, was put in the input?