Errors, bugs, questions - page 1859

 
Can you please tell me the build number of the latest version of MT4, 1065 - is this it?
 
Rashid Umarov:

I have seen the picture. It's not reproduced in my MQL5 code. But I see that it is recognised in MQL4 in the same way as in the picture. It means that

  1. either the editor has decided that this is an MQL4 code for some reason - in this case you should provide it to Service Desk with details
  2. Or it was originally used as MQL4, and then was renamed to MQL5

No other options

Originally it was an indicator for 4. I opened it from mql5 editor, saved it under a new name and started editing it for mql5. I did not have any errors during compilation, but it did not want to work. I started to look into it and noticed this difference.

After reopening the file in the editor, the order has coincided with the reference.

Not very convenient (if you don't know about this behavior beforehand).

 
fxsaber:

Is anyone else having a playback? The tester settings are as follows

Result

I.e. a run of 250K ticks in 16 minutes.


How is it possible to measure performance degradation in the tester? I tried GetTickCount and global variables, it fails.

GetTickCount may help. How do you use it?
 
kaus_bonus:


I checked it out. it took too long to finish, so no tester prints.

this?

https://www.mql5.com/ru/forum/190147#comment_4880571


Yeah, that. Can't find how it was measured. Maybe it was done for MT4. I couldn't get it to work in MT5 off the bat.
 
Vladimir Karputov:

Why don't you give us all the information? The fuller the question, the better the answer. What about tabs "Tuning", "Parameters"?


Please:

Setting:

setting

Settings (but there is nothing of interest here):

options

Optional: If optimisation is disabled, changing the parameter has no effect:

Parameter value "true":

nastr01


Param01

result01

Parameter value "false":

gfhfv02

result02

 
Yury Kirillov:


Please:

Setup:

Settings (but there is nothing of interest here):



You have a different number of trades in the two cases. Your trading mode is 1000ms. Check the log (you are displaying the result of an attempt to open a position, right?). It is very likely that you will have several REJECTs.
 
Yury Kirillov:

How can this be? Problems with the tester.

I am testing an EA. Compiling, getting results:

Recompiling, testing, getting results:



How can this be? The code has not been changed, settings have not been changed. Random functions are not used. Moreover. The parameter exMode2Cont is commented out everywhere in the code (except the external parameters section).

It means that changing the value of the USEFUL parameter in some cases affects the result of testing?

What is thebuild number?
 
Slawa:
What is the build number?

Bild 1583
 
Vladimir Karputov:

You have a different number of trades in two cases. Your trading mode is 1000ms. Check the log (you output the result of attempted position opening, right?). It is very likely that you will have more than one REJECT.


Naturally different number of trades, on the same data and with unclear influence of the disabled parameter in the code.

A suspicion arose that there is an influence on what cores are tested: local or network.

At the moment I'm checking this assumption...

 

It has been clearly established that the test result depends on which cores, local or network, are being tested.

Test1 local

а03

р03

Test2 networked

а04

р04

Test3 networked

а05

р05

Test4 local and network

а06

р06

After that I do not know who to trust! :-)

Well and it became clear why the single test gave unchanged results - it's done on the same local kernel.

I shudder to think what would happen if we closely analyse testing in the cloud.......