Servicedesk. Complaints, suggestions. - page 8

 
Alexey Kozitsyn:

In MT4, using a similar scheme, the data is not loaded. I.e., initially there are about 2050 available, and in order to get more history, you need to scroll deeper into the history.

This is what I was telling you about. Apparently, there is no other way in MT4. It has always been that way.

But MT4 always synchronizes data of TF and indicators by itself and the programmer does not have to take care of it at all (if only he has previously downloaded the history through the archive of quotations or through the chart). In my opinion, this is a greater advantage than the ability to automatically download the history of MT5.

 
Alexey Kozitsyn:

In MT4, using a similar scheme, the data is not loaded. I.e. initially 2050 are available and to get more history you need to scroll the chart deep into the history.

I also noticed one peculiarity. When scrolling, if new bars are loaded, prev_calculated is reset to 0 (works like "refresh" button).

Dear developers. Please comment if there is any way to swap history without using chart scrolling (MT4!).

My Expert Advisor is very good.

You cannot download the history without scrolling the chart programmatically, only emulate the PgUp key - it was written about on the Fours forum.

You can also pump up the history data by running the tester several times in "All ticks" mode. The timeframe being tested and the timeframes below up to M1 will be swapped to the maximum - several attempts will be made to download all data available on the server, without paying attention to the setting of the maximum number of bars on the chart.

 
Slawa:

In quadruple, it's the same as before.

It is impossible to pump up the history without scrolling the chart, only by emulating the PgUp key - it was written about it a lot on the quadruple forum.

You can also pump up the history data by running the tester several times in "All ticks" mode. The timeframe being tested and the timeframes below up to M1 will be swapped to the maximum - several attempts will be made to download all data available on the server, without paying attention to the setting of the maximum number of bars on the chart.

Unfortunately haven't visited the quad forum. Thanks for the clarification.
 

Wrote to servicedex #1342540 on another subject. At first they replied and asked for a code. I provided the code and detailed screenshots describing the problem. In reply for a few days - silence. And I just asked if they would fix it or not.

What's the point: when accessing timeseries in the MT5 tester, if the start date of the test is earlier than the start date of the history for a symbol, you can't get data from its timeseries throughout the test, even on those dates when this symbol already has history. Instead, the tester sends the error"XXXYYYY: There is no history. Please make sure that XXXYYY history is available on the trade server".

The multicurrency analysis often uses that subset of symbols from the common set, for which at the current moment of the test history has already begun. And as the history for other symbols from the set starts, it connects them too. I consider the absence of such a possibility for the emphasized multicurrency terminal as a great disadvantage.

Even after sending the request to servicedex, I found out that the SymbolInfoTick function works well in this situation, despite the impossibility to work with timeseries. It means that in order to create normally working programs we have to invent a crutch that builds data of necessary timeseries during the test by itself fixing price changes of each tick.

 
This seems to be a limitation of OHLC on M1 testing mode, in "All ticks" mode everything works.
However, the test run time of my EA from 2000 ao 2016, which is programmed to act only on H1 bar openings, in OHLC mode on M1, is 40 seconds
In "All ticks" mode it runs from 2000 to 2016 for 6 minutes, which is almost 10 times slower. At exactly the same level of testing accuracy. I think that "crutch" with creation of required TF via SymbolInfoTick will work much faster.
 
Ilya Malev:
This seems to be a limitation of OHLC on M1 testing mode, in "All ticks" mode everything works.
However, the test run time of my EA from 2000 ao 2016, which is programmed to act only on H1 bar openings, in OHLC mode on M1, is 40 seconds
In "All ticks" mode it runs from 2000 to 2016 for 6 minutes, which is almost 10 times slower. At exactly the same level of testing accuracy. I think that "crutch" with creation of required TF via SymbolInfoTick will work much faster.
Better to remind SD of the problem once or twice without crutches. Developers are busy people and sometimes miss requests. And this, from the description, is an obvious bug in the tester.
 
Servicedesk wrote that the behaviour has been corrected. Thank you!
 
Ilya Malev:
Servicedesk wrote that the behaviour has been corrected. Thank you!
You're welcome! (If it's addressed to me and not to SD).
 
Alexey Kozitsyn:
You're welcome! (If it's addressed to me and not to SD).
Actually it's addressed to SD, but thank you for the thread too)
 

Application

Error in the storage

Open, Started: 2015.09.29 17:09, #1311805

Hanging unanswered ...