Development plans for the MetaTrader 5 Strategy Tester - page 13

 
Сергей Таболин:

And I specifically asked for a flag to be set after each compilation that this is a different program, that the previous data is no longer relevant. And you want to get back to that bedlam? I am against it!

EX5 hash is stored in each opt-file. Therefore, even a recompilation without changing the source code is a new EA. And it is right.

I was talking about treating each line of cache as a set-file of a single run. No one prevents you from loading the set-file of another EA into your own one.

This is exactly what I would like to do here.


Now, if there is a variable "MyName" in the EA. And the set from the left EA has such a variable. Then, when we load this set-file, the variable MyName will change to the value in the set.

It is logical that the same behaviour would occur when working with the cache. There, in fact, each line of the pass is a set-file of settings.

 
fxsaber:

The EX5 hash is stored in every opt file. So even recompiling without changing the source code is a new EA. And this is correct.

We were talking about counting each line of cache as a set-file of single run. No one prevents you from loading the set-file of another EA into your own one.

This is exactly what I would like to do here.


Now, if the EA has the variable "MyName". And the set from the left EA has such a variable. Then, when we load this set-file, the variable MyName will change to the value in the set.

It is logical that the same behaviour would occur when working with the cache. There, in fact, each line of the pass is a set-file of settings.

I understand what we're talking about. But! The set is designed for a specific version. Hence, a compromise can be made. Delegate responsibility for the correctness of the sets used and other things to the progressor. To do this, it is sufficient to record the version of the software. If it has not changed, it is one thing, if the proger has changed the version, then ....

 

Forum on trading, automated trading systems and trading strategy testing

Undocumented... (bugs, opportunities...) MT5

Sergey Tabolin, 2019.05.13 09:23

================

2. How to catch the optimization end date programmatically?

Preferably make it possible to define this date (like TESTER_END_DATE).

 
Сергей Таболин:

I know what it's about. But! The set is designed for a specific version. Hence, a compromise can be made. Delegate responsibility for the correctness of the sets used and other things to the progressor. To do this, it is sufficient to record the version of the software. If it has not changed, it is one thing, if the proger has changed the version, then .....

What is wrong with a set for such an EA

input int i1 = 0;
input int i2 = 0;
input int i3 = 0;


to apply to such an EA?

input int i1 = 0;
input int j1 = 0;
input int i2 = 0;
input int j2 = 0;
input int i3 = 0;
input int j3 = 0;

Now in MT4/5 everything works perfectly in such cases. Similarly, there is no reason not to do so from a set of sets - cache Optimizer.

 
fxsaber:

What's wrong with a set for an EA like this


to apply for one of these?

Now in MT4/5 everything works perfectly in such cases. Similarly, there is no reason not to do so from a set of sets - cache Optimizer.

There is a reason. The optimizer cache is a cache of a particular program. It is intended solely for it. The single test should be launched only with the program, with which it has been created.

You can manually load a set from the first example into the second one, adjust additional parameters and everything will be fine. But it is too much to run a single test from the optimizer with another EA. You can imagine how many tears will be shed on the forum at once because of it.

 
Сергей Таболин:

But running a single test from the optimizer with another EA is too much. Just imagine how many tears will immediately be shed on the forum because of this.

It's a pity you don't understand. You can't even think of a scenario of tears. It's hard to talk when understanding of the Tester's work is disproportionate between opponents.

 
fxsaber:

I wish you could understand. You can't even come up with a scenario for tears. It's hard to talk when understanding of the Tester's work is disproportionate between opponents.

It's one thing to be a Tester, and quite another to be an Optimiser. There is no need to confuse the red with the wet.

I understand your message very well, that's why I'm against ))))).

 
Сергей Таболин:

A Tester is one thing and an Optimiser is quite another. There's no need to confuse red with wet.

The arguments are nil, unfortunately.

 

Forum on trading, automated trading systems and trading strategy testing

MetaTrader 5 Trading Strategy Tester development plans

Renat Fatkhullin, 2019.09.02 23:03

  1. Let's rewrite the source data preparation mechanisms to reduce agent synchronisation costs

    The acceleration will be especially noticeable on local agents, where you won't have to pump large volumes and won't have multiple copies of historical data.

Is it possible to keep only one copy of price data for all local Agents in RAM? Right now memory consumption is quite irrational.

 
fxsaber:

Is it possible to keep only one copy of the price data for all local Agents in RAM? Right now memory is being used up quite irrationally.

Seconded.