Chatter about the MT5 strategy tester - page 16

 
Andrey Dik:

My last guess is that you are drunk! I don't see any other explanation.

There is no need to write every tick to a file! RAM is used for operative storage of incoming ticks and packs of ticks are dumped to disk, not each tick! Just use your brain, eh. Holy shit ****************! You don't have to write every tick to a file and the meta-quotes don't do that, they're not stupid. If that's what you do in your aplach, well, one can only sympathise with the backwards-driven Jobs...

Are you being rude again? Is that your way of being rude when you're being pushed around? You should be polite and courteous.

I'm not saying you have to write a tic. Cache will take care of that.


I've been trying to explain a different point to you throughout this thread...

How do you consolidate the data into one TICK - with 100 terminals ticking in

I.e. in the end we need to record one TICK, not 100.


Well, how could I pass by and not remember one interesting proverb: drunken drunk - a fool never.

p.s.

What will happen to the steam heating

 
Yuriy Zaytsev:

I have been trying to explain to you throughout this thread a different idea...

HOW DO YOU COMPREHENSE THE DATA IN ONE TICK - when you receive ticks from 100 terminals

So you will end up with one TICK instead of 100.

Please clarify the question.
 
Andrey Dik:
Please clarify the question.

You know, both Renat and Slava understood me and gave you and everyone else the information that you cannot write into the file - which collects data from different applications (terminals) - without a special manager.

By asking you to clarify your question, you don't seem to know what you're talking about.

 
Andrey Dik:
Clarify the question, please.

Maybe I can be more specific?

A pack of ticks is being accumulated for recording... Another tick is received and at that moment the recording to the file should take place. That's exactly the tick Yuri is talking about.

 
Yuriy Zaytsev:

You know, both Renat and Slava understood me and gave you and all the information that you cannot write into the file - which collects data from different applications (terminals) - without a special manager.

By asking you to clarify your question, you don't seem to understand what you're talking about.

Clarify your question, please.

Renat is Renat. Slava is Slava. Yura is Yura. Renat thinks with Renat's head. Slava thinks with Slava's head. Yura thinks with Renat's and Slava's heads (and now Alexei is trying to think for Yura as well). Clarify the question, please.

Clarification of the question is necessary not so that I could give you an answer to it (the answer to it was worth 10000$ yesterday, which you have missed), but so that you understood it, that clarifying the question helps you understand, and maybe finally conduct an experiment with links to a shared folder.


SZZ Renat and Slava are company representatives, so they cannot speak against the interests of their company. In this matter, the solution is against the interests of the company because it will require man-hours, the economic return of which is questionable for Renat.

 
Alexey Viktorov:

Maybe I can be more specific?

A batch of ticks is accumulated for recording... Another tick arrived and at this point it should be written to the file. That's exactly the tick Yuri is talking about.

It's not what happens with every tick, it's when a certain event occurs and not when the next tick arrives.

For example, the reset to disk file occurs when the buffer is already full and there is nowhere else to put the next ticks.

Either the terminal is unloaded and its task is to write to the files of the buffer - which it has not reset yet and keeps in the memory.

---

So pay attention: Incoming, ticks - of course system collects them, but does not write them immediately to the file, it collects them to the cache - but this cache has its own for each terminal! then rebuild the cache, with information at a convenient time in the file

If terminal is one and he writes data into his database - he has FIRST THICK, THEN SECOND THICK, THIRD THICK and he will put them consequently first into the buffer, and then it will put them consequently into a file.

(I have not described it in detail - assuming that people here understand it)

----

Now let's imagine that 100 terminals adds TIKI to cache, moreover each terminal adds TIKI to its cache

And then comes the event, the buffer is full, and they HERACK - each by itself reset buffer on disk to a file :

  • - and each terminal doesn't have to reset at the same time as the other terminal.
  • - and again not coordinating the write with the other terminal
  • - They will definitely get stuck in a situation where one terminal writes - i.e. locks the file - and the others just sit there and wait and wait until they dump their full caches to disk
  • - applications are running independently and they have no idea about each other.

IF ALL THIS IS RESET TO THE SAME FILE!!! ?

And now pay attention to the question, what will be in the file? That's right, CHAOS from 100 terminals.

 
Andrey Dik:

Clarify your question, please.

Renat is Renat. Slava is Slava. Yura is Yura. Renat thinks with Renat's head. Slava thinks with Slava's head. Yura thinks with Renat's and Slava's heads (and now Alexei is trying to think for Yura as well). Clarify the question, please.

Clarifying the question is not necessary for me to give an answer to it (the answer to it was worth 10000$ yesterday, which you missed the chance to get), but for you to understand it, clarifying the question will help you understand, and maybe finally conduct an experiment with references to a shared folder.


SZY Renat and Slava are company representatives, so they cannot speak against their company's interests. In this matter the decision is against the interests of the company as it will require the expenditure of man-hours, the economic return of which is questionable for Renat.

You are talking nonsense again.

You don't understand the questions and answers, so why should I bother explaining them to you or formulating them in any way? You should read what I write to others. Stand aside.

The couch is theoretically yours again, you have to maintain it somehow, make yourself comfortable. And don't generate more nonsense for a while.

 
Yuriy Zaytsev:

No, not exactly! The file is not written with every tick, but when a certain event occurs.

For example, the reset to disk file occurs when the buffer is already full and there is nowhere else to put the next ticks.

Either the terminal is unloaded and its task is to write to the files of the buffer - which it has not reset yet and keeps in memory.

---

That's what I said

Alexey Viktorov:

A batch of ticks is accumulated to be written...

Anything further is a refinement.

 
Alexey Viktorov:

That's what I said.

Anything beyond that is just a matter of clarification.

(I did not describe it in such detail - assuming that people here understand it).

I mean caching

The important thing is just how the ticks stack up :-). From one terminal, everything is clear - and from 100 post above painted.

That is why Renat writes that he needs a manager which will receive a tick from 100 terminals, handle double entries and write in the right place (firstly into the cache) one tick instead of 100 ...

So Renat said - it is not worth building a garden for 1% of users who will use a configuration with ONE database for a group of terminals, instead of going and buy a 3 terabyte disk

 
Yuriy Zaytsev:

(I have not described it in such detail - assuming that people here understand it).

I mean caching.

The important thing is just how the ticks fit in :-). From one terminal, everything is clear - and from 100 post above painted.

That's why Renat writes that we need a manager which will take a tick from 100 terminals, deal with doublings and write in the right place (first, into the cache) one tick instead of 100 ...

So Renat said - it is not worth building a garden for 1% of users who will use a configuration with ONE database for a group of terminals, instead of going and buy a 3 terabyte disk

At first I thought it was about downloaded history for the tester, but you and Andrey have shifted the conversation to recording history while the terminal is working.

Basically, let them each write their own history, but if we need to download the history from terminal's server, we may have used one tester and then had to launch the tester of another terminal... ...and again we have to download the existing history to it. So instead of this it's desirable to show the path to that folder with downloaded history data and not to load it a second or more times.

Or better yet, make a separate directory in a shared folder where the history will be loaded and divided by CA. And periodic synchronization of history from terminals, even from xxxx pieces, with the history, which was prepared for the tester, is not difficult to do, well, so in my view an amateur.

Reason: