MQL: security vs opportunities - page 5
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Come on, that was a bogus number right off the bat. Ask for files in memory, it's supposed to be much easier to implement than mapping and more efficient.
MQL: Safety vs. Possibilities
Renat, 2013.04.25 15:57
We just made pips so we don't have to use files.
There is no effective two-way communication. Pipes don't work, everyone has probably figured that out by now... Files in memory?
Another question - is it my understanding that you can't release to the market code which gets its info from a third-party source via bundles?
We just made the pips so you don't have to use files.
I understand perfectly why server pips in the terminal. Only and only (everything else is just excuses) for one task - to combine terminals for the sake of arbitrage. But that is not part of our tasks.
Who really wants to make inter-processor exchange, it's enough to implement multichannel pipe-server. But you can't sell that in a marketplace, and that's exactly the goal you're pursuing.
You are wrong about only and only.
Arbitrage, copiers, mixers (this is when you run several Expert Advisors purchased in a market and create a hedge position from them), ...
Now Joo needs it for his GA. You never know what the programmer's fantasy will come up with.
I mean that it's very difficult to predict what you may accumulate with a new feature.
And in fact you do not really need to anticipate (like not a priority), people have a need,
checked for safety, if so, why not give them the opportunity.
You know very well that any extension is a plus for the application language.
SZ
I understand that it's hard to understand the future solution, but there is already a solution - "files".
They are not satisfied with only one thing "beats a screw" all the rest is checked and rechecked.
The question is not about reinventing the wheel, the question is about giving a more organic solution to what already exists.
SZY
Here is yet another way of using virtual files: instead of a button interface you make a bitmap interface, live interactive and highly tricked out (with round buttons, for example) and compute all that in OpenCL on the fly. You could at least broadcast TV news on chart. You have an example of OpenCL, run it for a month and see for yourself what will happen to the drive.
We just made the pips so you don't have to use files.
I understand perfectly why server pips in the terminal. Only and only (everything else is just excuses) for one task - to combine terminals for the sake of arbitrage. But that's not our task.
Who really wants to make inter-processor exchange, it's enough to implement a multichannel server pip-server. But you can't sell that on the market, and that is the goal.
But what about within the sandbox of a single terminal? And purely within MQL5, without dll?
I understand that the difficulty is that the Expert Advisor on the terminal and the EA in the tester are separated in memory space due to the fact that the terminal and the tester are different processes. It seems that this very separation of the tester from the terminal leads to such a "chasm".
Well, what about within a single terminal sandbox? And purely within MQL5, without dlls?
I understand that the difficulty is that the Expert Advisor on the terminal and the EA in the tester are separated in memory space because the terminal and the tester are different processes. It seems that this very separation of the tester from the terminal leads to such a "chasm".
Just the expert in the tester (even on a physically different computer or even in the cloud) can transfer data to the expert in the terminal using frames. That is, there is actually no gap
Yuk. Uck. (upper left eyelid went into shallow vibration mode)
And back? You need to get a batch of information on the agent (intended only for the agent) to be processed at the start of the run and pass back the result at the end of the run.
Uck. Yuk. (upper left eyelid goes into shallow vibration mode)
And back? You need to get a batch of information on the agent (intended only for the agent) to be processed at the beginning of the run and to transmit the result back at the end of the run.
Yuk. Uck. (upper left eyelid goes into shallow vibration mode)
And back? You need to get a batch of information on the agent (intended only for it) to be processed at the start of the run and to send back the result at the end of the run.
It's both complicated and expensive (especially in the cludes). It is possible, though.
We won't do that yet.
Simply put, give the agent an input parameter other than the one set by the standard tester.
It's both complicated and expensive (especially in the cludes). It is possible, though.
We won't do it yet.
For information - our MQL5 Cloud Network cloud servers normally generate about 5 terabytes of traffic per day, sometimes up to 10 Tb.
If unlimited custom traffic is sent to this network with a guarantee that every agent can be reached, the network will not feel so good.