Menu
Sister Sites

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

MQL4 code help 12 replies

Harvester 14 replies

MQL4 Language Most Recent Version is it updated beyond the tutorial on the mql4 websi 6 replies

MT4 Indicators Usage Tutorial/MQL4 Reference Documents 36 replies

Attachments: C++ MQL4 Forex Harvester
Exit Attachments

C++ MQL4 Forex Harvester

Quoting JensItzig
FIXED API access must be granted from your broker. I understand why you are doing it, but i don't see that the reward will high enough for the effort you put in. Since you need to use an EA to communicate you are doomed and bound to MT4. Multithreading can only be achieved via multiple terminals (that's the only reason is see) but since you don't build a really big dashboard or multi pair EA you can achieve this also with MQ4. Since now i have found only a few things that can be not done by MQ4 (and mostly you can find a workaround). To show you...
Yea there remains some logic on MQL side of course to redirect the requests from C++ to the MQL API functions. But once that is done, the entire trading logic can be in C++. I respect you for having such confidence in MQ Language and I am actually surprised people can create amazing things with it. But as I said, I don't like it and I don't want to use it.
Quoting JensItzig
FIXED API access must be granted from your broker. I understand why you are doing it, but i don't see that the reward will high enough for the effort you put in. Since you need to use an EA to communicate you are doomed and bound to MT4. Multithreading can only be achieved via multiple terminals (that's the only reason is see) but since you don't build a really big dashboard or multi pair EA you can achieve this also with MQ4. Since now i have found only a few things that can be not done by MQ4 (and mostly you can find a workaround). To show you...
I agree with you. MT4 performance issues can be easily resolve using MT5. MQL5 have similar performance than C++ and MT4 have expire date since MT5 supports hedge

Quoting pascalx
{quote} Yea there remains some logic on MQL side of course to redirect the requests from C++ to the MQL API functions. But once that is done, the entire trading logic can be in C++. I respect you for having such confidence in MQ Language and I am actually surprised people can create amazing things with it. But as I said, I don't like it and I don't want to use it.
Using DLLs you can transfer code to other platform. And using pipes or file mapping you can send all mql info like quotes without have rewrite all in your C++ side. If mql4 had success is because you can overpass the platform limits easily
Try don't lose pants never...
Quoting braintheboss
{quote} I agree with you. MT4 performance issues can be easily resolve using MT5. MQL5 have similar performance than C++ and MT4 have expire date since MT5 supports hedge
I don't think that MT5 will kill MT4. What do we see the last years. MT adds the features from MT5 into MT4. The codebase is to massive for MT4 and there are as many reasons to switch to MT5 as to build a C++ bridge.

The only issue is the lack of multithreading and this is not solved in MT5 either. The few things that make MT5 superior over MT4 are that marginal. All the graphical features for indicator buffers can be easily build in MT4 (even better than the standard MT5 stuff). The additional event hooks, ok good for lazy programmers, but not a real improvement. Better backtesting features, since MT4/5 lacks using ticks this is also nearly worthless.

Quote
Using DLLs you can transfer code to other platform. And using pipes or file mapping you can send all mql info like quotes without have rewrite all in your C++ side. If mql4 had success is because you can overpass the platform limits easily

With DLL you don't become plattform or system independent. If you want to achieve this you must use something like Java or Smalltalk (when it's available for the different systems).
I don't want to hijack the thread. That's just my thoughts about your project.

If you feel comfortable mit C++, then use it. I will watch the outcome.
Quoting JensItzig
{quote} I don't think that MT5 will kill MT4. What do we see the last years. MT adds the features from MT5 into MT4. The codebase is to massive for MT4 and there are as many reasons to switch to MT5 as to build a C++ bridge. The only issue is the lack of multithreading and this is not solved in MT5 either. The few things that make MT5 superior over MT4 are that marginal. All the graphical features for indicator buffers can be easily build in MT4 (even better than the standard MT5 stuff). The additional event hooks, ok good for lazy programmers,...
About Metaquotes interests they are forcing kill MT4 forcing brokers to use it. You have many news about this and CEO words too. It's not for tomorrow but will be fast. About MT5 is true is not perfect but it's big improvement comparing with MT4.

I disagree in your graphical features affirmation. Mt5 is easier. You only get a MT5 color buffer with 256 levels and try do it in MT4 About other features it's true not big difference but it's normal. It's a evolution not a new platform. At least the base. OOP was a big improvement and performance is a big step too. If you see indicators handles you will see the big difference with MT4 icustom.

But really the question is because somebody want wrap a platform if you don't like. If you want replace a platform then you create one new, not use the old. While he have connect to mt4 he will be dependent to that platform then I don't see the real advantage
Try don't lose pants never...
Quote
I disagree in your graphical features affirmation. Mt5 is easier. You only get a MT5 color buffer with 256 levels and try do it in MT4

It's easy. Look at my screenshots. The indicators are build full graphical, so i can use 16M colors plus alpha. The graphical improvements are fine (as is wrote) for inexperienced programmers. If you think a step further and build your own framework for MQL you don't need such stuff.

Instead building such things the developers of MT should thinking about the shitload of problems that comes form their bad design. First of all the lack of tickdata, that will be the biggest issue and is not solved with MT5. Next, I guess for a lot of MQ developers it has more value to be able to get a correct 24 SMA on H1 for 10:24 am @ 11:30 am (that's not possible because of the bad design). At the end of the day you must develop all stuff by your own when you need rock solid libraries (and that's what i have done the last 3 years only to be able to do the stuff i want to do with MT4).

Please tell me one thing! Why should be someone using iCustom? The only reason is to use indicators you don't developed by yourself (mostly stolen in my experience). If you the developer of such an indicator you build a class that can be used inside the indicator(s) and ea(s).
Quoting JensItzig
{quote} It's easy. Look at my screenshots. The indicators are build full graphical, so i can use 16M colors plus alpha. The graphical improvements are fine (as is wrote) for inexperienced programmers. If you think a step further and build your own framework for MQL you don't need such stuff. Instead building such things the developers of MT should thinking about the shitload of problems that comes form their bad design. First of all the lack of tickdata, that will be the biggest issue and is not solved with MT5. Next, I guess for a lot of MQ developers...
I think I explained wrong. In mt5 one plot can have many colors only setting a number color index. In mt4 you need a plot per color. This complex a lot the code and slower too. About icustom depends of project. Normally you use it like plugin. If you want use some indicators you don't need worry about dependencies and that things. You can check my thread. Finally I out all indicators of EA and use it separately. Was a chaos use it inside EA because a bug in one indicator can break app execution.

About Metaquotes who used his products know are a crap with many bugs and missing features but... It's the most used platform. Then we have adapt to this
Try don't lose pants never...
Quoting braintheboss
{quote} If you want replace a platform then you create one new, not use the old. While he have connect to mt4 he will be dependent to that platform then I don't see the real advantage
I don't know how that would work. You need to work with a broker, and brokers only support a limited amount of platforms. One of them is MetaTrader. If you build a platform from scratch, you have no broker to connect it to.

Quoting JensItzig
{quote} Please tell me one thing! Why should be someone using iCustom? The only reason is to use indicators you don't developed by yourself (mostly stolen in my experience). If you the developer of such an indicator you build a class that can be used inside the indicator(s) and ea(s).
Good point. Ideally with Market data you can write all indicators from scratch and would not need to use indicator functions. However, if the build-in indicators work correctly, there is probably not much reason to reinvent them.

You sound experienced man. Glad to see someone working 3 years on tools to get to the goal. That's the spirit.
Quoting pascalx
{quote} I don't know how that would work. You need to work with a broker, and brokers only support a limited amount of platforms. One of them is MetaTrader. If you build a platform from scratch, you have no broker to connect it to. {quote} Good point. Ideally with Market data you can write all indicators from scratch and would not need to use indicator functions. However, if the build-in indicators work correctly, there is probably not much reason to reinvent them. You sound experienced man. Glad to see someone working 3 years on tools to get to...
I understand you need use the data for can do work your app but I think you are duplicating the work. When you uses one platform, even for obligation, you have a payload for use something is developed for others. This payload it's many things will not make in way you like. My question is when you finish your project which tool will be the first you will develop? And what unique features will have for use your "platform"?
Try don't lose pants never...
Quoting braintheboss
{quote} I understand you need use the data for can do work your app but I think you are duplicating the work. When you uses one platform, even for obligation, you have a payload for use something is developed for others. This payload it's many things will not make in way you like. My question is when you finish your project which tool will be the first you will develop? And what unique features will have for use your "platform"?
I am not sure if I understand you. There is no platform advantage per se, but the development environment will be much nicer and it will be easier to write good and maintainable code.
Quoting pascalx
{quote} I am not sure if I understand you. There is no platform advantage per se, but the development environment will be much nicer and it will be easier to write good and maintainable code.
But you can have same with a naked library. You don't need move all to C++ library. But finally this depends of needs of everyone. I don't know. I have my framework and that things but I try reuse all things I can inside MT4
Try don't lose pants never...
Quoting braintheboss
{quote} I think I explained wrong. In mt5 one plot can have many colors only setting a number color index. In mt4 you need a plot per color. This complex a lot the code and slower too. About icustom depends of project. Normally you use it like plugin. If you want use some indicators you don't need worry about dependencies and that things. You can check my thread. Finally I out all indicators of EA and use it separately. Was a chaos use it inside EA because a bug in one indicator can break app execution.
I got you even in the first step. I only need one buffer as reference for all informations and needs (this also prevent that someone uses my indicator via iCustom). Lazy calculations, more flexibility, use of alpha is included and still not available in MT5 (maybe the change form RGB to ARGB for buffers or add a buffer for alpha).

Using indicators as "plugins" via iCustom is nothing than a dirty hack. You end up calculation much to many data, the overhead is not worth the simplicity. Dependencies are not an issue because you can include them into the indicator you want to use. You mentioned the big step when OO was intoduced, so why you don't use it and make your base for indicators and the eas solid (every developer using c++, java, smalltalk what do this).

Quote
Good point. Ideally with Market data you can write all indicators from scratch and would not need to use indicator functions. However, if the build-in indicators work correctly, there is probably not much reason to reinvent them.

The build in indicators don't work correctly in different situations. Think about this scenario: A stochastic indicator that plots the M15 stochastic on M1 chart. Only using PERIOD_M15 instead PERIOD_CURRENT will not do the thing, because of the bad design of the timeframe data. If you want to have correct working indicators you are doomed to implement all of them in the correct way even when there is a standard indicator (you can use them only in the case when the higher period end it candle also). Make a test and write such an indicator and let it run several hours and then open a fresh chart with the same indicator and you will clearly see the difference. This is the reason that most of the users talking about that MTF indicators are "repainting". It's obviously no repainting it only the bug those indicators are implemented by MT.
Quoting JensItzig
{quote} I got you even in the first step. I only need one buffer as reference for all informations and needs (this also prevent that someone uses my indicator via iCustom). Lazy calculations, more flexibility, use of alpha is included and still not available in MT5 (maybe the change form RGB to ARGB for buffers or add a buffer for alpha). Using indicators as "plugins" via iCustom is nothing than a dirty hack. You end up calculation much to many data, the overhead is not worth the simplicity. Dependencies are not an issue because you can include...
Simplicity vs flexibility. You takes one depending project needs. Normally indicators have enough with plots. Don't need alpha or others things and mt5 have more flexible plots. About more control with indicators I explained you before. I had EA with both approaches and I know how works every way. Finally I decided do the things more simple. Really I can survive with icustom limits

About OOP why you think I'm not using? I'm C++ expert. Templates are my favorite feature. I use all features give me MQL. But one thing is make a helper library ( mine have more 300 files) an other is search the excellence. I want something for code fast and easy. I will not code something is available at platform if cover my needs
Try don't lose pants never...
Quoting braintheboss
{quote} But you can have same with a naked library. You don't need move all to C++ library.
What is a naked library?

Quoting JensItzig
{quote} The build in indicators don't work correctly in different situations. Think about this scenario: A stochastic indicator that plots the M15 stochastic on M1 chart. Only using PERIOD_M15 instead PERIOD_CURRENT will not do the thing, because of the bad design of the timeframe data. If you want to have correct working indicators you are doomed to implement all of them in the correct way even when there is a standard indicator (you can use them only in the case when the higher period end it candle also). Make a test and write such an indicator...
Would you mind explaining why this issue happens exactly? I don't know much about the indicators and I am curious to hear why it is broken. It may help me in the future to avoid doing the same mistake.
Quoting pascalx
{quote} What is a naked library? {quote} Would you mind explaining why this issue happens exactly? I don't know much about the indicators and I am curious to hear why it is broken. It may help me in the future to avoid doing the same mistake.
Its a personal way to call libraries with auxiliar code. Not main code. Its like a helper.
Try don't lose pants never...
Quote
Simplicity vs flexibility.

You started with the comparison which system has the bigger balls. 256 additional colors via buffer is not a groundbreaking feature it can be done also with MT4 (and even better), thats my argument. And you must agree that 1 buffer is more simple than 2 buffers.

Quote
About more control with indicators I explained you before.

No you don't! Regarding the dependency, you include your class file and that's it. It's the same work as #resource your indicator or even worse hope that the system will find it @ runtime (via name convention in the filesystem). For me more a nightmare then real dependency management.
Then the "perfect" interface definition when you use iCustom. Errors in calling the interface are only found during runtime, no compiletime checks.
Jens if you ever happen to want to code in C++, please drop me a note. You appear to be valuable. I am from Germany too btw.
Quoting JensItzig
{quote} You started with the comparison which system has the bigger balls. 256 additional colors via buffer is not a groundbreaking feature it can be done also with MT4 (and even better), thats my argument. And you must agree that 1 buffer is more simple than 2 buffers. {quote} No you don't! Regarding the dependency, you include your class file and that's it. It's the same work as #resource your indicator or even worse hope that the system will find it @ runtime (via name convention in the filesystem). For me more a nightmare then real dependency...
Buffer colors was just example. Not important if you use 2 or 100. Its easier with MT5.

MQL miss some features like namespaces. When you use same name conventions for your aplications this missing feature make add calsses a problem. That I wanted tell with dependencies inside project
Try don't lose pants never...
Quoting braintheboss
{quote} MQL miss some features like namespaces.
Does MQL5 actually support multiple inheritance, referencing/dereferencing, multiple template arguments and typedefs?
Quoting pascalx
{quote} Does MQL5 actually support multiple inheritance, referencing/dereferencing, multiple template arguments and typedefs?
Evidently MQL is a simple version of C++. But really you need all that features for make indicators? How many projects are enough big in MT for need full spec features?
Try don't lose pants never...
0 traders viewing now