Discuss better ways to manage MT4 external/internal variables

Hi MT4 programmers, users,

This thread is started to discuss ways to better control MT4 external variables than what the default MT4 methods provide. It is especially important for when a programmer only wants to provide .ex4 files, which means that the user cannot edit the source to change various default values.

Of course there is an ability to "Save" and "Load" a .set file at the time an indicator/script is added to a chart. However, the use of .set files is not automatic, and so it's a bit of a nuisance to always remember to "Load". Furthermore, suppose one has already added an indicator to many charts, but then wants to make a change that applies to ALL existing charts. A .set file does not accomplish this.

What does?

Well, the very generous FF contributor "hanover" has supplied a number of free MT4 programs, and he has established his own (private code) methods for the user to create a "Presets---Indicator Name.TXT" file, which goes into the C:/.../experts/files/ folder.

His programs have regular external variables of course. Consider for example his Recent News (v3) indicator. One variable is "ParameterFile = 'TXT'" (NONE is the default, but 'TXT' (or blank) is the value to look for the "Presets---Recent News.TXT" file).

If that file is found, it is parsed for legal variable names and values. THESE PRESET VALUES SUPERCEDE the values of his external variables in the "Inputs" section! (It works well, but one drawback is you do NOT see the actual value that is used, because the .ex4 has only the compiled value which it displays. The init() section finds the Presets and replaces the default value with an override).

Recently I posted some complimentary comments to Hanover (David) about his Presets-use-model, and CockeyedCowboy picked up on the discussion with some great ideas of his own (see post), and suggested I open a new thread to continue discussion... so here it is!

CockeyedCowboy (Keit), and others, please join in.

Kent (pips4life)

Here is another post I made on the topic which I'll repeat here because the subject is what this thread is about.

==========
Hi Keit (CockeyedCowboy),

Yes I'd like to see your code and some examples in use.

Do I understand, however, that the user will not SEE what the settings are? All variables will be blank upon attaching to a chart, but the defaults get written, first, to a file, and then if the user changes any var to non-blank, that new setting gets written into the file.

But apart from manually editing the set file, how does the user know his settings (if everything is blank)?


If the above is the use model, then I'm not sure I would use it but I'd like to see and study it anyway.

If you know of a way to use a Presets file AND to see those values in the Inputs section when you add the indicator to a chart, that would be fantastic. (To be clear, I'm fully aware how to Load a .set file, which obviously does show the user what the values are. This Presets is intended to set defaults *without* resorting to a .set file... that's the whole point of this discussion).


I've been kicking around another use model as an expansion on David's ideas. Here's a working draft. I'm just pretending to actually read values from a Presets file, but if David wants to share that bit of code, I'd be very happy! (I'm pretty sure the answer is no, so... oh well... I can do it eventually. The idea, however, is completely to David's credit. You (Keit) obviously thought of something similar a long time ago, so kudos to you too).



I just threw this together but I'm fairly pleased with it.

a) If MYVAR is left at "1" (the usual default), then if the PRESET = 2, the PRESET value is used, and no ERROR is issued (because the user deliberately created a PRESET and doesn't want to see any ERROR just to use it).

b) Similarly, if MYVAR is changed to "2" and PRESET=2, again there is no ERROR because even though the user changed it, the value is already what the PRESET said it should be.

c) However if MYVAR is set to "3" while PRESET=2, an ERROR is issued, because the MYVAR is neither the usual "1" nor the PRESET "2", but it *will* be ignored, and the user may be confused if not alerted to this fact.

d) If PRESET is <=0, then whatever the user sets MYVAR to be is what it is. In this test code, this pretend PRESET=0 means "there was no PRESET found for MYVAR".

Thanks in advance for any ideas/code/tips...

Kent

I should like to add that the "Presets" method of hanover's can also be used to control *internal* variables, not just external. It is probably best if they are global variables (defined before, not within the init() and start() blocks).

The init() code can parse the Presets file and override whatever is it's default value.


=====

Yet another method of control is the use of MT4-global-variables. (Push F3 to see if you have any defined). These Global Variables can be set and read real-time (such as per-tick with an indicator). If properly defined -- it gets a bit tricky -- global variables can pass some limited information from chart to chart.

One big drawback of these is they are only type "double" vars. No strings or other useful datatypes can be defined this way. At least one method to pass information between charts is to define a MT4-global that if one chart sets the value, other chart(s) can read (and possibly reset) the value to trigger some kind of action (like go read an external text file to get some detailed information on, say, some objects to draw on charts with the same Symbol() ).

Kent

First, When I memtioned empty value I was not really refering to the MT EMPTY_VALUE, but what ever would be deemed of NO value. When I decleare variables I always assign them an Empty or Clear value(depending of variable type) that is my way of knowing that the variable has no meaningful value set and can be disregarded. So when you see in my code

int VariableName = Empty;

it really means that it contains, at this point, no meaningful value. What ever you want to use to mean empty or no value is up to you.


If you cut your code down to if var == 0(empty) then use the preset if its any thing else but empty use what the user input. on exit save the these new settings to the file, the next time he starts the program the settings he has sellected last will be the new defaults. even though he has not changed the var == empty or not. The fact is you the user want to know what the setting in the file are? That must be communicated to the user in one way or another. Simple example let say you have a custom stochastic indicator with the usal setting of 5,3,3 and OB line at 80 OS at 20 as defaults contained in your set file. and all of which are externals which the user can set to his likeing. You can communicate these settings simply by using the indicators title. Granted its a very simple example, but its up to the programer to complish this in any way he deems right. you must also think of, do you the programer, want you the user to tamper with some of the settings? Like right now I am rewriting my lot sizeing program to control which orders that I the user can apply under certain conditions, all the parameters in that code are hard coded defaults by me the programer because me the user is not intitled to change them on the fly.

I have not used external parameters in my code in a very long time and I thing that MT way of removeing the code(stopping it) to make a change in a parameter then restating the code is not a very good idea. therefor I donot use them, except in writing, testing, and debuging, but never at run time.

So the real question is how do you get arround MT leaknesses? How do you make changes in your program settings without MT stopping your program when ever you open up the dialog box?

Having preset defaults in a set file is only part of the answer. How can you communicate to your program? In the screen shots, I give you a simple example of the user moveing an indicator level line, This one I call the Price Inefficicency Index, the first screen shot shows the default setting the 'a' on the left hande scale means 'automaticly set by the program' the user can simply move the line by draging it or opening its settings and entering anything he wants and the program will take the new value in its calculations. the 'm' on the left side now mean 'manually set' the number is the new setting the program is using. If I the user want to go back and get the program's setting I just have to drag the line, or any part of it, down to the bottom of the screen and the line will reset back to its default setting (as zero is not a value). Granted this too is not a very major development but it does give a good idea of what I mean about the user and the program communication between each other.



Keit

[edit] in this text when I say set file its not what MT calls a set file but a file that is created by the program itself and located in the file directory. The set extention is not manditory, in fact I use the extention .ini most of the time.

It getting late and I still have tings to do. but here is some code that I am working with right now, my lot size code. The file it creates is also included. will explain more later if need be. O well .ini file types are not allowed, look in the zip.


will explain more later

Keit

Hi Keit,

Your approach is interesting, and very creative.

I see that in this example, you saved only Int and Doubles to the .ini file, but can you also save any data types, like Strings, etc.? (I have zero experience with reading/writing binary data files. I suppose one can if a fixed file format is defined, but haven't a clue about a good way to go about this).

For a program such as RSI, it's relatively easy to imagine a way to tell the user what values the program is using for the few inputs. However, for something with a lot of inputs, it would seem impractical to me for a program like "Recent News" to create objects/labels/etc. to tell the user what values were used, and also to offer a way for the user to change it. For that, using external variables seems an easier way to go... unless I'm missing something in your use model.

Are your methods something you could imagine used for the Recent News kind of indicator? Or would the external variables still be the way to go?

If external, do you know of any way to automatically change the values that those external variables have as defaults to the values actually used by a Presets file? You've done a lot with MT4 others said can't be done, so maybe you're aware of something.

(The only idea that came to me is... one could have a Presets that is read looking for overrides automatically. Then whatever the program uses (a combination of defaults, presets, and user-entered overrides) could be written to another file, like (".USE" or ".NET" extensions). If the user wants to see the sum total of all variable values actually used, Edit the indicator and do a LOAD of that file and inspect the values. (One can cancel the Load after inspection). That said, this hardly seems worth the trouble as compared to just visually inspecting the Presets file and know that those entries are forced and everything else is at it says in the Inputs section. OR... as you wrote, the program could communicate to the user in some way graphically that certain values were used).


Thanks for a glance at your methods. If you want to add more comments or details, please do.

Kent

Kent

First, I have never published my methods of program (and user)communications any were, and to understand fully what I mean about communications I would have to publish some of those methods. I have posted some screen shots in the past, but not really got into any real discussions of how I did things. Mostly because most people want code and not knowledge. I have not posted any complete code projects because most wont run as a seperate unit most rely on more then one code block to run. Therefor I would have to post everything including libaraies, which I can not do. I can tell you how I do things and posts some code blocks as examples, but one must seek the knowledge and not not have the 'give me your code' additude.


I would also like to restate that I donot use the MT preset files as such. what I use is a data file created by a program and placed in MT file directory. All reading is done from those files and not the standard preset file. In which case we can create these files to meet the given needs with out any restrictions except for locations. Which one can bypass that, as well if so desired.



You can create these files to contain any data types including strings. They can be in text or binary formats and can even be encripted to make them unaccessable. (the file I posted above is a combination of binary and text if you get down to it.) It is best to have them in a fixed format as you say be cause one file can contain data from more then one program if the file is of a fixed length and the binary file lends itself to that. If more then one program accesses the data it will have to set the pointer to the start of its own data location before reading the file.



I have to be trueful, I have not run that program, so I can not commet on that.

Let see if I can explain in more detail what I said above.

If you have an

extern double UserInputVar = 0; // no value;
.....

then later when loading the data file, you check.

if( UserInputVar == 0 ) then UserInputVar = (data from file);
if the user entered a value that value is used in the program. if its empty then the default from the file is loaded instead.
.....

at the end of the program, in the deinit() we save the file but now the file is saved with the current values used in the program including the users inputs over writing the old defaults. That would change the files contens to those that the user entered as externals. therefor the next time the program is run the user no longer has to enter his settings as his settings are those contained in the file.

What the programer needs to do is give the user some means of a visiual on what those settings are. In so doing, the user can if so desired go to the dialog box and change them to something else. if the user does this a few times he'll going to end up with all his prefered settings saved to the data file and not need to visit the dialog box again.

If the programer doenot want to go throu the hassals of suppling the original data file he can have the program create one with all the start up defaults even if no value (empty) or (unknown) values are given. This will allowcate the space in the file for that data. Now the file doesnot have to contain just default settings, it can contain anything the programer wishs it to contain including runtime values. Which can be saved between program runs, values which would not be known when the file is created.




You do not need to write a different file, just rewrite the existing one with the users imputs. exception would be, if more then one user uses the same terminal, then you can have more then one file. but at least one external input would be needed, that being the file name to which the user is associated with. again if the user inputs a file name that doesnot exist the program will create a new default one under that name with the preset defaults and the same stated above would apply for each user except the user needs to supply the name of a file, but all other setting can be left empty.

Now you donot need more then one file here too if your code can set the pointer to different location of each user settings only one file is needed. If your program is such that diffenent settings are used for say different currency pairs. then the file can be so constructed as to have different defaults in the file and depending on the return of the Symbol() function a different pointer location can access different data and different user inputs can be saved in one file. only one file is needed even for different users, symbols, chart scales etc..

The file can be in text or binary and can be encripted to limit access. The end user doenot even need to know anything about that file. All he needs to know is that his prefered settings will become the start up defaults. No need to even open the file to see its contens. That is if the progarmer gives a means to see them on the screen.

That visiual can be accomplished in meny different ways it depends on each situation. It can be supplied on screen, or the user can on demand retreve it.


Keit

Kent

To give you some idea of user (and inter program) communications, I have posted a screen shot. I would say that half the objects you see on that screen shot are active and has or is capable of accessing program information and/or in altering program execution and parameters.

When I stated in an earlier post that altering the external input of MT is only half the story the other is communications. Not only between the user and the program but also between defferent program componets of a system.

I seen some people have auto order routines that the user draws lines on a chart and if price crosses it an order is sent. This is alright if limits or stop orders are used. but how would a user open a market order on the fly off the chart? drawning lines would not be very good here. Opening up the MT order dialog is time consuming and also takes up too much screen space. How ever haveing an active object on the chart that one drags and drops can do the same thing with out any other action, that is if the parameters of the order are precalculated within the program or in another program componet.

Behind that screen shot contains 12 seperate programs all running as one. The total lines of code? In exaccess of 25,000 lines (never counted them however). How does all this operate without out killing the CPU? Supprizingly it only uses about 20% of a less then optimum machine. If I programed this the way MT tells you to. Then this amount of code would slow down, if not kell, even the best computer. You see user externals are not the only thing I've over come but you need to know how the platform runs to understand how to change things. Which is another story in itself.

I have not exposed much of what I've done as there were no one interested in the how toos. Most just want code to snap out all winning ordres and would careless of how its done.

Things like creating what I called active libaraies were the libaray is actived in the init() function of another program and the library runs similar to any other program it can even have an interface which can differ depend on how it is actived, but want makes it different then a program is that it exposes itself (that is; its properties and methods) to calls from another program. How can a script run as a program and a libaray at the same time? and why would one want it to? That is something beyon the topic of this thread. If your really wanting to look at things differently. If your looking for the knowledge. If your willing to think for yourself. we can expand this topic to include other things. Also suprizing is that most, not all but most, of what I've done was in MQ native language.


Keit