SC2Mapster

Bug with array properties:

unit u=UnitGroupSelected(player)[1]; 
//compiler exception

unitgroup g=UnitGroupSelected(player);
unit u=g[1];  
//works

Bug with switch:

switch (true)
{
    case true: Print("True");
    case false: Print("False");
}

prints:

True
False

Even if I am not adding a break, it shouldn't run the 2nd case, right?

Suggestions:

• Copy&Paste as well as Undo support for the Dialog Designer
  
• Support for multiple declarations of class variables:
  

void foo()
{
    int x,y,z,a,b,c; //valid
}
Class Foo
{
    int x,y,z,a,b,c; //error
}

• Search & replace support for script of dialogs (currently it searches only non-dialog script files)
  
• After writing a typecast statement like (int), the suggestion box doesn't pop up, not even when trying to force it with ctrl+space.
  
• Don't take me to the script immediately after defining a dialog event. I want to be able to define a bunch of events first and then script them. Make me switch to the script by pressing enter or double-clicking on the event only.

Don't fix what ain't broken :)

+= was valid code from the start, but I didn't learn about it until like 6 months in.

Suggestion with very low priority: When you inject a script to a map and open it for the first time, a "Could not find triggers" error message is shown; after saving and re-opening the map, this message disappears. So, I assume, your editor deletes the entire trigger file, and the editor re-adds a blank one or something similar. You could create this empty file right away to avoid the error message.

Got some issues with your optimization/obfuscation stuff lately. I usually don't use it, just decided to test, if it still works.

I did send you a script, which caused an exception if used with the Obfuscate Strings option, it also occasionally produces invalid galaxy script when using the Use Short Names option, however I wasn't able to reproduce that reliably.

Quote from SBeier: Go

Good point with the mod files.. Ill look into that when I have the time. Although, you will probably need to give the compiler the script explicitly, since, if I remember correctly, you can't get the .SC2Mod file from the dependencies.

Fine with me, just give us the possibility to add a list of usable functions or a script file, which is not added to the map script per project.

€ I would propose anyway, that you add a "scripts" folder or something, which holds the SC2 script files (the native lib and stuff), so we can update the script files manually, if SC2 gets an update.

Quote:

Doesn't give me any errors.. Did you forget the * in the type?

I am pretty sure, I didn't, but I must have made another mistake, because it works for me now.

I would like to use the GAx3 mod for my map, so obviously, I would like its functions to show up in Galaxy++ as well. How would I do that?

Just adding the script as a file does work for the syntax, but all the functions get added to the map again, which conflicts with the mod, which of course adds all functions itself.

And again, an issue with a variable disappearing in the generated map script.

    a1=BeamCreatePoint(p1,p2);
    ActorSend(a1,"SetScale 5");
}
Wait(4,0); 
ActorSend(a1,"SetScale 1");

generates

    ActorSend(Beam_BeamCreatePoint(p1, inlineReturner3), "SetScale 5");
}
Wait(4, 0);
ActorSend(a2, "SetScale 1");

Attached script file; you will find it in the Detect Mouse file (and, as usual, prepare for a mess ;) )

In vJass, we could do something like this:

class Foo
{
    static thistype[10] fooList;
    static int fooCount=0;
    Foo()
    {
        fooList[fooCount++]=this;
    }
}

This creates a cyclic dependancy error in g++ (obviously, there is no "thistype", but using "Foo" causes the error). Any way to do something similar, or do I need to use an external array of Foo?

Also, can we create a new instance for a struct, without assigning it for a variable?

Foo* f = new Foo(); //instead of this
new Foo();          //I want that

Error with operator overloading: If you do something like this:

point operator *(fixed f,point p)
{
    p.x*=f;
    p.y*=f;
    return p;
}

the resulting function will look like this:

point *O1(fixed f, point p)
{
    p = Point(PointGetX(p)*f, PointGetY(p));
    return Point(PointGetX(p), PointGetY(p)*f);
}

which is no valid Galaxy syntax, the function name must not include an asterisk.

If you do it the way it is supposed to be, this is not a problem, because:

point operator *(fixed f,point p)
{
    return Point(PointGetX(p)*f, PointGetY(p)*f);
}

this will just be inlined, and the function won't even exist in the generated script.

Initializer
{
    int i;
    int i;
}

produces an error message, but when you double-click it to get to the line of the error, you get a File not found error.

I would like support for this:

bool b = (0 <= angle < 90);

Oops, how did I miss that ;)

Suggestion: Add something to avoid errors caused by the order your initializers are invoked. Maybe an optional initializer priority or something.

If I have one initializer, which, for example sets tons of variables, I would like to be able to use these variables reliably in other initializers, without having to manually call them in a specific order.

Syntax-wise, it could work like an "overloaded" initializer to not break backwards compatibility:

Initializer{ //no argument, priority defaults to some value, possibly 512 similar to render priority for dialogs
     DoStuff();
}
Initializer(513){ //higher priority, gets invoked first
     DoStuffPreviously();
}

Yes, for dynamic arrays, it is not needed anyway, there might be cases, where it makes sense to do that (once resize works again, that is ;) ).

Quote from SBeier: Go

Are you sure the int[0] i; presists to galaxy code? It should be removed.

Seems so.
Source code:

int[0] i;
Initializer{
    Print(i[0]);
}

Generated:

void Invoke(string methodName);

int[0] i;


bool Init(bool testConds, bool runActions)
{
    UIDisplayMessage(PlayerGroupActive(), c_messageAreaSubtitle, StringToText(IntToString(i[0])));
    return true;
}

void InitMap()
{
    Invoke("Init");
}

void Invoke(string methodName)
{
    trigger initTrigger;
    initTrigger = TriggerCreate(methodName);
    TriggerExecute(initTrigger, false, true);
    TriggerDestroy(initTrigger);
}

Quote from SBeier: Go

How exactly do you define this 0 sized array? I thought I fixed that one.

int[0] i;
int[] j= new int[0]();

Nothing special. No error messages, can use them in functions, causes ingame error. Behaves exactly like it did pre-patch.

Documentation error for the null-pointers:

As of 3.0.0, null checks are now reliable. What this means that if a pointer is uninitialized, pointing to a deleted object, or has been assigned null, writing (p == null) will return false true.

Can confirm the typedef - issue, couldn't get any kind of typedef to work at all.

Some issues I mentioned earlier seem to work properly now, as you mention in the changelog (wait and local re-use havoc)

Others haven't changed (like replace and undo issue, 0 size array), but you are probably aware of that and consider them lower priority.

Suggestion: Add typecast (and maybe implicit cast, but that would depend on your personal taste) from byte/int/fixed to bool, if 0 then false, else true.

Resizing a dynamic array does not seem to work anymore,

    int[] i=new int[0]();
    i->Resize(1);

causes a Method not found error.

Also, if you write i-> , the autocompletion does not suggest length or resize, while for writing i. it does, but it still doesn't work (->lenght seems to work while .length does not; autocompletion suggests the other way around).

And a question: Is implicit resizing supported? I cannot test the result right now, but writing

    int[] i=new int[5]();
    i[10]=5;

throws no error for me.

Heh, I do this kind of stuff everywhere, in case I need the dialog item for something later on ;)

Using replace will always replace the characters behind your cursor with the replacement word. Also, undo does not work for replacing; it adds the replaced word again, but does not delete the replacement.

Quote from SBeier: Go

What I meant was to use global fields instead of local variables. It won't merge a field with a local variable.

Okay, thanks.

Unfortunately, similar bugs happen all over my map script. I would need to replace a huge amount of locals with globals to fix all the bugs, and for my messy script, it is very hard to spot the exact locations this malfunction happens, especially because it is not reported as an error.

I would need to check the entire generated script vs the source files and once you fixed the bug, I'd have to change it back... might as well rewrite the entire map script xD. Wish me luck.

Quote:

So far, I should have the compiler part done, and about half of the suggestion list done. But once I get that done, there will be some time with bug fixing.

Glad to hear that :)