| 08-21-2007, 03:52 AM | #1 |
vJass's text macros allow some interesting ideas like generic programming. It allows you to create arrays which you can treat as first class members. This gives you the ability to pass them as parameters to functions which is extremely useful and basically helps makes generic utility classes more viable.  JASS:globals constant integer SIZE=20 endglobals //! textmacro declareArray takes TYPE,TYPENAME type $TYPENAME$Array extends $TYPE$ array[SIZE] //! endtextmacro //! textmacro mathBPCode takes TYPE, TYPENAME static method max_$TYPENAME$ takes $TYPE$ a,$TYPE$ b returns $TYPE$ if(a>b)then return a endif return b endmethod static method min_$TYPENAME$ takes $TYPE$ a,$TYPE$ b returns $TYPE$ if(a>b)then return b endif return a endmethod static method findMax$TYPENAME$ takes $TYPENAME$Array list returns $TYPE$ local integer i=1 local $TYPE$ max=list[0] loop exitwhen i>SIZE if( list[i]>max)then set max=list[i] endif set i=i+1 endloop return max endmethod //! endtextmacro //! runtextmacro declareArray("integer","int") //! runtextmacro declareArray("real","real") //! runtextmacro declareArray("iComparable","iComparable") struct Math static integer MAX_INT=2147483647 static integer MIN_INT=-(2147483648) //! runtextmacro mathBPCode("integer","int") //! runtextmacro mathBPCode("real","real") //! runtextmacro mathBPCode("iComparable","iComparable") static method round takes real x returns integer return R2I(x+0.5) endmethod static method floor takes real x returns integer return R2I(x) endmethod endstruct More over, though as you probably expected these first class arrays do no such compile time checkings so the seemingly horrible JASS://foo is an integerArray which has a size of 20 foo[-1] //this is valid? foo[21] // this will compile as well Yup, these first class arrays behave as pointers. So why not treat them as such. We can now have "real" pass-by-reference semantics. JASS:type stringPointer extends string array[1] function StringsAreImmutable takes string s returns nothing set s="Strings are immutable, so whatever you thought this was doing well it isn't" endfunction function StringPointersAreMutable takes stringPointer sp returns nothing // sp refers to the actual stringpointer, sp[0] holds the value of this string pointer set sp[0]="StringPointers will change the actual reference though." endfunction .... local string s="String" local stringPointer sp=stringPointer.create() set sp[0]=s call print(s) // "String" call StringsAreImmutable(s) call print(s) // "String" call StringPointersAreMutable(sp) call print(sp) // "StringPointers will change the actual reference though." This can be useful when needing a means to output multiple parameters from a function and creates cleaner looking code JASS:function retrieveUnitFromGC takes string key1,key2,unitPointer up return bolean //returns false if it fails ... //logic here endfunction if(!retrieveUnitFromGC("Map1","Hero",my_up))then print("No save data found, creating new hero") //logic else print("Hero found of level "+I2S(GetHeroLevel(my_up[0]))) endif Unfortuantly, vJass still has a long way to go. We still need to deal with the normalization of the language (natives act like C code, non-natives like Java?) and building an even optional standard library can be very useful. A richer syntax for the text macro would be nice to but beggers can't be choosers |
| 08-21-2007, 04:20 AM | #2 | |
Quote:
Unless I am misunderstanding you, vJass does expose their lengths JASS:type aaa extends integer array [5] function aaaa takes aaa returns integer return aaa.size endfunction I'll add the syntax error for [-1] and [oversize] on arrays.... |
| 08-21-2007, 05:20 AM | #3 |
| 08-21-2007, 06:57 AM | #4 | |
Quote:
At run-time? That would slow things a bit....  |
| 08-21-2007, 12:37 PM | #5 |
| 08-21-2007, 01:00 PM | #6 |
| 08-21-2007, 01:18 PM | #7 |
[quote] JASS://foo is an integerArray which has a size of 20 foo[-1] //this is valid? foo[21] // this will compile as well@ Just checked and jasshelper does give a syntax error when you use an index higher than the size. adding for -1 |
| 08-21-2007, 02:58 PM | #8 | |
[quote=Vexorian] Quote:
The help file never mentioned the .size field parameter so thats why I assumed it didn't exist. I only tested -1, I didn't actually try a value greater than size I'd actually really like to see extension methods ala C# orcas as a means to do wrapping of common.j methods ie. JASS:struct IntUtil method toString takes this integer returns nothing I2S(this) endmethod endstruct //later on you can now do this 1.toString() And Pipedream is right this version of reference arrays isn't very interesting. Unless you use a memory array and a pointer struct that has an address. That way you can do Ptr.adr=locationInMemoryArray, and then ptr[0]=stuff. But thats dary type of stuff. and a richer macro language so that we can design our own constructs and make the heniously ugly macros look prettier. We should get rid of the whole //! runtextmacro syntax. Throwing $ $ around everything should be suffecient if you really are concerned with people accidentally using them. That way we can create certain inline functions and constants that the optimizer would anyway inline. i.e. JASS:DECLARE $MYID$ 'I001' DECLARE $ISNULLOREMPTY(x)$ ((x==null) or (x=="")) Also perhaps you can add something like a static identifier to structs i.e. JASS:static struct math Pipedream I mean stuff like this JASS://Currently we have native struct/object methods like // these are essientially static methods on the the structs GetUnitName(GetTriggerUnit()) //non-native means of call struct methods... MyInstanceOfStruct.getName() //constructor for native Location(1,2) //constructor for nonnative pair.Create(1,2) //deconstructor for native RemoveLocation(l) //deconstructor for nonnative p.destroy() |
| 08-21-2007, 04:35 PM | #9 |
| 08-21-2007, 05:33 PM | #10 |