JassHelper 0.A.2.B - A vJass and Zinc 2 Jass compiler

[Barade]

Originally Posted by Anitarf

That's what keyword is for:

JASS:

library Test
    private keyword Test2
    struct Test
        private Test2 member

        public static method create takes nothing returns thistype
            local thistype this = thistype.allocate()
            set this.member = Test2.create.evaluate() // Works now.
            return this
        endmethod
    endstruct

    private struct Test2
        static method create takes nothing returns thistype
            return .allocate() // It won't work unless you declare a custom create method.
        endmethod
    endstruct
endlibrary

Regarding the problem with the constant string, it seems to incorrectly identify // inside the string as a comment. You can just write the url without the "https://" part, it works just as well. If you absolutely must have that, you can use string concatenation to work around the issue.

Unfortunately, there's another bug with /// comments after constants in 0.A.2.B:

JASS:

 
public static constant string defaultOrderErrorMessage = A_TEXT_TARGET_TALKS_ALREADY ///

says

Line 39697: Syntax Error, unexpected: end of line?
Line 39697: (At generated code): constant string s__ATalk_defaultOrderErrorMessage= A_TEXT_TARGET_TALKS_ALREADY /

Besides static if statements shouldn't ignore //! import statements (they don't ignore //! runtextmacro as well afaik).
I have some options which can be defined globally using constants and require more files to be imported.
Is there any interest in continuing the development of the JassHelper or to create a new vJass compiler? I would like to drop some unnecessary things, to allow more flexible encapsulation of declarations and to get something like "protected" into the language.

[Bribe]

This is a mostly-finished project but it is not under development any longer.

There is cJass which has more features but it is less stable than vJass.

[Skater]

Originally Posted by Barade

[…]
Is there any interest in continuing the development of the JassHelper or to create a new vJass compiler? I would like to drop some unnecessary things, to allow more flexible encapsulation of declarations and to get something like "protected" into the language.

Ofcourse there is interest in a new vJass compiler. It's not like the whole jassing community dependents on it.
Having multiple options for compilers is always good.
But if you want a new vJass compiler it should actually be just that: a vJass compiler. No "dropping unnecessary things" and no new features.
Not initially. First of all, we need a stable, clean rewrite for the current vJass. Then we could do some experiments.

If you want some radical changes maybe build another language on top of vJass. I'm all up for some jass-experiments.

[Barade]

Originally Posted by Skater

Ofcourse there is interest in a new vJass compiler. It's not like the whole jassing community dependents on it.
Having multiple options for compilers is always good.
But if you want a new vJass compiler it should actually be just that: a vJass compiler. No "dropping unnecessary things" and no new features.
Not initially. First of all, we need a stable, clean rewrite for the current vJass. Then we could do some experiments.

If you want some radical changes maybe build another language on top of vJass. I'm all up for some jass-experiments.

ok but I have to figure out how many people are actually still using vJass and how many of them could help developing a new compiler.
For example, if I am going to use Boost's Spirit library it would take some time for myself to become familiar with everything and I could be sure that probably no one else of the vJass users could help me.
As you can see it took a very long time for Vexorian to get to the current state of JassHelper, so it would be much better to have a team.

[cohadar]

Vexorian you cunt.
I asked you to make a for loop in vJass like 3 years ago.
And instead you go and make ZINC.

WHY IS THERE STILL NO FOR LOOP IN VJASS?


for i = 1 to n do
endfor

for i = a downto b do
endfor

while i < 10 do
endwhile


This right here would be a real contribution to vJass.

[Bribe]

I don't think he is any longer maintaining JassHelper.

Even better would have been optional parameters, nested textmacros, textmacros as arguments, and/or fix the annoying bugs that compile with normal WE but not with pJass.

But a new parser in general made in a proper programming language would be the best solution overall.

[cohadar]

Ok since Vex is no longer updating jasshelper I decided to take over.

Is someone doubts that I can do it, take a look at this:
http://sourceforge.net/projects/parapascal/

I am not interested in zinc, I will update only vjass.

The first thing I will do is to add proper loop constructs.

@moyack
Prepare to add following keywords to TESH:
for, to, downto, endfor,
while, do, endwhile

[BBQ]

If you are really going to take over (I just hope you aren't kidding, even if you "don't care about Zinc"), then the first thing you should do is fixing the order in which the initializers are fired, so that people won't nag about resources that aren't initialized within modules.

[cohadar]

I am dead serious.
I have both the skill and the will for the job.
But loops come first since they have greatest impact on ease of coding.

EDIT:
I have had a fist successful source code compilation.
Vex did not include parser builder version in the readme.txt and latest version did not work.
I had to install 3 older versions until I found the right one.
Source code looks terrible, one pas file with 13256 lines of code and no comments...
But believe it or not I have seen worse and survived.

[curi]

What are functions like this for?

function sa___prototype70_HydromancerAbilityEffects takes nothing returns boolean

And can I disable them?

We have 1400 of them in The Kingdom of Kaliron map. Our Korean translators asked about them because a lot of text for quests is in the map twice in prototype functions.

There's also another prototype thing:

call TriggerEvaluate(st___prototype58[i])

I don't know what that is either. I don't see info about prototypes at http://www.wc3c.net/vexorian/jasshelpermanual.html

Thanks for any help.

[Nuclear Arbitor]

i'm guessing that the first one places an affect. that may be related to the return bug though, which no longer works.

i think the second one evaluates something. it appears that there are non asci characters in both though. either that or someone has some odd naming conventions.

as soon as someone with more scripting experience drops by you be able to get more help

[curi]

the prototype functions are auto generated by vjass. our map has a function called HydromancerAbilityEffects. then vjass makes a second prototype version with that naming convention.

[Here-b-Trollz]

The prototype functions and triggers are the reason you can call stuff that hasn't been declared yet using .evaluate(). vJass creates a trigger and a dummy function for each function, as well as a prototype function for each set of parameters (i.e. function Generic takes unit X, real Y returns integer), then adds the dummy as a condition (and an action if you use .execute()). When you call .evaluate() you are actually calling the appropriate prototype function, which sets generic argument globals and then calls TriggerEvaluate(prototypetriggerarray[functionIndex]—the trigger stored in this index has the dummy function added as a condition (and possibly an action). This is also why we can store functions in structs or whatever—we deal with their indexes in the prototype array, not with the actual code variable.

[WaterKnight]

Trying to convert a .j file with jasshelper via command line

Code:

"jasshelper\jasshelper.exe" --scriptonly <commonJPath> <blizzardJPath> <inputJ> <targetPath>

as it is stated in manual.

It starts and everything, finds my input file but appends it to the contents of some other script file before checking it. I do not know where jasshelper reads that other script from (although I have worked with it before). None of the input parameters contain it. I cleared the logs of jngp, deleted .j and .wct files in its directory that had contained it and the maps including backups.

Does jasshelper/jngp inject the paths to the last saved maps/scripts anywhere? The code it currently adds is actually the one it should convert later on (gets modified however) but because of this incident, everything is doubly declared.

[WaterKnight]

Kay, got it now. Uberfail. There was a remnant of the old method I used. The .wct I merged with the external code contained an //! import line, reading from a path I temp-saved my assembled script before.