[epScript] Circular import problem when referring to another module's variable within EUD function (new problem introduced in 0.9.9.0)

[Dr-zzt]

This is a minimal working example:

main.eps:

import b;
var problem = 0;

function onPluginStart()
{
    py_print("Hello!");
}

b.eps:

import main;

function troublemaker()
{
    main.problem = 1;
}

function funcptrexmaple(f)
{
    return 0;
}

var trouble = funcptrexmaple(f_troublemaker);

This code works fine until 0.9.8.13, but won't compile in 0.9.9.0 or later, due to circular import. My guess is that EUD function pointer content evaluation was lazy before but for some reason it's eager now.

[Dr-zzt]

I did some more experiment, and it's actually seemingly not related to function pointers. This is a smaller example:

main.eps:

import b;
var problem = 0;

function onPluginStart()
{
    py_print("Hello!");
}

b.eps:

import main;

function troublemaker()
{
    main.problem = 1;
    return 0;
}

var trouble = troublemaker();

Works fine until 0.9.8.13, doesn't compile in 0.9.9.0 or later.

[armoha]

It's not just EUDFuncPtr but applying to every epScript globals: https://github.com/armoha/euddraft/blob/master/CHANGELOG.md#bugfix-7

Bugfix

• [epScript] Don't wrap global consants with ExprProxy unless there's forward-declared function call.
  • No need to write globalConst.getValue() in most cases.

    Improved

• [epScript] Don't add initialization trigger for global variables when their initial values are constants

var trouble = 1;  // since euddraft 0.9.9.0, this does not execute any trigger on game start
// before 0.9.9.0, this runs/wastes trigger to assign 1 on game start

// in euddraft 0.9.9.0, above code is rewrite into this:
const troubleRewrite = EUDVariable(1);
// works similar to `static var`