The Battle for Wesnoth Forums

walker

I think I have it working. :D

I had to ensure that the unit vars had the correct value when the ability filter saw them plus, as they are static values I didn't need to use unit variables.
I wanted this code to work with variables and this seems to work:

First, make the coordinates explicit values ...

@gnombat,Soliton

Thanks for your posts, been prioritising another issue that's now (I think :fingers_crossed: ) resolved.

Update
:inspect shows that the actual filter (for unit and location) doesn't see the unit variable; i.e. x="" and y="".
My bad; should've looked there first - checked the ...

To access a container you can try: wml_vars.__children['return_guardian'][0].__attributes['return_x']

I suggest to test these things interactively on the formula console (press f) in-game first. (i.e. my_leader.wml_vars will print a representation of wml_vars of the current side's leader ...

...So this is always going to happen eventually.


Thanks for that last info-nugget, it helped. :D

Based on somewhat sparse testing I think I have two working return_guardian variants. :shock:

return_guardian_u , only uses unit variables for return co-ordinates.
return_guardian_ur as above ...

claws

What about if they don't have any moves remaining?

Is it possible for the same unit to be processed multiple times by a micro ai?
If so then yes I suppose it could happen.
Bit doubtful tho' TBH because this issue only arises when I use the version that doesn't have return_x, return_y keys inside ...

I suspect this would work better if you just used simple scalar variables instead of container variables, since Wesnoth tends to get confused if you try to do complicated stuff with container variables.


Good point: that'll be my plan B if I can't get this to work with unit variables - but it'll ...

You say you want location dependent ability, but your test is about variable dependent one.


Ah... my mistake.
I thought:
[filter]
formula="x = wml_vars.return_guardian.return_x and y = wml_vars.return_guardian.return_y"
[/filter]

Would dynamically test if the unit's current x,y location ...

I guess this is complaining that line 12: if guardian_u.variables.return_guardian then references an empty guardian_u ?


Yes, AH.get_units_with_moves() might not return any units. In that case guardian_u will be nil . You need to check for this before trying to do anything with guardian_u ...

Am setting up a location dependent regen ability but it doesn't work. :annoyed:

Here's the ability definition:

#define ABILITY_WOODWARD_REGENERATES AMOUNT
[regenerate]
[filter]
formula="x = wml_vars.return_guardian.return_x and y = wml_vars.return_guardian.return_y"
[/filter]
value={AMOUNT ...

Baffled again. :annoyed:

Plan was to simplify working version of return_guardian_uv to only use unit vars, partly as a lua exercise.
However I'm stuck.

Edit: using v1.19.27 on windows.

Here's the working code:

mai_return_guardian_uv.lua
--[[ mai_return_guardian_uv.lua

Modified return ...

...So instead of writing this:


return_loc_uv[1], return_loc_uv[2] = guardian_uv.variables.return_guardian.return_x, guardian_uv.variables.return_guardian.return_y


You could write this:


return_loc_uv = { guardian_uv.variables.return_guardian.return_x, guardian_uv.variables.return_guardian ...

Well, got it working much quicker (and more easily) than expected. ^_^

Not exhaustively tested but seems to work as expected plus no errors. :shock:

Only niggle was lua didn't like testing for the nonexistence of a variable inside a nonexistent container (fair enough I suppose) so I just tested ...