Stun events leak to unrelated games

[ProditorMagnus]

Game and System Information

• Version: 1.17.17

Description of the bug

After starting new game without any units with stun and using inspect, event list includes

__empty_lua_event=yes
first_time_only=no
name="attacker_hits"
nonserializable=yes
[lua]
	code="<function>"
[/lua]
[filter_attack]
	special_type_active="stun"
[/filter_attack]

Steps to reproduce the behavior

No response

Expected behavior

Event is included to game only when there is reason for it, such as unit placed with that ability.

Additional context

No response

[gfgtdf]

Does it actually break the implementation? If it's really just function="<code>" i don't see how it could work.

[ProditorMagnus]

I would assume it was tested when this version of event was created. Whether it works or not doesnt matter for this issue context.

[Wedge009]

I don't know the significance of the code="<function>" events, but it looks like they were being introduced from at least 1.17.10 (I compared with current 1.16.9 release, starting with a fresh Tutorial campaign and inspecting the events). There is a seemingly irrelevant die event with this lua code.

For the stun-filtered attacker_hits and defender_hits events, these seemed to have been introduced between 1.17.12 and 1.17.13, at cc40d5c by @CelticMinstrel.

It looks like it comes from the move of the stun lua code into data/lua/stun.lua. So maybe it's necessary to make the merfolk units part of the core even if the event leak is not desirable?

[Wedge009]

As of 1.17.18+dev (71e83f7), I see four additional events over 1.16.9 release:

• An additional die event, I think from feeding.lua in Refactor the game event system to be based on Lua at the core #5663.
• The aforementioned attacker_hits and defender_hits stun events.
• An additional side turn end event, also part of data/lua/stun.lua

My guess is the <function> is just how the inspect module renders the lua code but is otherwise functional. Perhaps the question specific to this issue is how lua functions should be rendered in the inspection dialogue, if at all. But the title of #5663 appears to suggest that events should include lua code.

[gfgtdf]

Oh right, I totally didn't see that this is just about inspect and not about savefiles.

Then I'd say this is not a problem. Maybe hide it from Inspect if it's annoying to have it there?

@ProditorMagnus could you please elaborate why you think that this is a problem?

[CelticMinstrel]

Does it actually break the implementation? If it's really just function="<code>" i don't see how it could work.

That's not the literal definition of the event. The event is registered with the Lua API (wesnoth.game_events.add). The gamestate inspector does its best to try and serialize it but is unable to decompile Lua code, so you get the string "<code>" instead.

Anyway, why exactly is this an issue? The stun event is clearly filtered so that it'll only trigger when a unit with stun attacks. It may be present in all games but it won't ever trigger if you don't have a stun unit. (If it did, that would be a bug.) I'm very tempted to close this as "won't fix".

[ProditorMagnus]

I suppose its acceptable then. But when it is global event it should be at least documented that stun is reserved identifier now.

[CelticMinstrel]

What do you mean by "reserved identifier" in this context?

[ProditorMagnus]

https://wiki.wesnoth.org/AbilitiesWML#The_.5Bspecials.5D_tag Any other tag is valid, but will result in a special that does nothing but report it is there.

Since stun will now be active no matter if its unit is in game, it should be listed.

[CelticMinstrel]

Ah, that's what you're talking about. Sure, it could be listed there. It should maybe also mention it's defined in stun.lua; that only matters to someone making a custom core, but still.