make ammo effect able to run EoC

[GuardianDll]

Summary

None

Purpose of change

I don't like how new damage types are used only because they can run EoC on attack

Describe the solution

Add ammo_effect field, that runs EoC

Testing

That's the problem, i use next EoC to test thing

{
  "type": "effect_on_condition",
  "id": "EOC_DEBUG_MESSAGE",
  "effect": [
    { "u_message": "DEBUG MESSAGE U.  Alpha: <u_name>, Beta: <npc_name>; <context_val:proj_damage>" },
    { "npc_message": "DEBUG MESSAGE BPC.  Alpha: <u_name>, Beta: <npc_name>; <context_val:proj_damage>" },
    { "message": "DEBUG MESSAGE GLOBAL.  Alpha: <u_name>, Beta: <npc_name>; <context_val:proj_damage>" }
  ]
},

but for hellish reason result looks like this:

So u_message is not run, despite u_name being assigned correctly. Any another u_ effects do not apply. I have no idea why it happens, my code uses the same tools as any another eoc-running function. I suspect npc_ do not work correctly either

Secondly, because of projectile ability to miss the target, EoC can be built either with beta talker as victim, or without beta talker whatsoever. It complicates the thing, because user is not able to use npc_ functions, because, well, we don't have an effect to check is beta talker presented (and i don't even know can such effect exist).

---

UPD:

Andrei helped to solve both issues, so it's good now (theoretically)

Additional context

@andrei8l sorry to bother you, but may i ask for some help?

[andrei8l]

@andrei8l sorry to bother you, but may i ask for some help?

I'm assuming you're asking why only one message is displayed.
First issue is that talker implementations were split into talker_X and talker_X_const without const correctness throughout the code. This should "fix" u_message:

diff --git a/src/npctalk.cpp b/src/npctalk.cpp
index 94f785f0dc..cf06a46029 100644
--- a/src/npctalk.cpp
+++ b/src/npctalk.cpp
@@ -4330,11 +4330,11 @@ talk_effect_fun_t::func f_message( const JsonObject &jo, std::string_view member
     return [snip_id, message, outdoor_only, sound, snippet, same_snippet, type_string,
                      popup_msg, popup_w_interrupt_query_msg, interrupt_type, global, is_npc]
     ( dialogue const & d ) {
-        Character *target;
+        Character const *target;
         if( global ) {
             target = &get_player_character();
         } else {
-            target = d.actor( is_npc )->get_character();
+            target = static_cast<talker const*>( d.actor( is_npc ) )->get_character();
         }
         if( !target || target->is_npc() ) {
             return;

Second issue is that npc_message only works for Characters (not monsters) and is explicitly disabled for... NPCs. I have no idea why.

Cataclysm-DDA/src/npctalk.cpp

Lines 4339 to 4341 in bfeb1ff

	if( !target || target->is_npc() ) {
	return;
	}

because, well, we don't have an effect to check is beta talker presented (and i don't even know can such effect exist).

If you meant a condition, then try this

diff --git a/src/condition.cpp b/src/condition.cpp
index 41e06c54f8..8a5321553f 100644
--- a/src/condition.cpp
+++ b/src/condition.cpp
@@ -1314,6 +1314,13 @@ conditional_t::func f_player_see( bool is_npc )
     };
 }
 
+conditional_t::func f_has_beta()
+{
+    return []( dialogue const & d ) {
+        return d.has_beta;
+    };
+}
+
 conditional_t::func f_no_assigned_mission()
 {
     return []( dialogue const & d ) {
@@ -2554,6 +2561,7 @@ parsers_simple = {
     {"u_is_furniture", "npc_is_furniture", &conditional_fun::f_is_furniture },
     {"has_ammo", &conditional_fun::f_has_ammo },
     {"player_see_u", "player_see_npc", &conditional_fun::f_player_see },
+    {"has_beta", &conditional_fun::f_has_beta },
 };
 
 conditional_t::conditional_t( const JsonObject &jo )

[GuardianDll]

I'm assuming you're asking why only one message is displayed

I didn't expect the issue would be in f_message. I also tried to use u_pain() to test, and also got no effect - does it mean it also has similar issue? how many functions has such issue overall?

If you meant a condition, then try this

Yes, but would it even work? In order to check beta exists you need to reach beta, which is nullptr, no?
I'll test it and confirm

UPD: no, it works as you expect it, and doesn't cause any errors. Thank you!
I still have question about u_pain(), does all u_ functions require rewriting to make them consistent?