Dealing with characters and corpses

[Pi-C]

I've lately developed a morbid interest in corpses (especially corpses of characters) and want to know as much as possible about them. But there are some limitations that make it impossible to get the data I need:

Usually, a character will be connected to a player. When a player is killed,

• on_pre_player_ died,
• on_player_ died,
• on_entity_ died, and
• on_post_entity_ died

will be raised in this order. The event data of on_player_died contains the player index, and this can be used to get player.character and player.character.unit_number. When on_player_died has run, the player is removed from the character. When on_entity_died triggers for the player's character, one can still access entity.unit_number, but entity.player is nil. I deal with this by storing event.player_index and player.character.unit_number in on_player_died, and adding the other data in on_entity_died. Splitting it up like this is not nice, but it works.

But characters are not necessarily connected to a player. They could be completely independent entities (these are not relevant in my case, just mentioning them for the sake of completeness), and they may also be associated with a player. I want to get character.associated_player, but there is no way to do that:

• on_player_died won't be triggered when an associated character dies.
• In on_entity_died, both entity.player and entity.associated_player are nil.
• From on_post_entity_died, one could get corpse.character_corpse_player_index, but there is no equivalent for associated players.

One way to cope with that would be to keep track of which characters get associated with a player. However, as there is no event for that there is no reliable way to be notified when that happens. (I could add a remote function that other mods call when they associate a character with a player, but other modders would have to know about it.) Another way would be to go over game.players and store the results of player.get_associated_characters() in regular intervals (on_nth_tick). That's also not reliable because a character may be associated with a player and killed before the association has been stored. (The alternative is to minimize the intervals by acting in on_tick instead, but I don't want to do that for obvious reasons.)

In my opinion, it would be best if the problem was solved in vanilla. I've several ideas:

• Add corpse.character_corpse_associated_player_index (on_post_entity_died). This seems to be the easiest way.
• Raise on_entity_died before on_player_died. Hopefully, this would preserve entity.player and entity.associated_player, but it could have unintended effects.
• Add a new event on_pre_entity_died, which would be raised before on_player_died.
• Add new events on_player_associated_character and on_player_disassociated_character.

I don't want to make a modding interface request for all of these, though, because each of these suggestions would solve my problem. What do you think I should ask for?

[Pi-C]

When on_entity_died triggers for the player's character, one can still access entity.unit_number, but entity.player is nil. …

But characters are not necessarily connected to a player. They could be completely independent entities (these are not relevant in my case, just mentioning them for the sake of completeness), and they may also be associated with a player. I want to get character.associated_player, but there is no way to do that:

• …
• In on_entity_died, both entity.player and entity.associated_player are nil.
• …

Found the solution myself now: If a character has been killed, on_entity_died won't have entity.player, but if there was an associated character, entity.associated_player won't be nil. So why didn't it work as expected? Quite simple: There was another mod active which would disassociate player and character in on_entity_died. Unfortunately, that mod was loading before mine, so it would remove the association before I had a chance to get the associated player …