no_leader=yes broken if the map file has a starting position for this side

[sevu]

This are three related bugs, one is a regression, one is fixed in 1.13.
Following situation. You have a map designed for 6 players, with six starting positions.
You have two scenario files which use this map

One scenario has a seventh side for an AI unit with no_leader=yes and hidden=yes. This one

• shows up as 6p map in MP Create
• shows 6 player slot in MP Staging (the screen after MP Create)
• In the game the seventh side has no leader. As expected

The other scenario file is for three players, and the AI is side 4. In difference to the other scenarios the sides are 1-3, 4 is the AI. The leaders for side 1-3 are placed with the x,y keys, ignoring the starting position defined in the map file. Side 4 has no_leader=yes and hidden=yes.

• shows up as 5p in MP Create. That's a bug. Why assumes it 5 players? It could be because side 3 starts at the position which is in the map file the starting position of side 5. But it doesn't seem to be, happens when using a random other place too. Same in 1.12.
• shows 3 player slots in MP Staging. In 1.12 it wrongly showed 1,2,3 and 5,6. 4 probably not because it is now the hidden AI side. Happily this is already fixed.
• In the game the AI side has now a leader, starting at the position which is defined in the map for side 4. This does not happen in 1.12

That's the UI part which I mean in MP Create, in case it's unclear.

[gfgtdf]

I an not 100% sure what this report is about so i'll just say what i know:

The beviour of no_leader was changed in 1.13 during the spmp patch:

It now just means that the mp initilisation code will not generate a default leader. This means if your side has a type= defined, the presence of type= will take priority over the no_leader attribute. in 1.13 no_leader defaults to true for [side]s in [scenario] but not for [side]s in [multiplayer]. In particular no_leader should no longer be used in sp, just omit the type= attribute.

shows up as 5p in MP Create. That's a bug. Why assumes it 5 players?

It probaly does a calulcation like this: 1) 6 starting positions and one [side] with controller=ai gives 6-1=5 sides.

[sevu]

The Ai side has no type= attribute either... It pretty much has nothing besides some [unit]s
Full code looks like this: https://bpaste.net/show/0c01247fcf2a

I have a look at your change in an hour.

[Vultraz]

Looks like that number is the number of sides that a player is allowed to control:

		num_players_ = 0;
		for(const config& scenario : data_.child_range("side")) {
			if(scenario["allow_player"].to_bool(true)) {
				++num_players_;
			}
		}

I'm not sure how that's the wrong behavior.

[gfgtdf]

what he is surpoised about is that the game engine generates sides for each starting position in the map, so if your scenario has 4 [side]s but your map has 6 starting positions, the engine will add 2 [side]s to the scenario wml.

[Vultraz]

Well that kinda makes sense. Why have a starting position without a side.

[gfgtdf]

Well that kinda makes sense. Why have a starting position without a side.

in his case becasue he wanted to reuse a map file for multiple mp scenarios (one beeing for 3 players the other beeing for 6 players)

[gfgtdf]

i don't really have a strong opinion on whether to keep that behviour or not, but in any case it should be properly documented.

[Vultraz]

BTW, I also don't understand the bug he says is fixed.

[gfgtdf]

this is marked as fixed becasue the main part 'no_leader=yes broken' is fixed, #2605 is still open.

[jyrkive]

I have now merged #2605 as well.