Merge lp:~gunchleoc/widelands/bug-1619157-random-ai-none into lp:widelands

[GunChleoc]

Random AI no longer selects the empty AI.

Since this is clearly a regression and the fix is tiny, I am in favour of having this in Build 19.

[bunnybot (widelandsofficial)]

Continuous integration builds have changed state:

Travis build 1350. State: passed. Details: https://travis-ci.org/widelands/widelands/builds/161981847.
Appveyor build 1192. State: success. Details: https://ci.appveyor.com/project/widelands-dev/widelands/build/_gunchleoc_widelands_bug_1619157_random_ai_none-1192.

[TiborB (tiborb95)]

Looks good (code + testing), It seemed that "very weak" AI is picked less then other types, but maybe it was just coincidence...

review: Approve

[GunChleoc (gunchleoc)]

Yes, probably just coincidence.

@bunnybot merge

=== modified file 'src/logic/single_player_game_settings_provider.cc'
--- src/logic/single_player_game_settings_provider.cc	2016-08-04 15:49:05 +0000
+++ src/logic/single_player_game_settings_provider.cc	2016-09-22 17:42:29 +0000
@@ -142,7 +142,7 @@
 			do {
 				uint8_t random = (std::rand() % impls.size());  // Choose a random AI
 				it = impls.begin() + random;
-			} while ((*it)->name == "None");
+			} while ((*it)->name == "empty");
 		}
 		s.players[number].ai = (*it)->name;
 	}
 
=== modified file 'src/network/nethost.cc'
--- src/network/nethost.cc	2016-08-04 15:49:05 +0000
+++ src/network/nethost.cc	2016-09-22 17:42:29 +0000
@@ -180,7 +180,7 @@
 				do {
 					uint8_t random = (std::rand() % impls.size());  // Choose a random AI
 					it = impls.begin() + random;
-				} while ((*it)->name == "None");
+				} while ((*it)->name == "empty");
 				set_player_ai(number, (*it)->name, true);
 				newstate = PlayerSettings::stateComputer;
 				break;