Improve behaviour on failure to monster teleport failure (#3839).

It's fairly easy to reproduce by wizmoding Zig:1 full of plants and
then teleporting a monster in there: it will probably end up in a wall.
As the mantis issue says, this will basically never happen in a real game,
but if it were to do so, just keeping it in place (with an appropriate
message seems like the best behaviour).

1 files changed, 4 insertions, 9 deletions

diff --git a/crawl-ref/source/teleport.cc b/crawl-ref/source/teleport.cc
index 92e18e03ee..18a6dc0f79 100644
--- a/crawl-ref/source/teleport.cc
+++ b/crawl-ref/source/teleport.cc
@@ -303,16 +303,11 @@ void monster_teleport(monster* mons, bool instan, bool silent)
 
     coord_def newpos;
 
-    if (newpos.origin())
-        _monster_random_space(mons, newpos, !mons->wont_attack());
-
-    // XXX: If the above function didn't find a good spot, return now
-    // rather than continue by slotting the monster (presumably)
-    // back into its old location (previous behaviour). This seems
-    // to be much cleaner and safer than relying on what appears to
-    // have been a mistake.
-    if (newpos.origin())
+    if (!_monster_random_space(mons, newpos, !mons->wont_attack()))
+    {
+        simple_monster_message(mons, " flickers for a moment.");
         return;
+    }
 
     if (!silent)
         simple_monster_message(mons, " disappears!");