From: divverent <divverent@d7cf8633-e32d-0410-b094-e92efae38249>
Date: Tue, 9 Oct 2007 21:35:38 +0000 (+0000)
Subject: last patch by LH fixed speedhack 1.0: -20ms + 30ms = 10ms, but moves by 30ms
X-Git-Tag: xonotic-v0.1.0preview~2860
X-Git-Url: https://git.rm.cloudns.org/?a=commitdiff_plain;h=2756f5666efa5ecd3855e3887412fe7eac61a9e5;p=xonotic%2Fdarkplaces.git

last patch by LH fixed speedhack 1.0: -20ms + 30ms = 10ms, but moves by 30ms
this patch fixed speedhack 2.0: (cl_movement 0) -10ms + (cl_movement 1) 20ms = 20ms client physics
and now also fixes speedhack 2.0a: (cl_movement 0) -10ms + (cl_movement 1) 20ms = 10ms server physics and 10ms client physics


git-svn-id: svn://svn.icculus.org/twilight/trunk/darkplaces@7619 d7cf8633-e32d-0410-b094-e92efae38249
---

diff --git a/sv_user.c b/sv_user.c
index 5fc53adc..70d3bafc 100644
--- a/sv_user.c
+++ b/sv_user.c
@@ -371,6 +371,8 @@ void SV_ClientThink (void)
 {
 	vec3_t v_angle;
 
+	//Con_Printf("clientthink for %ims\n", (int) (sv.frametime * 1000));
+
 	SV_ApplyClientMove();
 	// make sure the velocity is sane (not a NaN)
 	SV_CheckVelocity(host_client->edict);
@@ -628,6 +630,12 @@ void SV_ExecuteClientMoves(void)
 		}
 		// now copy the new move
 		host_client->cmd = sv_readmoves[sv_numreadmoves-1];
+		host_client->cmd.time = max(host_client->cmd.time, sv.time);
+			// physics will run up to sv.time, so allow no predicted moves
+			// before that otherwise, there is a speedhack by turning
+			// prediction on and off repeatedly on client side because the
+			// engine would run BOTH client and server physics for the same
+			// time
 		host_client->movesequence = 0;
 		// make sure that normal physics takes over immediately
 		host_client->clmovement_skipphysicsframes = 0;