Posts: 1,981
Threads: 63
Joined: Jun 2010
Hi all,
It's not the first time I have this bug...
Sometimes I need 11 bullets in survivor with SMG to kill a guy... Or 14 bullets for example when they don't pickup any health (according to accuracy stats)
How is that possible ? Accuracy stats are wrong ? Maybe I missed something
Thanks
Posts: 57
Threads: 2
Joined: Jun 2010
Posts: 739
Threads: 20
Joined: Jun 2010
Posts: 82
Threads: 5
Joined: Jun 2010
IF SP33DY GONZALES CAN'T USE AN SMG, YOU KNOW THE GAME'S BROKEN
Posts: 1,981
Threads: 63
Joined: Jun 2010
:D yes in fact, the server seems to "cancel" some hits
Posts: 739
Threads: 20
Joined: Jun 2010
(26 Feb 11, 12:16PM)Luc@s Wrote: :D yes in fact, the server seems to "cancel" some hits
yes, but why?
Posts: 1,718
Threads: 169
Joined: Jun 2010
because chuck norris preferes CS:S...?
Posts: 1,823
Threads: 20
Joined: Jun 2010
Eleven bullets to kill 100/100 with SMG is normal, but fourteen is not. If you want to test, make yourself a cheat client and use it on your own server with anticheat. You'll notice the dropped hits, but see your own accuracy still go up.
Posts: 684
Threads: 36
Joined: Jun 2010
As I promised to Sebacious, I will check this soon.
The system needs a better tuning.
Posts: 739
Threads: 20
Joined: Jun 2010
(26 Feb 11, 02:12PM)Brahma Wrote: As I promised to Sebacious, I will check this soon.
The system needs a better tuning. sounds good
Posts: 1,981
Threads: 63
Joined: Jun 2010
(26 Feb 11, 01:10PM)Gibstick Wrote: Eleven bullets to kill 100/100 with SMG is normal, but fourteen is not. If you want to test, make yourself a cheat client and use it on your own server with anticheat. You'll notice the dropped hits, but see your own accuracy still go up. 14 on CTF
But 11 on survivor :p
Yes I already noticed huge shots drops when I made a server-side accuracy calculator.
Even on my own connection ?!
Posts: 153
Threads: 3
Joined: Jul 2010
Posts: 684
Threads: 36
Joined: Jun 2010
(26 Feb 11, 05:15PM)Luc@s Wrote: Yes I already noticed huge shots drops when I made a server-side accuracy calculator.
wait wait wait... "server-side accuracy calculator" means a custom server without anticheats? :O
Posts: 3,780
Threads: 33
Joined: Jun 2010
It is largely lag-based, but I find it happens against bots as well...?
Posts: 684
Threads: 36
Joined: Jun 2010
"bots as well"?!?!?!
* Brahma suicides
Posts: 1,981
Threads: 63
Joined: Jun 2010
(26 Feb 11, 06:36PM)Brahma Wrote: (26 Feb 11, 05:15PM)Luc@s Wrote: Yes I already noticed huge shots drops when I made a server-side accuracy calculator.
wait wait wait... "server-side accuracy calculator" means a custom server without anticheats? :O
I code many mods for AC even if I don't use them.. but I made this server-side acc. calculator because I need to get accuracy from the server for a project I'm currently working on.
Posts: 684
Threads: 36
Joined: Jun 2010
So, these drops have no relation to the anticheats!?!??!
Posts: 1,981
Threads: 63
Joined: Jun 2010
26 Feb 11, 10:49PM
(This post was last modified: 26 Feb 11, 10:54PM by Luc@s.)
(26 Feb 11, 10:41PM)Brahma Wrote: So, these drops have no relation to the anticheats!?!??!
if these shots drops are related to the problem we are talking about here, and it's really likely, then yes, it shouldn't have relation with the anticheat.
But it needs to be confirmed... :/ I thought about the anticheat because i saw it could invalidate some actions... but you must be right
EDIT : maybe I just failed my accuracy code :P I'm going to do some tests
Posts: 3,780
Threads: 33
Joined: Jun 2010
27 Feb 11, 07:50AM
(This post was last modified: 27 Feb 11, 07:51AM by V-Man.)
I find it happens to me if my client has a sudden fps drop due to some external thing like an antivirus self-updating. Maybe the hitsound doesn't register a few of those bullets (nearly point-blank, no joke) and those bots have nearly 100 armor.
Posts: 1,981
Threads: 63
Joined: Jun 2010
27 Feb 11, 11:31AM
(This post was last modified: 27 Feb 11, 02:07PM by Luc@s.)
Ok I think I found the problem (in void processevent(client *c, shotevent &e), serverevents.h)
if(!gs.isalive(gamemillis) ||
e.gun<GUN_KNIFE || e.gun>=NUMGUNS ||
wait<gs.gunwait[e.gun] ||
gs.mag[e.gun]<=0)
{
return;
}
// and especially this doesn't work as it should do
wait<gs.gunwait[e.gun]
Server looks to think many shots are too fasts
(I tested, displaying warning messages when "wait<gs.gunwait[e.gun]" was true)
It happens with Rifle, SMG and sometimes pistol if you shot fast;
I'm also surprised this is not part of the anticheat : if it is also in the anticheat code, it must be redundant. And anyway there are too many false positives ! (sometimes, about 10 % !)
Posts: 799
Threads: 52
Joined: Jan 2011
27 Feb 11, 03:21PM
(This post was last modified: 27 Feb 11, 03:22PM by Roflcopter.)
(27 Feb 11, 11:31AM)Luc@s Wrote: ...
Am I understanding this correctly as: If a player fires two AR bullets -- the first arrives after 130ms, the second after 240ms -- the first will be after more than 120ms delay so will be accepted but the second will be only 110ms later so will be discarded. If this is the case it is very broken but simple to fix.
Am I right in thinking it can simply be fixed by changing:
Quote:wait<gs.gunwait[e.gun]
to
Quote:wait+40<gs.gunwait[e.gun]
which is far more permissible?
Posts: 739
Threads: 20
Joined: Jun 2010
Posts: 441
Threads: 13
Joined: Jul 2010
Oh god. This works so good once the bugfix has been applied to the server :D!
Everthing hits :)
Posts: 684
Threads: 36
Joined: Jun 2010
Yes... this piece of code looks like strange (lag sensible) and very redundant with the anticheats (since the anticheats have a better code to take into account lag/cheat in the shot).
But this piece of code is there from a long time... (before 1.1, and before I entered the devs team). And this means the 1.0 must be dropping shots as well.
But I am not sure... I need to check this. And unfortunately I cannot check it now.
@ Lucas and Roflcopter
Can you check carefully if this code is really necessary, or if the dix proposed by Roflcopter can fix it (wait+40<gs.gunwait[e.gun])?
I believe this is too much sensible to lag effects... and I also believe this piece of code could be simply ripped.
Posts: 1,981
Threads: 63
Joined: Jun 2010
27 Feb 11, 04:29PM
(This post was last modified: 27 Feb 11, 04:29PM by Luc@s.)
it should be directly ripped then... but I can test if you want
Posts: 684
Threads: 36
Joined: Jun 2010
Yes... please!
Talk to flowtron... I have no access to the sources for the next week, but I believe you could rip/change this piece of code and test with the current full official game (including anticheats).
Congrats to you, for your effort in improving this game!
Posts: 799
Threads: 52
Joined: Jan 2011
Yes -- this completely fixes other players' problems in a test server I made for SVN. We played with some UK and Australian players with the check completely removed and had no issues. Bullet registration seems, for once, perfect.
Luc@s, thank you!
Posts: 1,823
Threads: 20
Joined: Jun 2010
Your SVN server probably didn't have anticheat, did it? Do we need to compile a server with this fix and anticheat to see if anticheat is doing anything?
Posts: 799
Threads: 52
Joined: Jan 2011
(27 Feb 11, 05:41PM)Gibstick Wrote: Your SVN server probably didn't have anticheat, did it? Do we need to compile a server with this fix and anticheat to see if anticheat is doing anything?
It was recompiled by me -- that means there won't have been anti-cheat doesn't it?
I recompiled the server without any modifications too and the result was a definite improvement we felt. That is, it's hard to say how much this particular line of code is contributing. I think the anticheat may be at fault some of the time too, though.
Posts: 1,981
Threads: 63
Joined: Jun 2010
(27 Feb 11, 05:03PM)Roflcopter Wrote: [...] ok :)
Brahma Wrote:[...] np ;)
Ok 40 is not enough (but 50 is). But is it really useful with the anticheat ?
|