(20 Dec 14, 11:22AM)Luc@s Wrote: (20 Dec 14, 09:17AM)Sveark Wrote: (18 Dec 14, 10:48PM)Luc@s Wrote: the issue was fixed client side for 1.2 release.
(19 Dec 14, 01:20PM)Luc@s Wrote: the "old" patch fix is indeed useless
I had an opportunity to try Larry's patch on 5 different machines with different hardware configurations and data centres. The patch was applied to the 1.2.0.2 version of the server. The result is simple: in 4 cases the hit-drop counter showed a nonzero value. I had 0% of hit drops on the only server, the one with the best configuration (a 1 Gb/s port and an Intel Xeon E5-1660 v2 CPU). By the same token, the next configuration after the said one in a descending order was a 100 Mbit/s port with an Intel Core 2 Duo T7700 CPU. Maybe you indeed sorted the trouble out in your perfect world, but the reality gives an example of the contrary. The further word I hand to you.
If you tried different servers with the same patch, it makes no sense. And if you are trying to show that a poor connection and hardware can affect the gameplay, well, o'rly ?
What i want to see is a comparison between the official binaries and the "patched" server (and by patch, you have to be clear : do you just mean the anticheat was removed ?)
As roflcopter i could not detect hitdrops (and i had a bad internet connection at the time i performed the tests).
My methodology was to compare the amount of hits detected on my side to the amount of hits reported by the server.
I tried it on different servers : some servers i did not know and some servers i knew to be running with the anticheat.
The only drops i had were because of my ping when i shot someone who was already dead according to the server.
You have to explain your methodology better.
You are making things unreasonably complex. The methodology I used consists in counting shots that are ignored by the server. As you might know, there are 2 places in the source code that may cause hit drops in theory:
1) serverevents.h, 40-th line:
1.2.0.2 source code Wrote: if(!gs.isalive(gamemillis) ||
e.gun<GUN_KNIFE || e.gun>=NUMGUNS ||
wait<gs.gunwait[e.gun] ||
gs.mag[e.gun]<=0)
return;
2) serverevents.h, 175-th line:
1.2.0.2 source code Wrote:if(e.shot.millis<c->lastevent) { clearevent©; continue; }
One shot carries maximum of one hit (except for the shotgun), so dropping whole shot causes dropping potential hit. In my tests, I only counted the shots dropped by the first block of code. Also, I have to say I had a 6-th test machine where I had zero possible unhits. The quality of that VDS' connection was explicit shit with ping getting higher and lower in a very short period of time as well as with some packet drops on a 2-3 Mbit/s uplink load. However, on that server, I had the lowest (throughout the test) ping of 50 msecs. I don't think the ACAC is capable of affecting one's hit statistics, but I think that a relatively high ping and going beyond the gunwait deadline may somehow correlate.
The test clearly showed unhit existance, mainly for mid-ping (100 msecs) cases. Having this in mind, I come to believe Larry's hitfix patch is still viable, even for the 1.2.0.2 version.