Zandronum

In Such problems that i recive of the 3.0 are just for me not good.

i've been experiencing the new release wich was my 16th birthday where i play DM, CTF and LMS for ZDoom wars experiences, just i've seen that i get lag sometimes, much lag outs, same as you say about shots do not register very well.

i also think that i've tried to play CTF Wads along of people with or without mods i've been noticing that i just see a lot of unexcepted things from projectiles and shots.

Maybe this ticket has to do with it?

I've noticed it too in the early beta versions. Related topic: https://zandronum.com/forum/viewtopic.php?f=4&t=7315
I never got around to getting a demo of it happening

Have to say I noticed this as well, I barely had any dropped shots in 2.1.2. and now I drop them regularly in 3.0. Thought cl_noprediction fixed that issue, but apparently it doesn't really do anything at all.

Yep, from shooting people behind walls way after they took cover to incredibly awful desyncing weapons. (They start firing way later and your ammo is long gone...) The netcode got worse and this is affecting competititve and non-competitive modes.

Did the competitive community not test 3.0 at all? If they did, why didn't they report this, especially if it seems to have existed for so long?

Sean wrote: ↑

Sun Oct 22, 2017 4:31 pm

Did the competitive community not test 3.0 at all? If they did, why didn't they report this, especially if it seems to have existed for so long?

Exactly. If it is as bad as claimed in this ticket, either the competitive community ignored our many calls for testing or something outside Zandronum changed and its not 3.0's fault (e.g. servers running on different hosts with a less stable connection).

Anyway, to debug this we need more information. First, you should double check that 2.1.2 actually works better than 3.0. For this, you need to test 2.1.2 and 3.0 under the same conditions (same client machines, some server hosts, some settings, etc.). If there are differences, please provide demos so that we have something to investigate.

Sean wrote: ↑

Sun Oct 22, 2017 4:31 pm

Did the competitive community not test 3.0 at all? If they did, why didn't they report this, especially if it seems to have existed for so long?

It's true that it was barely tested because the main server cluster that competitive players prefer was not TSPG at the time, and most likely still isn't.

As for testing this, not sure exactly what kind of information would help here. Server settings haven't changed over time so that's not one thing to look for hopefully. (In regards to NJ that is) I know many people have fiddled with their unlagged settings by now so that's one thing to try.

The biggest complaint is the shots not properly registering and when they do register it's at seemingly impossible moments such as hitting people behind walls. The latter mostly happens with people of high ping difference being able to shoot low ping players behind walls, but the other way around is also happening. Other than that, I think we need more people to chime in on this.

Sean wrote: ↑

Sun Oct 22, 2017 4:31 pm

Did the competitive community not test 3.0 at all? If they did, why didn't they report this, especially if it seems to have existed for so long?

With what active competitive scene are you referring to? The only stuff happening now is TDM tournaments.

It would be worthwhile to have TSPG allow for a short while the last beta build of 3.0 to be selected when running a server. At least with this, the tic buffer could be turned off and it would be easier to see if it is a cause or the cause of the reported complaints.

There were plenty of CTF games going on during the period where TSPG was forced to be 3.0, for example. Said games were always on Funcrusher, which stuck to 2.1.2 untile the final version.

Gotta agree here too, some shots i make seem to be BS and some times i miss when i knew i shouldve hit the enemy

Yup, priv was actually in a jolt of activity before 3.0's release. Konar didn't and still doesn't give a fuck about his servers and I don't know why AlexMax didn't just host 3.0.

I'm one of the last people to complain about dropped shots, but I have been dropping shots too often for my liking as well.

Ru5tK1ng wrote: ↑

Sun Oct 22, 2017 5:38 pm

It would be worthwhile to have TSPG allow for a short while the last beta build of 3.0 to be selected when running a server. At least with this, the tic buffer could be turned off and it would be easier to see if it is a cause or the cause of the reported complaints.

Agreed!

I can attest to the problems people have mentioned. There is definitely something going funky with the unlagged code. Here is a video I have of one instance of shots just straight up not hitting their target:

This was recorded from my POV demo, which matched up with what I saw in the game at the time it happened (and hence why I went back through the demo and re-recorded).
If you want to investigate the demos, the links are below:
Patience's perspective (same as above video) (low ping, ~60ms or less)
Alfonzo's perspective (high ping, ~220ms or more)
(It's on the first match map, so demo_skiptonextmap, then do so again if you wish to skip the intermission, then demo_skiptics 17100 to land about at the spot where the video takes place)

There have been a number of instances as well where I have somehow hit people clearly out of sight around corners, where I fired for the sake of it and ended up scoring a kill. Watching the spectator perspective of someone with high ping (like 110+) is truly a delight as you'll often see them shoot at thin air and people who were behind them or off to the side or something will take damage/die.

One thing I suspect may be part of it is if the unlagged tic buffer is too lenient. Take for example, in Counter-Strike: Global Offensive, the unlagged on a 64 tic server allows for a packet timestamp difference of 200ms. Meaning, if the shot would have been on target on a player within a 200ms timespan, then the server will register the hit. 200ms is a little less than the average human reaction time of 250ms, and CSGO players don't move all that fast to begin with. This is where I think the tic buffer could be too lenient on players with higher ping vs lower ping, although this really does not account for shots dropping when it's low ping vs high ping. This is also just theoretical observation though, and I would like to look through the source and see for myself, which leads to my next point: A couple of folks and myself actually want to investigate into this ourselves as well. While we can probably find the code on our own, if you have any pointers on where we could start looking at for unlagged code and such, that'd be a helpful head start. One thing that could certainly be started on is some improved network debugging tools: console commands (obviously marked as cheats) to throttle/choke packets and add latency, a means of showing what the client saw vs. the server, and so on, would probably go a long way in resolving netcode problems.

[EDIT] oh yeah, wads needed for the demos:

Does the problem happen with both "cl_ping_unlagged 0" and "cl_ping_unlagged 1"?

EDIT:

Decay wrote: ↑

Mon Oct 23, 2017 2:01 am

I'd also like to saying that bumping into other players, especially higher pinging players, is particularly atrocious. It feels like your orientation gets completely skewed and the other player's position ends up somewhere completely different.

This should not be any better in 2.1.2. Does anybody notice a difference between 2.1.2 and 3.0 in this regard?

Latest 3.0 Beta is available on TSPG now

Decay wrote: ↑

Wed Oct 25, 2017 1:11 am

After trying out the previous 3.0 beta debug version and turning sv_useticbuffer off, I felt there was a -MASSIVE- improvement. Granted, I had 18-20 ping, but every single shot that should've hit, actually hit. Felt like LAN.

Occasionally there would be player jittering, but the shot registry was a ton better.

Yep there seems to be a small jitter period for both players. For a moment the other player will jitter from the other player's POV and vice versa. This doesn't always happen but it's worrying...

Amount of shooting behind walls has also decreased to a very small minimum. However this version has a problem with your corpse sometimes teleporting to your spawn point, and flickering across the screen giving away enemy position.

Interestingly, I was able to hit Ivan and make shots on him that I used to make in 2.1.2, though I did notice at the same time as him when things started to desync, but it would fix itself during our duels most of the times. Ivan is someone in 3.0 that I struggle to play against on the basis that I have to hit him only during specific times, I can't hit him very easily and it goes towards a lot of higher pingers, shots don't hit like they should.

After playing in my tournament vs Ivan or anyone for nearly a month on 3.0, I can tell the difference in versions instantly.

Ivan wrote: ↑

Wed Oct 25, 2017 1:17 am

However this version has a problem with your corpse sometimes teleporting to your spawn point, and flickering across the screen giving away enemy position.

Now that I think of it, I actually recall the beta builds (specically the last few) interpolating teleports across the screen like that for some strange reason.

I also noticed an improvement between this and 3.0’s netcode, though I’m more curious as to why the tic buffer was rewritten anyway and not thoughly tested.