netcode is the code that handles the net

i think everyone has their own ideas for how quakelive handles things wrong, I'm quite happy with it.

You're quite happy how quakelive handles things wrong?

still better than xerp )

No dude.

Yeah, I guess I'm happy if it handles something wrong, as long as I don't notice, or think about it. I think that's good enough, for me.

so quake live netcode is like the police killing minorities?

no, it's more like ISIS killing you over and over again.

actually, yes, in a way, it's a compromise. police are allowed to exercise their judgement in order to protect the law and generalizing is part of the way our intelligence works. innocent will always be found guilty to some level, and I think the discrimination is a cultural problem, not a police problem. vote trump to make our culture less racist.

i'm glad you decided to dedicate more of your brain to this than some appeal to the lowest common denominator audience members such as funnyb.

now, since you admit that it's a compromise, are you willing to admit the black people actually exist, and that their personal experiences, in many cases, are valid and justifiably legitimate?

if so, then maybe you are prepared to mature enough to accept that there are people who are mentally affected by netcode issues that don't trouble you, whether because of some personal difference, or difference of location, or some other variable.

don't be afraid to empathize with people who suffer. it doesn't have to mean you have to suffer with them, or that you must debase yourself due to some guilt of privilege. it can simply mean that you are capable of understanding experiences alternate to your own, and accepting that reality and the universe are not black and white.

it will make you a better person. you can do it. i believe in you.

how many you know is a trump supporter? If you had a pie chart of how many people around you/ you know/hangout with, are supporting trump what would the % for trump be? If you are traveling with a train that goes 90km/h and you need to cool down your train coupe, for how long you need to lower the window of your coupe to bring down the temperature down to a normal room temperature? Note that you cant remove any clothes, if doing so there is a black afro american with a huge dong getting the morning woodz.

There is a difference between an algorithm compromising for humans and humans compromising for humans. When it's an algorithm, the best you can do is shut the fuck up and blame yourself. This message is approved by Trump.

the algorithm is out of many people's control, and its effects vary by case, scenario, circumstances, etc.

whatever the 'best' you can do is does not change the reality that there are humans involved in the experience, and that not every human values your perspective similarly.

That's so true, it's moot. Basically dialectics, but less useful.

it isn't 'basically dialetics', and it's very bad behavior to use the process of labeling and marginalizing a diverse range of human experience so that you can disregard it.

at least allow yourself the intellectual honesty to recognize it for what it is before you disregard it.

The code that is made to simulate lan play on online servers where there is ping involved. Because of that code you don't have to lead your shots depending on your ping. Try installing vanilla quake 3 and connect to an online server and you will see that depending on your ping your shots are gonna lag (you fire and then 50ms later the shot will actually happen). Simulating no ping when there is actual ping is never perfect, it always requires some compromises. As stated above, different people dislike different things about ql netcode. For example ql netcode works really well for hitscan weapons till about 45-50 ping, but the difference between 10 and 50 ping rockets is drastic.

even worse for plasma :(

or machine gun :'[

pretty sure mg falls into the "hitscan" category, which meph is saying is NOT affected between 10-50ms.

I wouldnt say its not affected, just to too much

i suppose that's what i meant.

all true, except the first sentence, because I am pretty sure netcode is used on lan play too.

Of course. There will always be delay no matter how close the clients and the server are to eachother. No data transfer is instant. Sure you won't be able to feel the difference between ping 1 and 2, but the code works with all pings =).

I tend to notice problems with shotgun when the route/server is less than optimal. One other huge issue is how quakelive is unable to handle sound in high-action/high-load situations. These two always seem to go hand-in-hand to make the experience frustrating.

Yes I've seen grenades dance on my opponent's head, yes I've seen a match-deciding pointblank shotgun blast hit an invisible shield instead of killing my enemy, yes over the years I've seen all kinds of warps and lags and stutters and phantom shots miss when they should have hit...

...but for all it's flaws it still feels pretty fucking smooth and doesn't detract one bit from the idtech 3 engine's unparalleled feeling of being the engine the universe could be built with, and it still allows me to shoot some green robots of guys that are literally 1000s of kilometers away from the comfort of my own home with absolute satisfaction and guaranteed lulz, and that is something we tend to take for granted.

People like Weird who say they stopped playing because the netcode completely ruined the game for them are actually just whiny manchilds who are trying to blame their shortcomings on external factors. Fucking first world problems indeed.

I had to giggle about this invisible shield.

That invisible shield is legit though.

you can always play Q43A

http://www.esreality.com/post/2807027/18-popu...pid2807655

and paypal me the money and i`ll still release the source so `yall can do that tweak the 1 line thing

http://www.esreality.com/post/2805153/no-more...pid2807789

This would all be perfectly agreeable if the cpma mod didn't exist, with netcode so good that it allows for cross-continental play on competition level (same with reflex btw). QL's netcode certainly can't be called bad, but it does leave a sour taste to know that we could see rapha/evil or dahang/cypher weekly if id had managed to make it at least as good as their predecessor's.

doubt it, all these rejects like decent graphics to look at

http://1337mag.com/app/uploads/2015/04/166403...23e_b1.jpg

this

I don't think you grasp the grave travesty of the situation.

NetC0de is wrong.

Latency is wrong.

Losing is wrong.

Vote Trump, he will make neTcodE great again.

the datrumpy code

BAN WI-FI USERS!

because they ruin so many competitive games? no, they just are an annoyance in FFA where win/loss doesn't matter.

You sound like evil reptilian.

whodat? (c) J.T. Money 1999 bitches

would you play tennis with a racket?
Would you eat ice cream when it's hot out?



exactly.

Yes I've seen grenades dance on my opponent's head, yes I've seen a match-deciding pointblank shotgun blast hit an invisible shield instead of killing my enemy

I doubt those bugs have anything to do with the netcode.

Had the exact same issues, more than once(probably 100 times?), still agree with you. The netcode is great as it is. And I'm not a 20 ping player no.

I started to heavily dislike competing... because I felt like I could take on the top16 players at the time on lan (as btw I also did online quite often, not winning all the time but having a fairly decent record) but I couldn't do shit against a random guy on wifi or against a +forwarding LG heavy player on ping 50.

ayyyy boyz, top contender at that time on lan is here ^
your fucking ego is hilarious

love when euros that complain about 50 ping

I didn't mind playing in US tourneys on ping 150 to 225 against people who pinged 20 - 50. Heck I even played some fun games against Prozac and Unleached while they were in Japan on a nice stable 370 ping. And I also had some matches vs a few Australians on ping 999 (according to ql, it felt more like 500 tbh).
But I had huge issues playing on ping 20 and getting obliterated by an lg that was clearly missing me.

I've never seen LG any other way.

That just means you need to adapt your play style against some people. Yes wifi-warriors are out there, however some wifi connections are not as bad as most people make it out to be.

adjusting to someone else his connection.... ye... not gonna do that.
And you're wrong, ALL wifi connections are fucked up horrible for quake games.

You also can't alter your game to play against a wifi opponent. It's just worse than playing against actual cheats. Against actual cheats I'm usually the favorite to win the game, against wifi-warriors nothing works.

there are plenty of decent players who play decently consistently, regardless of networking-related troubles.

your sensitivity to these issues is a handicap, like a person missing a leg. some people missing a leg will strap on a prosthetic and push themselves to do better. you're the guy who cries in the shower about how his missing leg makes it harder for him to live life the way he wants to live life.

grow up, loser.

nah I'm the guy that wants to solve the issue for the rest of the community/gaming in general. I've moved on and virtually stopped playing.

I'm downstairs to my router, about 15-20 meters away and have used wifi, no issues at all, even against people they've not once complained about warp/lag etc.

I know for a fact that ain't true. I may be inactive but we've played a few games and I said you felt like you were on wifi. So at least once someone said it to you.
If people would know how wifi feels you'd get it a shitload more often as well.

confirmation bias.

? I didn't know he had a wifi router before we played.
It just felt like it, and my feeling is pretty much always correct (sometimes they have a torrent program running or multiple people on a single line etc which often feels somewhat similar (there are differences btw))

that all sounds very rigorous.

you still don't 'know' if he has a wireless router, nor if he plays quake live using a wireless connection to that wireless router. you're only assuming that it's true because it supports your preconceived notions.

i would like to see the evidence that your feeling is pretty much always correct, because my feeling, which is also pretty much always correct, is that your feeling is not pretty much always correct, but that you only admit to having feelings when your feelings are correct.

only if you present me proof that your feelings are pretty much always correct.
(bullshit argument)

And it's true that I can't "know" that he is on wifi. What is true however is that when I say that someone is on wifi they usually admit that they are. (or in some cases that they have torrents on or that there are multiple people using the same line)

regarding the veracity of my feelings, the quality of our arguments there is the same. i have directly countered your statement using the same argument, and so your 'bullshit argument' is moot.

interesting that you should use language such as 'admits', which implies guilt, as though a person using a connection to a wireless router is somehow a criminal. it's almost as though you have a preconceived notion that you seek evidence to support.

1) No the quality of the argument is very different.

I say he's full of shit because I told him myself that he was on wifi, and he claims never to have been told by anyone that he was on wifi. Even though he IS on wifi.
You tell me I'm full of shit because you don't trust my ability to see who is on wifi and who is not. Which is clearly not a useful argument.

2) How else would you name it when I claim something and someone else acknowledges that it actually is the case. It's kinda the definition of "admits" isn't it?

the quality of the argument is not different at all. we both have feelings that are pretty much always correct, and neither of us have any substantial or tangible evidence. now that this has been exposed, you should 'admit' the intellectually defunct nature of your egotistically arrogant masturbatory claim, and then move on and grow up.

to address point two, here is a definition of 'admit': confess to be true or to be the case, typically with reluctance.

i see no reluctance. i see somebody plainly stating that they are using wireless, and who is not convinced that there are issues with his specific connection. i see somebody who reinforces his claim to using a stable connection by citing third party experiences and feedback.

i see you bickering with this person using the same intangible and unfalsifiable nonsense. how sad for you to be so clueless to your own ineptitude, and to, as an untrained puppy, fight the collar of enlightenment i've placed around your neck.

calm down, weird. breathe. relax. it's okay to 'admit' when you are desperately deluding yourself. it's by this process of 'admission' that you can grow as a person. i believe in you. you can do it.

I have evidence of the actual guy in question agreeing with me on my claim. You don't have any evidence.

Typically I use words with their intended meaning.
You can bitch and whine about American redefining words to mean something that they do not. I'll stick to definitions.

Heck if you have a huge dislike for 'admit' then I can just use acknowledged.

My claim is acknowledged by the guy in question, your claim is not. Therefor my quality of my argument is stronger than yours.

admit
/ədˈmɪt/
verb (mainly transitive) -mits, -mitting, -mitted
1.
(may take a clause as object) to confess or acknowledge (a crime, mistake, etc)
2.
(may take a clause as object) to concede (the truth or validity of something)

there is no issue of nationality. you do not typically use words with their intended meaning. you didn't even bother to understand the definitions you cite in your reply.

here's the definition for concede:

admit that something is true or valid after first denying or resisting it.

you're plainly incorrect. 'admit' it. 'concede' it. please, so that you may stop embarrassing yourself.

you do not have evidence of the guy agreeing with you on your claim. you have evidence of a person stating that they use wireless, and of that person disagreeing with your claim that it is detrimental to other players' experiences.

learn to reason.

claim: All wifi sucks.
sTPHN: I use wifi, nobody has ever complained.
me: I've played you and complained afterwards about the quality of your connection. (you = sTPHN)
you(anonymous scared guy): confirmation bias
me: I didn't know he was on wifi.
you: crap argumentation about English including 'feeling'.
me: explaining why the 2 arguments are different.
you: going whining about English.
me: no you mean American.
you: hell no learn to reason.

I don't see how any of my claims are in danger because I use the word 'admit' rather than acknowledge. (which is perfectly fine English)

further, here is the definition of 'confess':

admit or state that one has committed a crime or is at fault in some way

i understand that english may not be your first language, and so i will forgive your incompetence in english argumentation.

the server is running a full game with out the renderer
the client is running a full game with the renderer

to keep them in sync, there are snapshot`s that are written at the end of every frame on the server and sent to all clients, which then conflict with client prediction which is used when your ping is above the servers max threshold, the first thing the server gets at each frame is the user commands sent from clients to the server

so when you ping is back to normal the client will resync to the server`s next frame and "teleport" to the synced destination

id games just run everything on server and client so everything ends up conflicting with each other whether you have bad ping or not

It's like when neo first realized what the matrix was and the agents are high ping russians who warp everywhere and hit everything.

Big hitboxes = good netcode, tight hitboxes = blame Syncerror for that missed rail shot

Hi rail_that_went_through_model_or_hit_something_that_should_not_get_hit

some of the people responding in this thread have stupid opinions. they want to make points that invalidate the experiences of others, so that they can celebrate their own ignorance or capacity for tolerance. here is some detailed documentation that will closely address your question:

http://www.ra.is/unlagged/

there are many different approaches to multiplayer fps networking. essentially, the server has to decide a gamestate between multiple clients. these clients are separated spatial and temporal dimensions. the server determining the gamestate from each client is an interesting challenge. some clients are laggy and update less frequently than other clients, whether because of some hardware or networking issue. some clients update inconsistently, for whatever reason.

there are benefits and detriments to any system that attempts to create a real-time experience for actions at a distance. while one benefit of unlagged netcode can be predictable hitscan point-and-click behavior, one detriment can be inconsistencies in other game behaviors, such as shots appearing to hit but not actually hitting, or being on the receiving end of shots that, from one client's perspective, may seem impossible or unbelievably difficult to make. part of the issue is that any given client's game state is not in the precisely same state as the server's when the server decided that damage was dealt, and so where the client may be some frames ahead of the state when damage was dealt, the player position is now advanced to behind a wall or at a difficult angle, and the damage registers.

some people seem to be extremely sensitive to this, and for them the experience of playing is spoiled. some people in this thread use that sensitivity to masturbatorially congratulate themselves for their own tolerance.

Doesn't matter what you think. I don't really care what netcode does and that's ok with me. Cheers everyone.

what i think does matter. your lack of care has been noted. thanks for sharing.

some people are stupid. they have opinions and their opinions are not like other opinions. one of the two opinions are masturbatory. or both.

some people are stupid opinionionated masturbators sensitively masturbating their opinions in threads and threading masturbation into opinions.

are you serious?! i can't even

do not even

You must be odd

God I hate people with opinions.

you and me LOL

But that is, of course, just your opinion

#govegan

I just prefer to have a solid sense of the latency. Ql doesn't give you that control, and even punishes those who do have solid ping and aim. Qls netcode doesn't break the game, but it's still more flawed than previous titles use of netcode. Although, I'm sure on the lan its absolutely solid.

So I sum it up this way:

Unlagged is flimsy plastic gameplay.

Quake, q2, and q3a is solid metal gameplay. Your ping is your responsibility.

Modern faggots prefer a disingenuous, lan-like experience which has goofy flaws (hence they be deemed faggots).

Your confusion is understandable, because it's such a loaded word, with a big history behind.

The base is that 'net code' is all the code that handles networked play. Every game you play with multiple machines (so not just split screen on the same one, but actual machines connected somehow) has it in some form. Quake uses the paradigm of a server controlling the game to which clients connect. The clients send the user commands, and the server replies with information like the position of entities, sound events, damage events etc.

Now, the main thing to understand is that latency is unavoidable and kills the game, and programmers have to hide it somehow. Latency on LAN is almost negligible, but online it quickly goes up. Consider that quake comes at a time where internet connections were slower and latency substantially higher than today.

So, there's a lot of trickery going on in any game played online, to try and give you the illusion that you are actually playing lag free.

The first concept, that is quite crucial to understand is client prediction. To exemplify, if you load the game and press forward, you immediately see yourself moving forward. In reality, your client is just showing you a prediction of your position. In the meanwhile, the client sends your command to the server, the server computes your next position and sends it back. At that point your client checks that everything is correct, and if it is, you will not notice anything. Problems in the network however may mean that your client shows you a prediction and then has to correct.

Now prediction has important applications, like your movement. Without it, you would feel like playing drunk, as all your movement would be delayed. So it's kind of a must have. Quake applies it to more things. If you pass over an item, you hear the pickup sound immediately. Again that's your client predicting it (there was a notorious bug where your client would incorrectly predict that and play the sound, but the server would then say you did not pickup the armor/mega, which would confuse the heck out of players).

Another important concept is backward reconciliation, which is where Quake Life differs from some of the earlier Quakes. When you fire a rail in Quake3, you have to aim slightly ahead of a moving target, to compensate for the time it takes for your command to reach the server. Quake Live has a form of 'anti-lag' or 'backward reconciliation' where instead you can fire the rail where you see the opponent, and when later on the server receives your command, it sort of 'checks what you had on screen' (doing some math) and if in that scenario your shot aligned, it assigns a hit.

This sounds very nice from your point of view, but it can be a disaster for the opponent, especially if your connection has high latency. To understand this, imagine you show yourself and quickly hide again. I have a shitty connection, but I shoot at you and the anti lag grants me a hit. By the time the information reaches you, you are well safe and sound behind the pillar, except that now the server says you are hit. How could you avoid that? Well you basically cannot, and it's confusing as fuck, because it's as if I am shooting through walls and you just have to suck it up.

So, dodging against a anti-lagged high ping opponent is very hard for certain weapons. You will see weird (as in the player who posts here from time to time) complaining about this especially for the lighting gun. And he's right, in that it's very unfair to run away from a high ping anti-lagged opponent that chases you with LG, because even from you point of view you are evading and hiding, he stills hits you and kills you, and that can even happen with you not being able to see the opponent well (as he's still behind the corner).

At the same time, with earlier Quakes, competitive matches required super long server searches, because a small difference in ping made a world of difference. Now with Quake Live it's easier, mainly because there are advantages and disadvantages for any ping.

yep.

Fantastic! Love the explanation.

Should be added that ql currently has issues with handling other players clientside.

In short, player models are being skewed slightly in the direction they are moving. This happens when the client tries to smooth out (interpolate) the other players movement.

This and some extrapolation issues causes the rail through's on dodging enemies and even happens on lan (can be fixed with higher tickrates when on low ping conditions).
It also causes rockets to appear to be closer to the player then they actually are, eg: rocket hits at players feet but does around 10-15dmg while a rocket farther away is doing 45-50dmg.
If you pay attention you will notice that you even on low ping can hit certain players behind or beside their player model with hitscan weapons (lg,rg,mg etc).

The only way to "fix" this is to either start using conditional client side hit detection or just outright removing any client side smoothing,prediction for player entities.


Reason why i feel confident in what im typing here is because i have managed to replicate the same issues on a different engine.

Did you implement unlagged on ioquake3?

No, i tend to stay away from the q3 sdk alltogether tbh.

Can you provide more details? I have been thinking of trying to do the same.

Not really, if you are interested in doing something similar you are best off learning the basics of it by yourself.

I am also rubbish at explaining the simplest of things. :)

I'm fairly confident that only happens with playerlean on.

Interpolating should be flawless and QL doesn't have any extrapolation of other clients (not unless you force timenudge on or run out of snapshots).
Extrapolating anything in an other way would break the unlagged code (which btw does feel pretty broken), only way to fix the unlagged code would be to implement the same extrapolation in there... Which reduces the effect of unlagged, which isn't happening.

It can be that the animations of the player-model do not stick exactly to the hitbox (which is the case for all models) but it's not strictly behind, although some models do seem to be further infront of the hitbox than others due to visual effects..

change the snaps and forcelanrate cvars

http://www.holysh1t.net/quake-live-commands-list/

snaps is probably per second, so knock it upto 100 for lan rate 1

probably get a snapshot every 10 frames, if your on fiber

if I'm on fiber (internet) then lanForceRate wouldn't do anything?
Because it detects that I'm not actually on LAN with the server.

Putting the sv_snaps to 100 probably means (not entirely sure) that your rate has to be above the 25000 that QL usually allows. In which case strange shit can start to happen (assuming you have it on 25000).

The lanForceRate ignores that 25000 cap and just sends regardless, but only if your IP makes sense to the server (aka you are on LAN with the server).

snaps probably just sync player origin physics, then projectile physics etc every X ms

your client needs the projectile spawn origin, then uses prediction, so if your ping is off, a snapshot will resync it, so you get a hit sound, but the hit has`nt registered serverside etc

lan mode normally just disables client prediction, which you could do if you had 20mbs fiber and playing duel or a few clients on your computer

You're not really making a lot of sense here m8.

lan mode = realtime syncing of all physics for all entities in a game
internet = syncing of physics every x ms depending on what your snap cvar is set to

if you have alot of clients on a server, all of the physics for all entities could easily reach 5mb of data to be sent to each client

there is no compression, like looking at a ascii model in note pad and all the point origins

ye but how would that come into play to the original issue of a lagging hitbox on lan?

it depends on your ping, and whether lan is syncing the physics in realtime

id think everyone still has v90 modems

use rtp instead of udp

http://www.cisco.com/c/en/us/td/docs/ios-xml/...ession.pdf

for modern games with more data to be sent

so then why do the lagging hitboxes occur on lan according to pitch?
If it's because, like you call it, "syncing in real time" is on?

I fail to see the logic that links your statements and the hitbox stuff together sorry...

hitboxes would only lag if all physic`s were not being synced in real time

hitboxes on a server would be linked to the player physic origin, if a client ping is too big, then the client would be using prediction, so then the client has to wait for another snapshot for the true physic origin to be synced to the server which would update the hitbox clientside aswell

if a snapshot misses projectile physic every X a mount of snapshots, then clientside prediction could think the force of another entity has moved the projectile, while server side the projectile was`nt moved

ye but pitch said there was an issue on lan.... Which I found strange.... But you claim that physics are being synced on lan..... And when they are in sync there are no issues...
Therefor I fail to see what your original post is about.

I can understand why online things might seem strange (although tbh your story doesn't really help me, but fortunately for you I am the evil god of netcode so I get that)

im not saying they are synced, im said it what the cvars probably do

entity physics will be synced one way or another

Physics are never in sync as long as you have a client/server model.
The only thing you can really do is try and lower the latency by upping the frame/tick-rate on server and upping the client framerate as much as possible. (in quake engines).

The same thing is applied to the player models if they are predicted on the client, entity keeps moving before it recieves another packet.

Add in the fact of small amount of aircontrol,rockets,spamming input etc and all of a sudden there is a lot of room for error.


This is completely dependent on the client doing some kind of interpolation or prediction though.

All we can really do is try and make some sense out of it imho.

I make sure to always use playerlean 0 combined with railtrail 0 to try and avoid that.

When it comes to animations going outside the hitbox that usually only happens when in death anim. Movement animation does not from what i can tell go out of bounds.

In worst case you can always use smaller models instead of keel or tankjr, which i have found they accentuate the issue a bit more.

Keep in mind that the issue im talking about is more prevelent in people with connctions that have a lot of ping jitter eg: when flooding your connection to the point where you start to get packetloss.

Oddly enough it happens a lot on servers where players are on lan condition aswell, so in both extreme cases.

So how do you get this bug on lan conditions then? Because I've never seen it in lan games.

I have seen very strange behavior on dodgy pings... but on lan not a single time. (well.... there was the hitbox 'capsule' which turned out to be a 'cylinder' which had some strange behavior for a while)

Only explenation i can come up for it happening on lan is low frame/tick-rate servers and the client trying to extrapolate the player movement.

Like i said, i replicated the behaviour in another engine, i fixed the issue with upping the tickrate tho this only fixed it on lan.

but ql doesn't extrapolate anything on lan. It's 100% interpolation.
(except then for the 2-5ms delay due to networking which have the 'client prediction' as MM called it (that you go forward instantaneously on your screen without having to wait for server validation).

So a low frame / tick-rate doesn't matter. The client should adjust to the server settings without flaw as to the location of the hitbox. (Of course they might've implemented stuff wrong... But in Q3 it's fine and only timenudge forces any kind of extrapolation.)
*edit: assuming that the LAN connection is stable of course.

Here is the thing, we don't actually know what ql does on the client as we do not have the source to check for ourselves.

Assuming it still works as q3 does while having a ton of changes to things like pmove just doesn't cut it. (wall clips, acceleration and jump behavior and the recent brush double jump bug)

Again, if you are trying to predict player models movement while having a lower tickrate that happens, you cannot update faster then the server can provide thus we predict.

If we are talking standard q3 behavior then no it does not do any extrapolation outside of timenudge, but this isn't q3.

right...... But you know that there is a bug on lan which nobody has ever complained about?
And you know that because in different engines there can be an issue with the tickrate?

Sounds pretty far fetched to me...
But hey I can't know how it's implemented. Although I can probably count the guys more knowledgable about this than myself on 1 hand. (ArQon, SyncError, Sponge, Rhea and John Carmack)

Bug has been mentioned several times, people have been complaining about it on their lan hosted servers (online play).
Last mentioned of it in competetive play that i can remember is rapha briefly talking about it on a qcon stream.

Judging by the sarcasm and overall acidic tone of your reply i would say we are done here.

"people have been complaining about it on their lan hosted servers (online play)." what?
How do you manage to put "a lan hosted servers" and (online play) in a single line?

What kinda setup are you talking about?
A client and a server on LAN which also connects to the inet, and the opponents of the lan client join in through inet or something?

Or do you mean that they use an online setup for an actual lan server? (all clients being on the same lan but with a server that is setup for internet play)

as for the bug being mentioned several times... link it to me, as I've never heard of it before.

I am too lazy to search it, but from my memory I can confirm that rapha said something along the lines of:
I had shots going through toxjq and later toxjq had shots going through dahang.

I also can confirm that for years and even without playerlean/timenudge/jitter I have to hitscan behind the model at pings 30- ms or laning my brother.

But there is no reason to further discuss it:
- QL ain't moddable, so netcode won't change
- they smeared the flaws of the netcode by disabling cl_timenudge on lans and latching several cvars like cg_smoothclients, cg_projectilenudge, g_laghax, sv_fps, etc.
- those effects only affect a small group of players with low pings, precise senses, fast reflexes and strong mechanical execution

Unlagged is a double edge sword for both high and low pingers. While you get hit around the corner by high pinger, he gets hit around the corner too. This happens because when you peek and fire around the corner he gets hit before he can see you peek. This is really annoying when playing low pinger and he rapes you with rockets dancing around the pillar. These unlaged problems really do go both ways. In OSP you stood no chance with higher ping because if enemy peeks fast and fires you can't hit him cos of delay on your shots.

Summary: high pingers hit you when you're already past the corner and low pingers hit you before you even see em peek.

That's a good point. I wonder though if it's a result of unlagged, or just basic latency. The low ping player peeks, shoots, and the high pinger will get hit before he sees things. That with or without unlagged code. Unlagged might accentuate it slightly, bit I guess the issue is inavoidable.

One argument is consistency. Without unlagged, you can get used to how the game feels at a certan ping, even if high. With unlagged you must factor the opponent ping as well, which pretty much makes infeasible unless you play a lot.

There's then the issue of defining fairness, but that's where facts end, I guess.

Yea well, when you take in consideration how incredibly complicated it all is its amazing how good it actually works. If somehow I was forced to play osp online again I think I would kill myslef :)

I used to play with 100+ and sometimes 300 in quake2-3. You're right, in that I got used to, and had completely different, but actually executable safe style with higher ping, corners and setups being sure death traps. The current netcode makes them much less deadly because I can deal more damage back, despite of how unreliable and glitchy this damage seems. Ironically, I never play with 300 anymore, but 110 tdm is quite manageable. yaw #tdmpickups

The bias in your post is not appreciated. Low-ping players do not suffer high-ping players, which is what your post suggests. Rather both classes have to tolerate the abnormalities that arise from networked play.

If we momentarily disregard ping and look at the imagined situation we notice that the peeking player was exposing himself. In other words, he was putting himself in a shootable position, regardless of the players' pings. How do you avoid being shot in shootable positions? Don't put yourself in shootable positions.

Let us suppose both players were playing with (a theoretical) zero ping. The peeker would have then realized he was shot as he was peeking. However, if the peeker was playing with low-ping, and the attacker with high-ping, the peeker would have realized he was shot after peeking. If we reverse the pings (peeker having high-ping, and the attacker low-ping) the peeker would have been shot before he peeked. (Rather before he'd realized he'd substantially exposed himself.)

All of the above assumes the attacker successfully hits his target. If the peek's duration is 0.35 seconds it's 0.35 seconds on the attacker's client, regardless of pings, and he has that time to notice, target and fire at the enemy-model. If the attacker misses the target on his client anti-lag won't magically "grant" a hit.

To surmise: the peeker was hit because he exposed himself and either:
a) the other player was expecting it and hit;
b) he had adequate reflexes to react to the surprising move, and hit anyway; or
c) he pre-fired and got lucky.

Anti-lag doesn't grant shots randomly, it just verifies them. The process of verification has some bizarre consequences (from the uncomprehending-player's point of view), but those consequences are just delayed or expedited, not untrue.

If a low-ping player loses an LG fight to a high-ping player it's simply because the high-ping player tracked his client's enemy-model better than the low-ping player tracked his client's enemy-model. If the low-ping player then goes on to deplore anti-lag he's simply being a sore loser. Though he wouldn't be wrong in claiming his opponent won solely because of anti-lag. If anti-lag wasn't implemented it would make no sense for the high-ping player to be tracking (aiming at) the enemy-model, and it's unlikely the high-ping player would have such precognitive prowess as to be able to track the enemy-model's future position. He'd hit a great deal less than the low-ping player, because humans don't have precognition skills. Any hits would be attributable to luck.

Weird and his ilk simply yearn for the days where a low-ping was significantly more advantageous than it is today. I word it this way because low-ping is still advantageous, if for no other reason than the fact that projectiles are still "lagged".

Also, I want to point out Weird's comment where he says he played on servers with a variety of high-pings (150-999). He failed to inform us of the quality of those particular games. From my experience: playing on servers with a ping exceeding ~200 results in extreme stuttering, where your client is constantly, temporarily locking up and displaying a "Connection Interrupted" message.

Yeah, this is true. My post was biased, but not because of an agenda I have to push but merely because of a narrative effort to explain the reaction of people who were used to Q3 when they started playing QL, which is part of the OP question.

I said things like 'you cannot avoid that! it sucks' to exemplify the kind of emotional reaction is from the perspective of the player that at a particular moment is suffering from the obfuscation of lag. It was by no mean a discussion about fairness and correctness. Or at least that was not the intention.

If we are to discuss that, I think OSP gave the high pinger a more consistent feel of the game, but it was also giving incredibly low changes of winning. QL gives you a less consistent experience, but is more fair overall. The search time for servers in competitive play is a clear indicator.

Is this better?

Weirds point is that gameplay gets heavily focused on hitscan weapoins (lg/rg) because of the combinaion of hitscan and predicted movement.

Having a "warpy" connection will cause the clients view to peak around a corner and display the position of a opponent before the position on the server has changed.

You can argue if that is a major issue or not, but it is in fact happening. Me personally i am in the camp that we need lag reconcilliation but someting is just wrong with how ql works atm.

Right now ql suffers from a string of unrealiable behaviour when it comes to both hitscan and projectile weapons. This is an issue for whatever reason does not exist to this degree in games like reflex or qw (both use lag reconcilliation).

Me personally i am in the camp that we need lag reconcilliation but someting is just wrong with how ql works atm.

Right now ql suffers from a string of unrealiable behaviour when it comes to both hitscan and projectile weapons. This is an issue for whatever reason does not exist to this degree in games like reflex or qw (both use lag reconcilliation).

I am a high ping player and I h8 it. I would kill for having weird's ping, but what u say is true and unbiased.

There are 2 parts of my problem.

The first one is that it heavily favors +forward LG players. The higher the average ping goes (of both players) the more powerful +forward LG becomes, all other styles get less interesting at the same time. Slowly leading to a more and more LG heavy game.

The 2nd one is that the wrong player is punished in case of dodgy connections. And that that player is "leading" in how the netcode deals with it.
So if you want to play a +forward game.... Then get a reasonably bad connection. If like me you want to play a more balanced game... Then you cannot ever play against a reasonably bad connection. Because you are pretty much forced to play +forward.

Like I said back when I played... I felt like I could take on the top16 at the time (both online and on lan) but I couldn't beat people who had a bad connection (even those who were just around the top200 mark.... it was just impossible to beat someone who played according to the netcode without altering your own game to fit to it).

Nowadays most players have developed aim heavy styles, mainly with instant hit weapons, as a result of the above. Which I said would happen in 2009 or 2010 (somewhere during the later stages of the beta)... And has since moved more and more towards that.

I must say though that the LG to 6 damage has insanely reduced this problem. However(!) that doesn't mean that it's not still there. And still the only way for people on bad connections to win games is to play overly aggressive. The 6dmg just made it less dominant than it was before.

My games hardly ever stuttered with a "connection interrupted" on there (although on the Australian ones the message was always there I think, but the game play fluently).
The only times that it does become really difficult is when you get hit by something with a lot of knockback. And on 150 - 200 ping games you can react to that pretty adequately but on higher ping games that becomes a real issue.

And I really don't care about having an advantage because of low ping. Even in my Q3 time I played almost all of my games on servers that favored my opponents.
I yearned for a server state which is correct and follows the same possible gameplays for all players.

In Q3 I could adapt to MY ping and have an extremely stable gaming experience as a result of that. I could basically do the same stuff on every server and I was very alround in doing so (I played a lot of TDM & CTF in Q3 and switched to duel somewhere in between those only for a limited time.... although it was several years in total.. ;))
In QL one "should" have to adapt to the opponent's ping. But you don't control the actions of the opponent... which makes adapting to it INSANELY difficult. As a result most people nowadays only focus on hitting their shots, and the most stable way to do so is with the unlagged weapons.

The adapting to your own ping in QL only works if your ping is shit. In which case you will start to automatically play retardedly digressive (which leads to winning a lot of zotacs / gdata / 125fps cups). If you look at the results of those cups... and compare internet connections and playstyles... it should be abundantly clear that QL's netcode determines who wins and only very rarely cooller or killsen managed to sneak in a win.

[+] - I don't see server searches being mentioned much at all in QL netcode discussions.

Quake players worst nightmare.

netcode is space time continuum for quake. :)

Can someone explain how did the netcode differ between CPMA and QL? Also, why did they never get to implement xerp in QL?

I havn't got the technical explanation but in short xerp made the netcode too good and because xerp mainly helped hitscan weps people didn't like the change to a hitscan heavy style.

Q43A has all hitscans removed, except on the LG, every other weapon uses full projectile`s with physics and full clip boxes. and they are all 100% serverside

But hitscan weapons are the best weapons. :(

it's funny how ungrateful and delusional humans can be... some race in ferraris vs u in a Lada, but they will still whine and find excuses and they will demand u to race with ur tires flat as well... ts ts ts

https://www.twitch.tv/philian89/v/30625718


GREAT GAME XD

Ok

a good example of what's possible with backwards reconciliation

Ahah that's a great reaction.

Bitstream is a ffa player who tends to have warp problems from time to time.

Bitstream if you're reading this, make sure you have \cg_lagometer 2 set in your console. There is an option to enable it in the settings menu as well. It will show you, for the most part, what's happening with your connection. If youre seeing lots of spikes or red or yellow something is wrong with your shit.

Yeah, I have that enabled. I've decided that from now on, the moment I see my ping begin to spike a lot, I will join the spectators until it stabilizes again.

That would definitely be cool very polite of you, sometimes the people playing may not even care though. Just communicate!

Any idea what's causing the spikes?

My connection is just really bad and inconsistent, and I don't have ethernet so I have to use wireless.

need mtu on 510 for wireless, to many packet drops, send normal packets, and the reply back to say your computer has receive the packet takes twice as long

if you have fiber, mtu should be going to 6400, but the shit is still stuck on copper age

This is a good video on how netcodes work

you`d be better off with

The story kinda hinges on having to stand still in CS (making it a kiddy game imho) the influence of the netcode in QL is much more severe because all players are (pretty much) constantly on the move.
Which basically means that only the stuff around the 7 min mark is relevant.
Which basically tells you that the aggressive player (with instant hit weapons) is always at an advantage.
Also QL doesn't often have situations where the players die because of 1 shot. The fights are much more extended and there are decisions possible inside of fights, which get skewed depending on ping. Making adjusting your play style to your ping a much larger issue.

Your arguments with the netcode are valid but i think you exaggerate on the wrong issues that has infected online play. I too hate it how you can get hit around corners with the lg but i am able to escape 90% of my lg fights against high pingers without being lged to death simply because i've gotten used to it. What is really ruining this game is the fact that people's aim are getting better and hitscan is now pretty much useful in almost every situation sometimes even when you are close range against someone with rockets.

You cannot fix this problem by simply lowering the lg dmg because it is an aim problem. In the vq3 era wep selection was very dependent on which wep was the easiest to shoot with for the given situation. In ql wep selection is very dependent on which wep will deal the most dmg for the given situation and the fact is this change of style makes hitscan weps the number one choice.

nope, the issue is that you cannot dodge lg streams anymore.
People's aim has not significantly improved, they just altered it to the inability of the other player to react.

and how have they altered it?

The way i see it, the netcode as well as players improving their aim through cfg/setup etc have tightened the gap that was the lg/rocket dilemma and made the game more about hitscan and less about projectiles.

Back in the day most pros would use high sens because low sens was all about lg and to be pro rockets were more important. But as time went on people's aim improved and players have started using lower and lower sens while maintaining their good rocket skills tightening the gap between the two weps.

As for the netcode improvements, players could hit their shots more consistently now so players became even more free to use dangerous weapons in difficult angles/situations instead of just using the one that was the easiest/safest to shoot with.

With these changes the game became less about playing a slow strategical game and more about out aiming and combat skills. Where players in the past would hope to hit a lucky spam rocket in a difficult angle instead of dealing measley lg dmg, they started using lg in those not so difficult anymore angles as it became easier to hit (consistency) and their aim (accuracy) improved.

The reason you cannot fix this problem completely by lowering the lg dmg is because that will only make ppl use the lg less and take less accuracy involved fights but won't stop people taking the consistency based fights such as the use of lg in those not so difficult anymore angles i talked about before. So unless they make aiming hard again quake will continue to become more and more aim and combat heavy. Projectile weps are also affected but to a lower extent.

So when i hear pros like tox say that all players these days aim the same, i think the opposite is true. I feel like more players have learned to aim at his level but also their is still a large number of ppl who aim crap like 1999 and their is wider variation in aim causing wep balance changes to be even more chaotic.

And thanks to these changes if a player loses to a strong lg player, he will more often than not lose to a player with 50% wep usage instead of the common 25% wep usage in the past. And with more peoples aim becoming better and better the gap in aiming skills is becoming larger and larger centering the game more around aim while the strategical side of the game has pretty much being mastered.

I wouldn't be suprised if 50 years from now when people have some super precise touch pad or some shit it will be normal for people to run in and attack in close range fights with rg instead of rl since it is more consistent if you hit everything as its hitscan so no guess work is involved.

Basically my hypothesis is that the reason lg aggressiveness works so well online isn't because of behind the wall type of shots (or though they are a small factor), The reason i think they work so well online is because the gap in lg aiming skills is becoming larger and although this is not as noticable between top players vsing each other usually because they have similar aim it is definitely noticeable between a top and lower tier player. A rape game in q3 was a rocketfest, now it's a lgfest.

Basically lg aggressiveness is mainly about abusing that the opponent has a much weaker lg, abusing that stronger lg means you can get more benefits out of every rl dodge.and abusing that the rl is delayed online.

rofl @ the strategical side of the game being mastered....

But anyway, the reason why hitting LG is easy is because it's become impossible to dodge it and move away thus nobody tries to anymore.

The difference with CPMA being that CPMA extrapolated the other guy for a large part as well. Meaning that you on your screen are aiming at a predicted model of where the opponent will, most likely, be.
Nowadays in QL you are aiming at the spot where the opponent was at.
So on the screen of the 'the guy that gets hit by LG' the LG doesn't actually have to hit.... it often isn't even remotely close. And thus, it cannot be dodged (as you have no way to tell if he actually hits, how he hits, where your own model actually is on the server (as you get knockback which influences how you could dodge)).
So basically the aggressor is always in the advantage because for him the screen is showing accurate information while for the defender it's hopelessly flawed.

Probably just a little too technical, but....... http://fabiensanglard.net/quake3/network.php

Interesting observation which may be irrelevant, but quickly skim reading that again it has an observation that Q3 picks packet sizes to target MTU 1500. IIRC, Win7 and later frequently have an MTU of 1300 for wifi networks. Wonder if that isn't helping?

Edit: and if you follow the link at the bottom to Brian Hook's write up you find the comment I was actually looking for. There was some weirdness with routers doing NAT and client port numbers getting reassigned is session that caused issues back in the day. Maybe the modern routers do something similar (but different) and that is why people are seeing strange effects with changing net_port?

put your mtu on 510 and your internet will be light speed