Predictor Code - Rewrite

[Cornered Rats]

For sometime now we've been having what the players have referred to as a, "lag issue." This has been the key focus of our game development effort as of the last few weeks as we've tried to nail this down. The good news is we're getting closer and I'll be going into some more detail about what this is, the effort involved and help set your expectations for the intended outcome.

View the full article on battlegroundeurope.com

[GrAnit]

I am glad you are working on the predictor code; it is extremely important to enjoying the game.

 

Can you provide a more detailed technical explanation on what is the issue with the code, how it is being addressed, and what the likely outcome is that we can expect?

[XOOM]

1 hour ago, Granit said:

I am glad you are working on the predictor code; it is extremely important to enjoying the game.

 

Can you provide a more detailed technical explanation on what is the issue with the code, how it is being addressed, and what the likely outcome is that we can expect?

I added some more information to the article from Victarus, funny you asked I was waiting on him to finish writing it up. To his credit I only asked today so, I guess since he's kinda busy I should give him some slack right? NOHP! 

[bmw]

Just read  Victarus  writeup.....good explanation for us DOS for dummies guys.

Hope it all gets worked out soon 

[GrAnit]

Thanks for the added detail.

 

Some time ago (many years, lol) I thought the previous Rat team did something similar but it caused a host of unintended  issues and was eventually removed.   For example, when the old predictor code was active you could change direction quickly and frequently while moving causing the predictor code to predict wrong.   To other players it looked like the player that was zig-zagging  instantaneously displaced from one point to another  as the 'real'position was updated from the incorrect predicted position.  This made it almost impossible to shoot someone because they were not where the predictor code said they would be.

 

 Have you looked into the problems in the previous incarnation of the code to avoid this type of issue?

[VICTARUS]

Unfortunately, a part of that is just the nature of the beast: If someone has a really poor connection, it's basically impossible to try to turn the information they send/receive into something that both looks good and is in any way accurate. The old code did a kind of decent-ish job at this (sometimes (sort of (arguably (I guess? (ok, not really))))), but it did it by guaranteeing you were lagging behind the latest info: If we're always playing catch-up then a little bit of lag just translates into moving the player a little bit later than usual. (Admittedly, I doubt this was how it was *designed*, but it's how it turned out.)

As for the quality of the new predictor, there's a little bit of rubber-banding if you start zig-zagging all over the place, but from what I've seen it's only really bad if you or the other player has a really poor connection. I plan to work more on the smoothness and physics prediction, but so far our everyone who's tried the client has gone "Yes, get this in the game immediately" — at first just because it fixes the horrible lag, but then because they've bounced back-and-forth between the new and old prediction and the difference pretty much feels like cheating.  (Really, the difference isn't obvious just running around, but the chance to actually take a shot at that guy when he comes around the corner is a *big* game-changer.)

[Sudden]

Despite all the improvements by the new team, this one has to take the top spot. Thanks for all the hard work.

[bmw]

Thanks for the update VIC....go to know those who are trying out see a vast improvement.

[Fohdron]

We have been spending hours testing this and so far we think we have something that is releasable with the knowledge we will continue to improve upon it since the code is maintainable now that it was re-written. 

From our testing it appears that we have improved majorly on the CP lag that players were being killed before the EI came around the wall/door,

We will be testing this further this weekend with a larger scale test with select individuals know to have been reported with the problem.

 

Pending nothing comes up from the test we should be good to release 1.35.5 early next week.

[major0noob]

why don't you just make dead guys bullets do nothing? it'll suck for high ping players but we all understand.

[Fohdron]

7 minutes ago, major0noob said:

why don't you just make dead guys bullets do nothing? it'll suck for high ping players but we all understand.

Not really sure that will help when that has nothing to do with the problem...

[major0noob]

why don't you just make dead guys bullets do nothing? it'll suck for high ping players but we all understand.

[B2K]

18 minutes ago, major0noob said:

why don't you just make dead guys bullets do nothing? it'll suck for high ping players but we all understand.

The problem is by the time the guy knows he's dead, he's already shot you...... and you're dead.... so your bullets would do nothing... 

[bmw]

As much as this code is needed................Im more interested in when town based supply will be ready.

[VICTARUS]

25 minutes ago, major0noob said:

why don't you just make dead guys bullets do nothing? it'll suck for high ping players but we all understand.

The issue being addressed here is position lag. The writeup I did in the first post is a good description of the overall issue, but due to a bug in the interpolation code it can get even worse sometimes — it's possible for a players' visible position to be behind by more than a full second or two — so in really extreme circumstances you can be upstairs in a depot capture building and be killed by someone who your client renders as just starting to run in through the front door.

To be honest, this new code might make what you're describing a non-issue. The actual problem is usually "A guy ran around the corner I was watching and we both died". The real problem there is that a moving player actually has a little bit of extra time to see their opponents with the current code (again, see the writeup in the first post for a simplified description of why). With the new code, that isn't the case anymore: When you see yourself round that corner, because of the prediction everyone else will see you round the corner at pretty much the same time. This should mean less getting instantly killed by players who are rounding corners.

7 minutes ago, bmw said:

As much as this code is needed................Im more interested in when town based supply will be ready.

Barring some last-minute bug, that's actually what I'm working on right now. No solid estimate as to how long is left — it's been a big change in terms of code, and there's a lot of edges to smooth out — but the bulk of the core stuff has been finished. 

[bmw]

45 minutes ago, VICTARUS said:

Barring some last-minute bug, that's actually what I'm working on right now. No solid estimate as to how long is left — it's been a big change in terms of code, and there's a lot of edges to smooth out — but the bulk of the core stuff has been finished. 

Great news!! Thanks for the update

[droptow]

Many thanks on the updates Victarus!

[vasduten1]

Glad for the updates!

 

This team is such a relief...

[captcrayon]

Anything that make the game better  you rats always do a great job! Keep up the great work!

[sideout]

Thanks for the hard work!