Our ReVolt Pub

sebr @ Feb 10 2014, 01:38 PM wrote: yesterday i get 3 stras in half of 1 lap on a 3 lap online race
there where more than 5 stars on this race, and i'm not including the track star ...
with only 6 racers ...

It's a little too much i think

I remember that i was last when i get them ...

About offline race, i never noticed this BUT i'm most of the time 1st 2nd or 3rd at the secound half of the 1st lap when i get my 1st pickup ...

(I always play offline simulation, but i'll try to go back to arcade and check it)

After some test:
Clockwork carnage with 30 toyecas Arcade 4 laps picks on
Each time I wait that ai cars made 1lap before i start => this way i stay last all race
The ONLY 1 time ai cars get star is when they accidentally drive on track's star ...
At all the races i only get 1 star/ race (and some race without star)
Weapon impacts seems a bit less powerful than in multi-player
car/car and car/wall collisions are as crazy in online or offline mode

Hi Huki, thanks for your good will in helping us again. And double thanks for explaining how the randomizer works.

I have a question now. I remember that some time in 2011/2012 the amount of pickups per players was increased a little. Is this feature still enabled in nowaday's patch?

Could you provide us a test build with this feature off?

Because I think another reason of the bigger amount of stars is because there are more pickups in race than in the old times, and this makes the game a bit hellish.

It was in 2012, look what i said =>>http://z3.invisionfree.com/Our_ReVolt_P ... p=22001369

The collision bug is always present on the last alpha in multiplayer mode. All impact (rockets, cars's collisions, walls's collisions...) are multiplied by 10. Why ? And there are too many stars during a picks races ! I did a clean installation to be sure that the problem did not come from me. Try to play with revolt 1.1, it's different. This patch RUINS the gameplay, ruins the fun !

And this about the number of picks per player =>> http://z3.invisionfree.com/Our_ReVolt_P ... p=22003749

Thanks for the link Dolo. According to that, the nowaday's patch has already reverted back the old amount of pickups per players. Okay, 1 possible cause of the problem to discard.

@SebR can you send me your clockwork folder with the toyeca ready to use in clockwork carnage? I want to make the same test that you did.

Edit:
@Cosmo, Nero, Stingox, Kosztya, zappa, sjampo, hilaire9, Last_Cuban, benvoliosaythat, sonkite, Gilou68, RobinEDTaylor, Why_try_harder, Monty, DrPepper, etc.. all the people that said that the weapons are overpowered in 1.2 Alpha, can you guys test offline too?

I was looking at some recent videos from SebR's channel. I assume they were using 13.0815 or 0820 (shouldn't make a difference as far as online is concerned). From what I can see, the weapon and car-car impacts look just fine. This video for example:
[youtube][/youtube]

You can see that the remote cars are very responsive.. even immediately after the countdown, they start moving without any delay - which wouldn't be the case in 1207. I could be wrong but having seen several others of the recent videos, everything seems just fine, really...

I wonder, are these recent complaints specifically about the 14.0208 build, has anything changed in this new release.. or just old problems that existed in 0815 / 0820 also?

And just a theory I have: remember that in "the old times" games frequently used to crash, nor was there any late join feature. Maybe because of the increased traffic and gameplay duration, people start noticing these "changes" which in reality is quite normal for re-volt.. I could be wrong, but it's still a theory I'm starting to consider..

Phantom @ Feb 13 2014, 10:25 AM wrote: @SebR can you send me your clockwork folder with the toyeca ready to use in clockwork carnage? I want to make the same test that you did.

toyeca ready to clockwork carnage

@huki : yes last video are with 13.0820 patch (i was only spectating)
and the weapon pb is the same on 13.0820 and 14.0802

Thank you my dear. ^^

Okay offline mode, here I go. Will edit this after test.

Huki, this is a video about a little issue between mmud's and my car. I didn't push his car, but it moved when I went close to him (and he finished, couldn't push accelerate button). Why did it happen? (the distance is between 0.5-2m)

[youtube][/youtube]

Here is an another video about that mmud stolen bomb from me when he was behind me with 1m. The other strange thing is that Vaidux couldn't stolen bomb when I hit him. (video from 0:10 to 0:16)

[youtube][/youtube]

And 1 more thing (about the collision again) here is the 3rd video about the same issue. Guys can overtake me in the tight way when I replace myself and they didn't fall down because of hitting me and their speed didn't decreased. (video from 0:30 to 0:40 and 1 more interesting happening from 1:08 to 1:12, Giacomo crashed in the first leg of the dino, but his crash happened near the second leg what is a little bit away from the first leg)

[youtube][/youtube]

Huki @ Feb 13 2014, 07:43 PM wrote:I was looking at some recent videos from SebR's channel. I assume they were using 13.0815 or 0820 (shouldn't make a difference as far as online is concerned). From what I can see, the weapon and car-car impacts look just fine. This video for example:
[youtube][/youtube]

You can see that the remote cars are very responsive.. even immediately after the countdown, they start moving without any delay - which wouldn't be the case in 1207. I could be wrong but having seen several others of the recent videos, everything seems just fine, really...

I wonder, are these recent complaints specifically about the 14.0208 build, has anything changed in this new release.. or just old problems that existed in 0815 / 0820 also?

And just a theory I have: remember that in "the old times" games frequently used to crash, nor was there any late join feature. Maybe because of the increased traffic and gameplay duration, people start noticing these "changes" which in reality is quite normal for re-volt.. I could be wrong, but it's still a theory I'm starting to consider..

You can't take a video from spectating mode to prove that alpha 12-13-14 are good...
I already showed you others videos from me with completly crazy gameplay with many crazy collisions, impacts weapons in racer mode...

Since few days, we tested some version from alpha 11.0426 to alpha 11.0825... All players with who i played are agree with me... The gameplay is really different about driving, weapons impact and cars impact...
And there was the late joining on it...

Phantom @ Feb 13 2014, 08:38 AM wrote:I think another reason of the bigger amount of stars is because there are more pickups in race than in the old times, and this makes the game a bit hellish.

Phantom @ Feb 13 2014, 02:55 PM wrote:Thanks for the link Dolo. According to that, the nowaday's patch has already reverted back the old amount of pickups per players. Okay, 1 possible cause of the problem to discard.

Yes, the pickup count was indeed restored.. but that reminds me of some other bugfix / improvement in 13.0815.
In races with heavy traffic, some weapon fires may be lost over network (i.e., Player1 fires a rocket and he sees it hit Player2, but Player2 doesn't get the message over network so there is no effect for him). Likewise some player could have fired a star weapon without affecting anyone..
Since 13.0815 the weapon fires are sent using guaranteed messaging so that they don't get lost during traffic. It could be a reason why people notice more firework impact / stars in the game. Maybe I should put back the un-guaranteed system for weapons and see if that helps.

Dolo wrote:You can't take a video from spectating mode to prove that alpha 12-13-14 are good...
I already showed you others videos from me with completly crazy gameplay with many crazy collisions, impacts weapons in racer mode...

Since few days, we tested some version from alpha 11.0426 to alpha 11.0825... All players with who i played are agree with me... The gameplay is really different about driving, weapons impact and cars impact...
And there was the late joining on it...

When we are sure (and with verifiable proof) that not one piece of code related to physics and collision was touched, and the network lag management is actually better than it ever was in any previous build, there has to be some small fix or improvement (as the one I mentioned above) that is causing the inconveniences.. definitely no "complete crazy gameplay" as you insist.
The weapon and car impacts are all random, and arcade mode (which I too prefer in online races, before you accuse me of bias) collisions are imperfect and "pushy".. a lot of things can "change" in each race.. you can't play a different build for a few days and find a definitive change in the impacts...

@Kipy: In fact all 3 of your points can be explained by lag.. not the bad kind of "lag", just normal internet lag. The price to pay for playing on the internet..

Huki @ Feb 13 2014, 10:20 PM wrote:When we are sure (and with verifiable proof) that not one piece of code related to physics and collision was touched, and the network lag management is actually better than it ever was in any previous build, there has to be some small fix or improvement (as the one I mentioned above) that is causing the inconveniences.. definitely no "complete crazy gameplay" as you insist.
The weapon and car impacts are all random, and arcade mode (which I too prefer in online races, before you accuse me of bias) collisions are imperfect and "pushy".. a lot of things can "change" in each race.. you can't play a different build for a few days and find a definitive change in the impacts...

You think that the lag management is actually better ???
I never see the lag when I play at the latest alpha ! You think it's better ??
I prefer see lag without no ****ing overpowered pushing instead to see no lag but with ****ing overpowered pushing...
On old alpha, we could see little lag... With this experience, i think it's better to see lag like we saw in revolt 1.1 or on old alpha 11 instead to see "hidden" lag... I think a player with a bad ping, so a bad connection is not a lagger on last alphas, but just a big bulldozer... It's like if the position of the car is not enough refreshed... or too much...

Sorry, just a very quick post because it's so late...

I've had a fresh look at the network code (which is not the part I was most aware of), and after having some difficulty to simulate a bad network between 2 local computers (with dummynet) I finally get an usable setup.

So I just fixed an issue with the sync code in v1.2 (I have not checked when it was introduced), as well as some other minor potential bugs. In my very last test I had the feeling the improvement was noticeable, but since I had to tinker a lot with dummynet I'm not 100% sure whether the previous issues I noticed were due to Re-Volt or a dummynet malfunction.

So I'm sharing this build F for you to test (this is just the exe that obviously has to be installed on top of a full v1.2 install). It doesn't feature the changelog of the last official alpha (it would be more close to the previous build I already posted in this thread.)

Re-Volt_14.0222_F.7z

All players in the race should use this version, and as usual the Legacy Compatibility in the Network settings should be turned off.

For info, I was able to play with a ping greater than 400ms and a bandwidth lower than 50Kbit/s. The remote car collision were obviously a bit weird, but this is perfectly normal with such a ping...

SebR and I tested your patch yesterday. He hosted the room, and basically... we got to test nothing.

Legacy Compatibility was off.

Every single time he started a race, it was a bad start, I couldn't move because I was stuck at either "Waiting for host" or "Waiting for: SEBR"... as I see him driving about, teleporting all over the place... although I'm not sure if that teleporting is his connection, or the patch.

Usually bad starts happen once (maybe a few times more in the 1207-era) and that's it. We tried at least 10 times with the patch but we were still given the same result.

Ok, thanks for the feedback guys.

That's weird, since I don't think that part was modified since my previous Build (D, E...).
I did notice this "Waiting for ..." message during my test but it used to last at most 1 or 2 seconds.

Do you know if he was able to send you F12 messages while you were stuck?

And for info in my test I limited speed to 28Kbit/s and added a delay of 100ms on both computer (ie. ping is 200ms). There was no simulated packets loss though. I have no idea what typical values we can expect for speed and ping between two players on Internet...

Anyway, I 've already improved this Build in several ways to prevent network messages congestion. It's not ready right now but I should update it very soon. We'll see if it fixes this issue.

Weird, I tested it yesterday morning and I got it working quite well, I need more time to detect if there are noticeable gameplay changes though, I only raced a few and it was a 1vs1 test. Dolo tested with SebR too and they didn't mention major problems, but it was another 1vs1 test. The real deal can only be felt in a more players room. Please give us more time before releasing any other improved build, if you release another one now people won't test this enough.

Oh and the limits you're using are very correct to simulate our networks. I normally have 100ms ping with people from my same country and approx 400ms-500ms with people from Europe so your numbers are okay and normal.

jigebren @ Feb 24 2014, 03:41 AM wrote: Do you know if he was able to send you F12 messages while you were stuck?

Yep, he was able to. So was I.

Just finished testing with 5 players this time on Build F, and it was a longer test (20 races, 3 laps).

At the moment I can say we didn't get any bad starts in any of the races, so the error mentioned by Nero doesn't occur here. But what I did noticed that was mentioned by nero is the teleporting.

In Alpha 0820 the position of the other players is more stable (doesn't teleport so much). Like Dolo would say, "the lag is there but you don't notice it in 0820".

In Build F the position of the other players isn't so stable, I saw the other players teleporting suddenly, like Nero mentioned. Not so much but a lot more than in 0820. And I asked everyone in the room to close all applications using internet before we started.

After this test on the Build F, inmediately we switched back to 0820 and continued racing. The teleporting is almost unnoticeable in 0820.

@Jig: What exactly should I pay attention to in my following tests? lag management? collissions? teleporting? or is there something else I should be looking at?

Thanks for the test. I recently added packets loss simulation during my own test (roughly one drop every ten packets), and I think it introduced something very similar to the teleporting you're mentioning. That is to say, the car rotation looks erratic mostly, and this introduces other glitches like the car looks like it's flying after a small jump, or the car looks like it hits a rocks even though the road is almost flat.

I believe these are outcomes of the original interpolation code. The fact is that we ironed out several bugs that used to disable this interpolation code. As a consequence, some interpolation glitches may have became more noticeable.

I could upload a build F with interpolation disabled so that you can see the difference (but it means temporarily reverting my new changes and a new compilation takes some times). Interpolation looks good in most case since it prevents "jaggedness", but since it has to forecast unknown upcoming data, it may also introduce glitches as well.

What exactly should I pay attention to in my following tests? lag management? collissions? teleporting? or is there something else I should be looking at?

Until next release, nothing specific I think. The build F should have fixed a bug that prevented players synchronisation before a race (except in case the ping was really good), which means: 1) the GO signal should be more accurately sync for all players, and 2) the interpolation code would no longer be disabled.
I'm afraid point 1) is not easy to check since you're obviously not in the same room than remote players, and since point 2) may make interpolation glitches more visible, it should not look like an improvement for a session over Internet (though it would on LAN session).
But if you already have more details about these two points let me know.

Edit:
NB: For the sake of accuracy, I said "interpolation" but you should read "extrapolation"...

nero @ Feb 23 2014, 11:47 PM wrote: SebR and I tested your patch yesterday. He hosted the room, and basically... we got to test nothing.

Legacy Compatibility was off.

Every single time he started a race, it was a bad start, I couldn't move because I was stuck at either "Waiting for host" or "Waiting for: SEBR"... as I see him driving about, teleporting all over the place... although I'm not sure if that teleporting is his connection, or the patch.

Usually bad starts happen once (maybe a few times more in the 1207-era) and that's it. We tried at least 10 times with the patch but we were still given the same result.

When we do the test, i was downloading something at the same time ...

but if i do the same with bild 11.0501, 11.0525, 11.0825 or beta 11.0208 i'm only lagging without bad start ...

Good job!
We said well the truth since all this time...

Tried again on GameRanger with Dolo, Phantom and Polenta today and this time it worked flawlessly. The last time I tried it out (with SebR), it was probably because of his connection (he was downloading) that it didn't work properly.

I'm not quite sure what these changes entail as I didn't really see the difference... mind you we didn't really get to test the pickups, so it would be better to test this with more players.

Let me add that I did not see other players teleporting either.


Regardless, you're onto something here, on the right track.

Hi Jig, we did more tests today. We have several good things and 1 bad thing.

Good things:
1 - We liked the new Tracks selector (Reverse and Mirrored don't work though).
2 - The impact of the weapons are normal, it is impressive the difference for me. Rockets aren't so powerful, shockwaves don't make you fly all over the track and when landing after being attacked by firework or shockwave the car lands normally which wasn't the case in previous alphas where weapons were dynamite. Electroshocks work quite normally, bomb pass work, even driving over oil is more controlable now, just like it was before the collission dilemma started.
3 - The following is a bit awkward: according to your changelog you didn't change anything in the collission specifically, only in the network management, or at least that is what we understood. But.. we have an agreement here with Dolo that touching walls feels better now and I don't get why.. in previous alphas wallbouncing was always buggy while in this build touching the walls feels better and wallbouncing is actually possible now.


1 Bad Thing:
1 - Teleporting - it makes it almost unplayable. Players (specially the host) fly all over the track constantly with rude position movements, radar going crazy showing that the drivers are at 200mts and suddenly at 23mts and then suddenly again to 200mts and back to 20mts again. It is impossible to detect the real location of other drivers. The host was mmud who has the fastest connection between all the online players as you know and we were only 5 (mmud, Nero, Polenta, Dolo and me). And for me Dolo and Polenta weren't teleporting so much but the host (mmud) was teleporting like i never seen before. I hope mmud can write his report from his perspective here. And this situation happened since the first race until the last race. Dolo says this is normal for re-volt and it was like this in the original re-volt but I don't think so. In 0820 this teleporting doesn't happen in such a rude way and we were playing with 0820 seconds before we proceeded to do the Build F Test. 0820 has more powerful weapons but the teleporting is less frequent which makes the other drivers positons more stable compared to Build F. Anyway that is my report. I hope mmud can write his own.

Phantom @ Mar 2 2014, 01:32 AM wrote:Hi Jig, we did more tests today. We have several good things and 1 bad thing.

Good things:
1 - We liked the new Tracks selector (Reverse and Mirrored don't work though).
2 - The impact of the weapons are normal, it is impressive the difference for me. Rockets aren't so powerful, shockwaves don't make you fly all over the track and when landing after being attacked by firework or shockwave the car lands normally which wasn't the case in previous alphas where weapons were dynamite. Electroshocks work quite normally, bomb pass work, even driving over oil is more controlable now, just like it was before the collission dilemma started.
3 - The following is a bit awkward: according to your changelog you didn't change anything in the collission specifically, only in the network management, or at least that is what we understood. But.. we have an agreement here with Dolo that touching walls feels better now and I don't get why.. in previous alphas wallbouncing was always buggy while in this build touching the walls feels better and wallbouncing is actually possible now.


1 Bad Thing:
1 - Teleporting - it makes it almost unplayable. Players (specially the host) fly all over the track constantly with rude position movements, radar going crazy showing that the drivers are at 200mts and suddenly at 23mts and then suddenly again to 200mts and back to 20mts again. It is impossible to detect the real location of other drivers. The host was mmud who has the fastest connection between all the online players as you know and we were only 5 (mmud, Nero, Polenta, Dolo and me). And for me Dolo and Polenta weren't teleporting so much but the host (mmud) was teleporting like i never seen before. I hope mmud can write his report from his perspective here. And this situation happened since the first race until the last race. Dolo says this is normal for re-volt and it was like this in the original re-volt but I don't think so. In 0820 this teleporting doesn't happen in such a rude way and we were playing with 0820 seconds before we proceeded to do the Build F Test. 0820 has more powerful weapons but the teleporting is less frequent which makes the other drivers positons more stable compared to Build F. Anyway that is my report. I hope mmud can write his own.

820 is not a reference considering it was buggy as many of them in the past ... But maybe teleporting was less important in the past with version 1.1...
And your connection was very crappy, phantom...

One last thing for me, with this fix, the sensitivity and range direction with my wheel is different in the game, and certainly more in line with what should be re-volt... (not yet like beta or 11.0525)

Yes, my connection wasn't the best yesterday. But still the way that 0820 handles the high ping is a little more stable than in Build F under the same conditions. If it is possible to convine the good weapons of Build F + the stable car position of 0820 it would be the perfect patch.

Phantom @ Mar 2 2014, 12:42 PM wrote:If it is possible to convine the good weapons of Build F + the stable car position of 0820 it would be the perfect patch.

And then add custom car skins and an online championship mode... although this is slightly offtopic.

Phantom @ Mar 2 2014, 01:42 PM wrote: Yes, my connection wasn't the best yesterday. But still the way that 0820 handles the high ping is a little more stable than in Build F under the same conditions. If it is possible to convine the good weapons of Build F + the stable car position of 0820 it will be the perfect patch.

To believe that it is either one or the other unfortunately...

Seems like we're going in the right direction... That sounds good.

Phantom @ Mar 2 2014, 01:32 AM wrote:1 - Teleporting - it makes it almost unplayable. Players (specially the host) fly all over the track constantly with rude position movements, radar going crazy showing that the drivers are at 200mts and suddenly at 23mts and then suddenly again to 200mts and back to 20mts again. [...] The host was mmud who has the fastest connection between all the online players as you know and we were only 5 (mmud, Nero, Polenta, Dolo and me). And for me Dolo and Polenta weren't teleporting so much but the host (mmud) was teleporting like i never seen before. I hope mmud can write his report from his perspective here.

This is a relevant info. The host being the most affected, it means that teleporting is actually an outcome of network congestion (since the host has more data to send than other players). BTW I don't think mmud itself has actually suffered much from teleporting (though his car is the most likely to be affected, it's only from the remote players point of view).
It would be interesting to know it the track you were playing with has a lot objects (like eg. traffic cone) since it should make this host issue more noticeable.

The radar issue sounds weird though.

I had spend quite some time recently trying to use another approach to the lag compensation issue (mostly for personal curiosity and also as a way to understand better the involved mechanics). Though I got some interesting results, I think it was not enough to go further in this direction, I don't want to lose more time. So I'll be going back to build F. If I can improve some points according to the previous report I'll publish a new build to compare.

jigebren @ Mar 2 2014, 02:29 PM wrote: It would be interesting to know it the track you were playing with has a lot objects (like eg. traffic cone) since it should make this host issue more noticeable.

Can you prive me with a list of of the tracks with lot of objects? I'm not sure what you mean by objects actually, for me all tracks have lot of objects and the teleports occur in all of them so i don't see why this is relevant.

Phantom @ Mar 3 2014, 09:59 PM wrote:Can you prive me with a list of of the tracks with lot of objects?

I mean tracks that feature "movable" objects, that is to say any object in this list:
"Football, Beachball, Trolley, Weeble, Basketball, Tumbleweed, Bucket, Cone, Lilo, ABC"

The following ones are disabled in multiplayer for performance reasons so they don't count here:
"Bottle, Can, Packet, ABC".
Animated objects like Copter, Train, Planets, etc. are moving objects, but not moveable, so they don't count here either.

The host sends moveable objects data the same way it does for his car, so this takes part of his upload bandwidth. That's why the host car is more likely to lag on track with a lot of moveable objects (compared to normal players or to a track without those objects).

BTW I already improved this point so this effect should be way less noticeable in next build...

I see. Thanks for the explanation. I didn't know that. I'll keep this in mind for future tests.

Yesterday we made more tests on Build F. This time I tested with 2 drivers from my same country (Chelco & Polenta), both with a 100ms ping to me. I was hosting, and look:



This is the kind of radar issue I meant. On Build F it happens constantly, mostly it happend on the first race (Toys in the Hood 1), Museum 1 and Ghost Towns where the radar went crazy like on 6-7 occasions during the same race. And this radar issue was noticed by Chelco & Polenta from their perspective too (meaning not only 1 player sees the radar showing crazy values when the car is right next to you).

First we started with Build F, apart from the radar going crazy the teleports were horrendous, and we were just 3 players who live very close to each other, and best ping. We did 13 races, 4 laps each. Results is that the teleporting affects the race since the first race (hood1) up to the last (ship2). It was pretty unraceable since the movement of the position of the car is rude, unpredictable. Also I should mention that for me (the host), both other drivers teleported a lot. And they saw me teleporting a lot too, but they didn't see each other teleporting so much (although the teleports were still horrible not as much as with the host).

After the last race (ship2) we preoceeded to repeat the exact same session under 0820, and yeah, everybody noticed the difference: car's position a lot more stable, 0% lag between us. However, the weapon impacts are a lot better on Build F and the steering is also better on Build F.

Greetings.

Edit: monty reports the same thing on his tests:

<monty> i see a lot of distance-meter (radar) with 14.0222
<monty> radar lagg in a lot of races
<monty> i had full line
<monty> no programs running
<monty> sebr was in front of me
<monty> but game said i was 1st
<monty> and so on..

Edit 2: Aaaaand.. today it happened. We convinced HIL to install the Build F, which means everybody else upgraded for first time so a test room over 14.0222 with 12/12 players was possible.

First comments:

<Kosztya-> Unplayable patch !
<Phantom> why unplayable kosztya?
<Kosztya-> - Everyone flying, many lagging player
<monty> lagg
<monty> radar lagg
<Nero> you're used to 0820 kosz

i must add that on Monty's test i didn't have any lag or teleporting
i was behind monty before the end and finish 1st at the last corner ...
Is that mean that i receive all data correctly but didn't send them in the same way ?

well, like i said, there were more than a few races in which i saw radar lagging

No lagg, no radar issue at all, works fine for me. Nice job I can say, more stable driving and collisions.

Ok, I'll try to check the radar issue. And thanks for the screenshot which clarified it since I have never noticed it so far.

Glad to see that even Hil joined the testing, though I'm afraid this is not THE build to spread to everyone for now. Since it's WIP, I consider it more alpha than the alpha.

I'll try to upload the next build soon as I'm be curious to see how it behaves, but I'll will likely not be suited for room with a lot of players.

I also lack one information that appeared to be crucial in my last test: when it's very low (for example 10Kbps for a 2 players session), the bandwidth does matter a lot. Can you post here (or PM me) a few relevant values of the upload bandwidth of the players in the room?
You can get it from sites like this one. And I actually mean the upload speed, which is usually way lower than the download speed. Unit should be Mbps or Kbps. Thanks.

EDIT:
I was definitively not able to reproduce the radar bug in an artificial degraded LAN session with 2 players, so I'll need more info. I believe the radar jump value is always the race length (for example 668m for museum), can you confirm it?
And when a bad value appears, how long will it last? a few seconds only, or rather something like 30 seconds, or until next time the starting line is crossed.

jigebren @ Mar 5 2014, 01:57 AM wrote: I believe the radar jump value is always the race length (for example 668m for museum), can you confirm it?

Yes it is so, I think you either see that you have been lapped (+the right amount of meters) by a certain player, or you have lapped a certain player, and I think it's always only one player, you don't see that you lapped everyone, or that you've been lapped by everyone.

Of course you will know much better what the consequence of this is, but just an idea... can it be a problem with lap counting? Like... the game 'forgets' which lap are other people in.

This bug has occured now and then since I am here, though, not a new thing.

In v1.1, this bug radar could appear but rarely. It happened with the players who had a very bad connection... It was not also frequent like today, but it could happen...

jigebren @ Mar 4 2014, 09:57 PM wrote: I believe the radar jump value is always the race length (for example 668m for museum), can you confirm it?

I still don't have this info, I will keep it in mind in my next tests. Thanks a lot for pointing this out. About how much it lasts , it never lasts the entire race. It randomly starts in some moment of the race, stays there for a lap or maybe two or even half a lap, and it fixes itself finally making the value go back to a logical number.

One clarification though, like Balint said I believe the radar bug has always been there since re-volt 1.0, it is not something 'new'. It randomly happened in the past. It isn't something actually added by this new build F. It happens in previous builds too, for example I tried today with 0820 again and the radar goes crazy in few ocassions aswell for most players in the room. Maybe this was more noticeable with the Build F because the Build F has a different management of the lag I believe.

What is indeed new in comparison to 0820 is the Teleporting. The way that the patch handles the car positions under high ping is different and this is something actually more disturbing and more important than the radar bug which is no more than numbers showing unreal values, while the teleporting is what makes multi-player games unraceable.

Physics is unrealistic online and offline
and many times i had experienced "fake" 1st
weird **** with radar

Here's new build G:

Re-Volt_14.0306_G.7z

(Use 7-zip if you can't open the archive. This is the exe only, so use it on top of a full v1.2 install).

And here's the changelog:

• Trying a different approach for Packets Per Second handling to avoid overwhelming low-bandwidth, still remaining responsive. Currently not perfectly suited for more than 2 players if at least one player in the room has a crappy connection.
• When ping gets high (roughly more than 400ms), don't extrapolate the remote car position above 200ms so as to avoid too noticeable prediction error.
• Minimize the bandwidth used by the host for moveable objects in the tracks (basketball, traffic cone) when they stand still.
• Avoid updating timers synchronization with host when ping is bad.
• Try to prevent radar bug when cars are around the starting/finish line.

Radar bug
The radar code is quite ugly and complex, but I think I have already identified a potential issue. I tried to avoid it in this build (this would not definitively fix the issue but should be enough). Also, since I believe this may also be a consequence of Teleporting, if there's less teleporting in this build it could lessen the radar issue occurrence as well.

PPS
This build is also an attempt to manage PPS (packets per second) more efficiently. Keep in mind though that in its current state, the host PPS will align with the worst player PPS in the room, so if there's one laggy player, the host will likely looks laggy as well (and probably other players too). So if there's actually an improvement, it should be mostly noticeable in head to head session for now.

Extrapolation / Ping
Some info about the ping (first and last points are roughly valid for all RV releases but second point is introduced in Build G):

• under 400ms, extrapolation is used. It means remote cars position is tried to be predicted with no delay.
• above 400ms, remote cars begin to be displayed with a delay (so as to cap extrapolation inside reasonable limit), but interpolation is still used.
• above 1.2 second, interpolation & extrapolation are disabled. Remote car info are copied as they're received. Anyway a ping higher than 1 second is not really supported, as it would not make much sense anyhow.

Your words are exactly what I saw.

We did the first test in GameaRanger. 4 players (Nero, DrPepper, Monty and me) -I know, we didn't respect the head-to-hear principle, will do 1vs1 tests later-. We did 12 races 4 laps. I was the host.
You know that GameRanger has a Ping detector whe somebody hosts a room. When I hosted the others saw my room showing 380ms ping. I thought it would be very laggy because my connection was the slowest and we are far away, but from my point of view the test was Perfect. Nobody lagging for me, the car positions were super stable even more than in 0820, no radar bug for me, physics and steering feel amazing, like the old re-volt. It was a pleasure to race on this patch.

On the other hand, the guys told me they noticed radar bug in few occassions, for monty it showed like if I was 4th (behind him) but I was in fact 3rd (before him). They noticed a bit of lag but I don't know what kind of lag exactly. Note that I was the host. I didn't notice any radar issues in all the races. Also no lag, they all seemed stable in all the races. I think teleporting is fixed.

For the host = great session, stable car positions, no teleporting. Lag was almost unnoticeable or don't disturbing for me.
For the guests = they all saw me lagging, i can't tell if the lag was rude or just radar lag. They'll have to confirm it.

I think the guys in RVHouse were testing it aswell

<hilaire9> with new patch I am the Jesus of Revolt

Don't take it literally.

1 question for you:
- We don't know what extrapolation / interpolation are. This is too technical language. Could you explain it what they are? Or in what way they affect the gameplay in more simpler words?

Phantom @ Mar 6 2014, 07:31 PM wrote: I think the guys in RVHouse were testing it aswell

<hilaire9> with new patch I am the Jesus of Revolt

Don't take it literally.

you can take it literally, i never see hil win so many times so easyly !

it seems that i lag much more with test F &G than with 13.0820 ... don't realy know why

Phantom @ Mar 6 2014, 07:31 PM wrote: - We don't know what extrapolation / interpolation are. This is too technical language. Could you explain it what they are? Or in what way they affect the gameplay in more simpler words?

Its basically a method for the game to predict what other players in the network are doing when you don't have information about what they are doing right now.
The purpose of this is to make an online race more fluently in case players have a bad connection where you don't get as much information as you need to display proper movement/action.
But it can of course also end up displaying weird results if the latency is too high (like the teleporting I assume).

Normally this doesn't really affect gameplay of course (at least for the average bad connection) since this is only used for the moments where you don't have any info about the player with a bad connection and even in those moments it depends on the implemention what happens in the game.
Once you receive the information from the player it gets properly updated again (which usually happens all in the millisecond timeframe).

However I don't really know how it is implemented in Re-Volt and if it works like its supposed to do so I can't say if its the reason for the teleporting, jigebren can probably give deeper insight.

When i hosted with this build G tonight, i remembered all the good times I spent playing revolt in the 2000s with version 1.1 !! Good work !

Thanks for the fast feedback. Well, it seems promising, so I can think about implementing per player PPS now (though I'm not sure which of point 1 or point 2 in the changelog has brought the most benefit).

Too bad about the radar issue. BTW we'll have to be clearer on what we're talking about: is it just the player position which is wrong, that is to say it display 3th instead of 4th but the distance look almost correct? or is it the radar bug we mentioned previously, ie. the distance that suddenly jump with +/- the lap total length?
I'll also need to know if it can occur anywhere, or rather when crossing the start/finish line.

Phantom wrote:For the guests = they all saw me lagging

Seems consistent with the current PPS code. I have hopes it will be improved by per player PPS management (but I don't want to be too confident).

SebR wrote:it seems that i lag much more with test F &G than with 13.0820 ... don't really know why

What do you mean by "I lag"? All the remote players lag? Just for info can you PM me your up/down bandwidth as suggested in my previous post.

About extrapolation / interpolation, I can't explain all details, it would be too long for this thread, but Kenny already draws it roughly, except that it's actually always used (even in LAN with very small ping). This neat schema will try to explain.

Packet C is sent from remote player at time C. After ping/2 ms it is received by local player at time c. When you receive it you can display it as is, but keep in mind that it is already outdated (because of ping). And what value do you use after c and before another message is received?

We have the choice to either:
- try to guess what the current remote car position is at time 1: it's extrapolation.
- apply the position as it's received, in that case the remote car position is always outdated for the local player. Not to mention an extra delay has to be added to allow interpolation.
Here we use A, B and C data to extrapolate (ie. to mathematically guess) what it going to happen between 1 and 2.

The other solution would be to delay everything and to simply interpolate between the values. That is to say at point 1 we use B data, and between 1 and 2 we interpolate between B and C. There is no prediction error in this case, but the delay is increased even more. This is not used in Re-Volt.

jigebren @ Mar 6 2014, 08:18 PM wrote: Too bad about the radar issue. BTW we'll have to be clearer on what we're talking about: is it just the player position which is wrong, that is to say it display 3th instead of 4th but the distance look almost correct? or is it the radar bug we mentioned previously, ie. the distance that suddenly jump with +/- the lap total length?
I'll also need to know if it can occur anywhere, or rather when crossing the start/finish line.

We were testing again, 3 players this time (kikizana, Balint and me). I was the host.

From my perspective there was absolutely no lag, the car positions were stable all the race. And I didn't notice any radar issues at all. Since nobody lags for me (the host), I can't tell you what they see but Balint told me I dissappeared from the radar in one occasion (in Ghost Town 2). Then he said that the value in the radar fixed itself and the position went back to normal after 30-40 seconds. I don't know if it is important but the radar fixed itself right after we all stopped racing. I don't know if it fixes because we stop moving so our positions are easier to locate or.. if it is normal that it fixed itself after a period of time. Balint also said the radar showed the value 531m from his perspective (and note that GT 2 is 490m lenght). From my perspective nobody dissappeared and the radar worked fine all the race.

Overall, this patch is awesome. The stability and the weapons feel so well. Also I think you fixed something regarding the problem with minimizing re-volt when -emulatefullscreen command is enabled, because it doesn't give me the "can't restore primary display" error anymore. It still has reduced fps when it is enabled but I think it will be about time for that to be fixed aswell. I just hope Huki doesn't include the new Time Trial system on this and I would be using this patch forever.

I late joined to the game, and at first I saw quite big lag (teleporting, cars not moving smoothly at all, but hopping around), but once I actually joined the racing, it was gone. After that I didn't see this sort of lag anymore.

About this radar bug. I was (naturally ) leading, I saw in the rearview mirror, that Phantom was like 10 meters behind me, but my radar wrote, that Kiki is behind me, with 50 meters (which was originally correct, only that Phantom was between us). Later we stopped and I let Kiki overtake me, so I see how far Phantom is, according to the radar, and it wrote, as Phantom said 531 meters. We stood still, and a little later it got fixed, so I saw that Phantom was actually about 40 meters behind me. So I can confirm again, that the radar was wrong with exactly 1 lap.

As I'm not in a good form now, so I don't see exactly, if the physics are the good old one (but if Dolo is pleased, it must be) but after this little test, I say that this patch seems really good. Well done!

Phantom @ Mar 7 2014, 01:42 PM wrote:I don't know if it is important but the radar fixed itself right after we all stopped racing. I don't know if it fixes because we stop moving so our positions are easier to locate or.. if it is normal that it fixed itself after a period of time.

ReVolt sends every 20 or 30 seconds (20s originally, 30s in some builds, 20s in last build G) some info about the player number of lap, distance, etc. These data are likely already outdated when received (since the car has moved between the remote player sending it and the local player receiving it), so for high ping I believe this might cause the radar bug (just after the car of a remote player is updated with these info).

But when a car no longer moves, the received data are nicely up to date since the car position don't change for both the remote player and the local player (even if the ping is high). So before the next 20s the radar will be back to a proper value (when the car is updated with new data).

I believe this code could be doing more harm than good, except maybe if a player went totally out of sync for a while for some reasons. Which means simply disabling it is not an option...

Balint12 wrote:I late joined to the game, and at first I saw quite big lag

In this case I noticed the race timer are no synced, so you see what's going on when the extrapolation code is disabled... This will be fixed eventually.

We tested the 14.0306 yesterday, in a 10 player race, and I have to say the players were not at all happy with the build. I, personally, had no problem other than the (very very serious) radar bug, but they kept saying "this build is not good" "why are we using this" "blabla" (maybe a bit more harsh)... I don't really know what they experienced, they said it lags, and that the radar is crazy. I don't really had a problem with lag, only with certain players.

About the radar bug: it was REALLY heavy this time. Sometimes I saw I was 2nd, when I was actually around 6th place, and it lasted quite long. This way it was really disturbing, cause it didn't happen only with 1 player at a time, but my radar didn't see half the people, almost throughout the race. Sometimes I had really no clue, which position I have.

What the reason for the other players to create that harsh critic about this build is, I don't know, they should report it. (Hil, DrPepper, Didaka, Rafster, Passion, Zappa... I don't remember the others)

With the 0306 patch, SebR teleports like crazy. He is practically invincible this way, nothing will stop him. Want to zap him? Nope, not gonna happen. Rockets? Nope. And what if he has a zap? You better pray he doesn't teleport to you.

Sting also lags like crazy, gliding all over the track and also being invincible because of it.