Thank you.

Quote from sicotange :

I have nothing worrying to report in relation to my Windows server (WS • Metropolis) using 0.6B3.

Am I right to assume that:
• The probability to encounter the error on a full server is much higher.
• The issue concerns both Windows & Linux servers.

- Network activity does also fill the buffer, so it increases the probability of encountering the overflow. But it can happen on an empty server.

- I think the timer / overflow error is not likely on a Windows server unless it is overloaded and fails to give LFS any CPU time.

So, it's good for any servers running on Linux / Wine, specially if they have the clock adjustment problem or ever see OVERFLOW in their log and sometimes have to restart their host.

Thanks Scawen.

Thanks.

Hopefully this solves all the issues

Quote from Scawen :

Here's another dedicated host update, 0.6B4

Thanks alot. New version is now set up on TC servers. Results will follow soon.
You're welcome to join

Maybe not right place to ask. But scawen can live scoring (lap times, finishes etc) be expanded another digit, because there have been quite a few ties in history of league racing on ovals and even on finishes not just qualifying.

Quote from Scawen :

Here's another dedicated host update, 0.6B4

Awesome!

Updating....

Quote from Bmxtwins :

Maybe not right place to ask. But scawen can live scoring (lap times, finishes etc) be expanded another digit, because there have been quite a few ties in history of league racing on ovals and even on finishes not just qualifying.

Pretty sure that is an engine limitation, I remember a discussion about that topic but can't find the link to it anymore.

Quote from Bmxtwins :

Maybe not right place to ask. But scawen can live scoring (lap times, finishes etc) be expanded another digit, because there have been quite a few ties in history of league racing on ovals and even on finishes not just qualifying.

Photo-finishes can be checked manually, I'm sure I've seen one where one car lost due to a crumpled front end.

Identical times happen in real life too, drivers are ranked in the order they set the time.

Just spotted a wrong userlist on LFS remote. Only LFS remote is doing this.

* Oh and numbers 1,2,3 are mentioned twice. Dunno why, just grabbed my attention. Can be something which is already going on longer.

.

What this? (Joos - car)
testing on server "Blackwood XFG XRG"

Quote from [Audi TT] :

What this? (Joos - car)

A .vob mod. Quite often in demo servers having a XRG on track. The funny part - usually guys with mods can't think that far that the mod is resulting the kick from server,seen guys registering new username and trying to connect again...

He's probably using Car Mod.

That's why appears JOOS-CAR

Thanks in infornation.

Quote from Bean0 :

Photo-finishes can be checked manually, I'm sure I've seen one where one car lost due to a crumpled front end.

Identical times happen in real life too, drivers are ranked in the order they set the time.

Car damage has nothing to do with time when LFS will recognize you have crossed finish line , infact LFS doesn't even care when you cross it with front end. Try doing this in BF1 and UF1 on some short costum layout, you'll see what I'm talking about.


EDIT:
Blue line marks first split, and car is left at position when LFS reported split time.

B4 Seems to have been working perfectly on the TC servers all evening.

I don't like to interrupt racing so I just finished completing things which needed to be done. Now lets see

Quote from DANIEL-CRO :

Car damage has nothing to do with time when LFS will recognize you have crossed finish line

It does when two cars cross the line at the very same time and then you (a human) judge who was first. The one with smashed front will lose by a few inches, so damages does matter.

After 6h:
(Server 0.6B4)

5x JOOS - CAR
1x JOOS - PLY COL SET CAR WHL ENG
1x JOOS - SET CAR WHL ENG

I have one innocent question regarding timeouts. ^^

I'm sure it's my ISP to blame and i'm fiercely working on finding better one, but i was curious, if there is any possibility of lowering chance of timeout in case of few (2-3) lost packets over course of 20-30 seconds, in some simple way in LFS dedicated host, or if it would require further digging into LFS's own netcode.
Under normal circumstances (low ping, on half filled or full server), it happens to me only in LFS, i've never had timeout in other sims or games.

The dedicated host will disconnect a guest's computer if it does not hear anything from it for 10 seconds. I considered increasing this, to maybe 12 seconds or so but decided not to because there is a problem with it. The problem is that if a guest is very laggy and packets take so long to arrive at the host, the game becomes unresponsive to other guests because of the "in-sync" part of the game. LFS does not continue processing TCP packets that have arrived while it waits for the packets from one of the guests. I think maybe this is a fundamental problem or feature of the way that the LFS packet system works.

Quote from Bean0 :

Photo-finishes can be checked manually, I'm sure I've seen one where one car lost due to a crumpled front end.

Identical times happen in real life too, drivers are ranked in the order they set the time.

kyoto 250 once (sean williams and someone else for 2nd or 3rd place)
But for qualifying purposes and WR purposes it would be better to have that. LFS is the only racing sim/game I play that doesn't do this. Even 12 year old NASCAR Heat has it down to the third digit after the period.
maybe in s3

Quote from Scawen :

LFS does not continue processing TCP packets that have arrived while it waits for the packets from one of the guests. I think maybe this is a fundamental problem or feature of the way that the LFS packet system works.

Fixing this would be a very good improvement for many reasons and one that I can think of is on is that on servers like cargame.nl s2 where there are a lot of people and when a crash happens you cant pit or spectate.

Quote from Scawen :

The dedicated host will disconnect a guest's computer if it does not hear anything from it for 10 seconds. I considered increasing this, to maybe 12 seconds or so but decided not to because there is a problem with it. The problem is that if a guest is very laggy and packets take so long to arrive at the host, the game becomes unresponsive to other guests because of the "in-sync" part of the game.

It's good thing you left it that way and personally I would lower it to 8 or 5 because even 5 seconds is a big lag imo and surely something wrong with that connection and it should be fixed not the game.(now a lot of you would want shout at me for what I said, please don't do it, I am aware of the facts)

Scawen, related to last part of your post: when a new guest is in the process of connecting to a server its connection process takes priority over pretty much everything. Existing users can't enter spec mode, can't shift+p, can't even change any of the options while a random even such as guest connecting takes place. If a guy happens to connect from another side of the planet con process can take up to half a minute cause of ping, effectively disabling everyone to spec, pit and don't know what else is tied to it. I mention spec & pit above else cause it prevents people to avoid avoidable accidents by specing or pitting.
I understand the reasoning behind it but i hope you can figure out a better way.

Sry for this not being in suggestion thread, i know you don't particularly like off topic suggestions; reason is you seem to be 'in the zone' of this packet thingy now and i wanted solidify your thoughts about packet system

22.5 hrs runtime 0.6B4

548 connects
3x JOOS - CAR (same user, one occasion)
1x JOOS : Resync RES (1)