Quote from duke_toaster :

Inflation is an issue if you want to compare drivers at different times. That's why chess players moan and whinge about the greatest chess player of all time. But if you aren't comparing different people at different times, ELO should do the trick.

There is Glickman and Chessmetrics as well.

Haven't checked those Glickman and Chessmetrics. The normal ELO is only to rate the results on the basis of win/draw/loose system. It would need some tweaking to rate race stats on time difference basis.

Quote :

World Football Elo Ratings uses different K (volatility) factors. If the win is by two goals, the factor is increased by 50%, 75% for a 3 goal win and so-on and so-forth. Maybe 1% per second?

Yes, but it needs the testing to choose right kind of values. If the values are not correct you get feeling the system doesn't track the player skill level correct. The time difference off course is affected by the lap count or actually the total race distance (those affect how much are allowed to loose). I understood MS made large number of statistical analyzing to choose their values for TrueSkill. The system in LFS would need some kind of testing too to choose them. For the best reliability it should be used in large number of servers and best if it was general LFS feature (stats in LFSWorld).

Quote :

If InSim provided us with packets on collisions, the wrecking thing would be simple. A set number of collisions (maybe excluding T1) ==> Spectate.

But in making formulas you need to do at least some handwaving. The tennis rankings (for example) do not take in to account court types, colours are not taken in to account in chess etc. All of those other than the wrecking are simple. Put enough fuel in, and make sure you do the pit stop. I get it out of the way early (second or third lap).

Yep. It was just the point in the "out of fuel" case that it doesn't really make the driver slow (can happen to a good driver too). And alter the skill rating (ELO etc) on the result of that race is quite hard. Basicly it would be DNF result, but again the issue of how much to substract from the skill level.

In the end I think the rating would show little bit more the racer reliability/consistency than the speed. But of course the speed would have an effect too. Then the rank should be calculated for every combo or atleast every car. Like I'm ok (I think) with XFG etc, but I suck with FO8.

Quote :

IMO an ELO system is better than the STCC system of points for positions, more points on higher servers as it's not related to strength of opponents. I mean, if you are racing biggie and bawbag you should get more points for coming second than for coming between me and someone else painfully slow.

Yes. Actually that's why I haven't made it so that the points are saved. Simply because you cannot compare them between different races. That would indeed need the ELO type rating system to make it wise. Without that the point amount would have a strong correlation to the amount of races driven and less to the skill level. And I've understood that the skill level should be the idea behind the STCC system. Atleast they give some kind of 'licenses' and allow only certain licensed persons to enter certain servers.

Actually I wouldn't like to see the ranking system to be used to categorize people into different classes with the sole idea to kick out people who are not high ranking enough.

@duke_toaster
No no. Some good ideas. Some of them require bit more coding. The idea of multiple servers connected to the same points table could be expanded to 1 lap time/stats log for multiple servers (but that is a large thing to do). Other points...not so hard.

ELO ranking... I studied it a bit. The original ELO system doesn't work well. Needs some tweaking but basicly it would still have issues. The MS TrueSkill is in some ways better (XBox), but more complicated. One of the bad sides is that all systems take some amount of races before the calculated skill level is settled to the players true skill. In racing the time difference should be the measured variable (IMO) (e.g. 0.01 sec difference is basicly a draw) and how much value e.g 1 sec difference is. It would take some intensive adjusting.

Another thing is that in racing all kind of abnormal things happen like: out of gas, being wrecked out of race, forgot to make the compulsory pit stop, etc. Should those be calculated to the skill level and how. And those who wreck their car, teleport to pits and begin the race again. Chess starts to feel so much simpler game.

I'm not completely abandoning the idea, but it needs more thinking. In the end you should think "does it bring anything to the game?".

Not just Israel. I don't consider Turkey to be part of Europe either (Turkey is in ESC and wants to be in EU). If you say that it is Europe because it has a little bit land on Europe's side, then Spain should enter African Union too.

Quote from otolikos :

Vcom improvement suggestions:

-Idle tool : when a player do nothing during 30s 45s or 1 min he's autokick in -spectate mode
...

Because you requested it again, I just did that tonight. Seems to work, but needs still some more testing (nobody wants to be spectated w/o a reason). (It works only in server and with players having account names (NOT demo?) )

User can give the time allowed to idle before kicking (e.g "45sec"), the speed limit (below which it is considered to be idling) and the warning text (e.g "Drive or spectate noob!") which is printed middle of the screen (together with the time left before being kicked). The time ticks down from 10... to 0 (seconds) with the text.

The other suggestions... I think them. Actually the second one is something I've been thinking already.

Quote from Vain :

Südschleife.

That'd be pretty unique.

Vain

What about an 'Old Ring' area with 3 tracks:
- Nordschleife
- Südschleife
- Combined (Nord und Süd)
Like it was in reality before they scrapped the Süd in the early 80s. There was also some shorter piece of track in Südschleife used for practising.

Then we would need some old classic car for those tracks.

@mrodgers

I don't like driver aids. However if you want to hear the "whee" sound it should be in NO position. You probably get little bit better times with that in NO position (hotlapping etc). Not realistic and definately not recommended to do in real life.

Have you set "throttle cut on upshift" as NO? Should be NO, except maybe FXR (gets engine damage).

EDIT: Damn, beaten

"...nb of races won by a racer and his total score points..."

In a way nice idea, but...

I haven't made any permanent (saved) statistics of the points, because it is not fair in 99% of cases. In F1 or WRC the group of drivers is always the same so the 10pts here is usually as well earned as 10pts somewhere else. But in LFS races you could score easy points if there is only a group much slower drivers against you. It should somehow take into account the count of racers and how good they are. So some kind ELO ranking points (like in chess) except for multiple players... hmmmm... ELO ranking... hmmm... got to go... got something to think

Quote from otolikos :

I've some problems to read .xml with ie7 but works perfectly with firefox and opera

you can see .xml edited files at: www.6traxgtrleague.tk

Yes. The M$ boys don't bother to put much effort improving their stuff. It usually complains some unknown unicode char. They don't put much effort in following standards either. If you want to see it as intended I would use some other browser (I use Firefox). Of course it would help a bit if people didn't go crazy using every odd looking char they can find in their names (because IE doesn't handle all them)

So far I haven't found an IE proof solution, sorry. (if you know one, please tell me too)

Btw. Besides atleast I had to download a new unicode font to see most of the special chars like the Japanese Katakana letters (otherwise they look like a box or a question mark etc depending the browser used)

Quote from otolikos :

Does it be possible to change the .xml output names?

for example _replace 03 by March and add some _ between numbers for bettter reading....

Currently no (hardcoded). If you really want that now, you need to make some script etc to change the names. Atleast the number system is language independent

Quote from Davo :

Anyone got screenshots or what this game is about?

Not quite sure, but you can watch a chunk of meat floating in a toilet

I have one thing you surely (said in a sarcastic voice) want to discuss ...

You know the "Cretan liar paradox"? The great Cretan philosopher Epimenides said someting like "All Cretans are liars".

So how can this be? If all Cretans lie, the statement cannot be true because Epimenides was Cretan (and then Cretans would not be liars and that contradicts Epimenides being a Cretan). Or if Cretans do not lie and the statement is true. But the statement just said that all Cretans are liars... :nut:

So think about it, but not too much... I read someone got nuts after thinking it too much.

/mso vcom out sb <car id>

For example /mso vcom out sb ALL
Car id is a 3 letter car or group id.

Comments on the manual line 716.

quick answers...

"Does it be possible to generate a daily and a weekly results? "

For best laps? Well I made this for the lastest version, but the release is still held back by "something" I'm still working on.


"can't generate sb.xml files"

You need to type it on the server console unfortunately. The new version would help greatly (see above).


"Does it be possible to upload on another ftp site where vcom isn't running?(my webspace isn't on the same ftp than dedicated lfs server)"

You can give the output directory where it is generated. If you cannot refer to this public web space (it can be on other computer too) from the computer where the vcom is running (i.e by paths like "~/public_html/lfs/results/" or "Z:/lfs/results", etc - in those 'public_html' and 'Z' would refer to directories over network), you need to do it manually (by transfering the output files via ftp etc) or by generating some auto run script.


"It would be cool if daily and weekly results could be in .html"

All web browsers can show XML/XSL without (much) problems. Today many programs save the data into XML. If you are not happy how the data is shown be free to make your own XSL files and changes and make your own graphics.


"What's the future of vcom? Do you still work on the project."

I think this is partly answered above. I have coded or will code the things I've been personally interested to try (past or future). I try to fix bugs if people report.

I personally use it to play music or see changes in setups (offline).

Oh... if Ati indeed gets it out so soon (somehow I understood it takes longer) then it indeed is a good idea to wait now. But anyway it hasn't been good for a consumer that only NVidia has had DX10 chip available. So really interesting to see when the first performance comparisons and prices are available.

The chip is quite good/fast (never heard about the card manufacturer). It should be good in performance/price ratio.

I was thinking the same thing from Sapphire for AGP (as a replacement for Radeon 9700). However I started to suspect it is unnecessarily powerful for 2.8GHz HT P4 (comments to this are welcome)

About GF8600. Does anybody have idea how much they are going to cost? As long as Ati (or AMD) has no chip for the same generation (to match GF 8xxxx cards) NVidia has relatively free hands to price them.

Yeah, well there was a catch if you wanted to use all the 4096 colors at the same time (in AGA Amigas the max was 262144 of 16.7M palette). Usually games used max 32 colors in old Amigas or in few occasions 64 (so called half-brite mode, where the other 32 where half brightness of the first 32). The AGA Amigas had all the modes in old ones plus at least 64non HB, 128, 256, and 262144 HAM8 color modes, plus the higher graphic resolutions.

Actually the nicest thing was all those odd features like having two screens visible at the same time with different dimensions, resolution, color count and palette (can be seen for example in Test Drive, where the data sheet is 640x200 and car screen behind it is 320x200 resolution iirc). And you could move them up/down left/right. Also the screen above could have a hole in it showing the background pic (but then the 2 pics shared same max color count or something like that iirc)... nuts

About 2nd Reality. It is about the same level as the best AGA demos IMO. But when it came it was great (I had it too). Especially the music rocked. The biggest problem in Amiga demos was that they usually had to fit 1 Amiga disk (880kB), because not so many demo viewers had a HD. Also not so many had more than 1MB of mem. In case of the 2nd Reality the music alone takes 587kB for Purple Motion and 383kB for Skaven (ScreamTracker modules). BTW: The final city flying music was reused in the Final Reality 3D gfx card test program (by Remedy, maybe c 97-98).
(I still have both the 2nd reality mods and the final reality mod)

Oh boy oh boy... now the old Amiga freak inside me is waking

About Atari ST being better.... dunno about that. I remember them being quite equal. Both had good sides, but Amiga was more popular.

I still had 2 Amigas last summer. One was A500+ with 40MHz 68EC030 GVP turbo and 8MB of mem and 120MB SCSI HD. It was quite kick ass computer and I felt that it was like 486. At least it was nice for simulators. I dropped them to junk yeard last summer... already regretting. :banghead:

There is really no need to convert old MED(or octamed) and MOD music (unless you really want to). Mod can be played by many players and med can be played by deliplayer (http://www.deliplayer.com/) among with all kind of other old Amiga music formats.

For emulator I would recommend Winuae (or other UAE ports for different systems). It runs programs really well and I have even Workbench 3.0 (Amiga's graphical OS) installation with professional gfx card (Picasso) emulation (you can get flicker free 1280x1024 in your 'Amiga').

It is python? I think I tried it year ago, but cannot remember if I downloaded anything else. I think they cannot stop you running it if you have the interpreter which runs the script. Just run the exe it has. Otherwise check if there are some other relay (sometimes called gateway) apps.

Just be warned that there might be some dual printings if some of the commands (!<something> style commands) collide with lfslapper. Also I don't know what those relay things do if some of the programs require mismatching settings in init (InSimInit) (one of the reasons I haven't finished relay in vcom).

At least with IE7 I can get SO4 (SC Long) like...
FXO_GTR_so4_Norbi1741.set

I think it should be "FXO GTR_..."

http://www.youtube.com/watch?v=wZ7YedEopp4

Yeah, was thinking, but...

the idea itself is good, but making it 100% proof is another thing. Damn annoying when spin out of track and be forced to spectate on the last lap. If it would be able to spot true idlers from those who are just waiting for a gap in traffic to enter the race.

Ok, now I start to get an idea what's wrong.

It clearly misses at least the track data BL1, so most probably it has an error between the rotate and the first track data section (BL1). Check again that it has that closing } for the rotate section. That means after the closing } for the last rotate/track section there must be closing } for the rotate too before the track data sections. ...


This same thing might be causing the rotator goes nuts. I guess it goes on and thinks the first track data section (BL1) is part of the rotator entries.

Argh... I knew this C looking format can cause trouble (but it is very good for making nested sections). I guess I have to make it fail to start if braces are missing or do not match otherwise

EDIT: Idle function?

Does it do that always? Because I cannot reproduce it myself. I tryed with a shorter list of 1 lap races and it always goes back to first.

The only error I could find from your setup is that it doesn't close the rotate section (} missing), but I guess it is just missing from this paste.

Anyway I look more to it (and try to reproduce it) later.

In the commands.txt file (LFS files) I found these

I thought that if I make a script (vcom.lfs) which has /exec (something...)/vcom.exe (something...) and run it with /hrun vcom, but it keeps complaining that exec has a bad parameter. I cannot run it even with straight typing /exec (something ...). I don't know if it would help because you must run vcom from it's own directory not some other like lfs directory. Of course I could implement some restart command in vcom, but it would be nice if the script thing worked.

You can run multiple instances of vcom. There are few things to note...

1) You need multiple copies of vcom (copy the same stuff into another directory). This is currently a must as you cannot give another setup file (another instance of the exact same exe would run the same setup file). I fix this to next version. Vcom installation is quite small so it should be a very big problem.
2) As you run the servers in the same computer (?) you need different insim ports for the servers and also for the vcoms. Disable the car radio (e.g. give mode=65471; in the setup file(s) ) as you don't need it in server (otherwise you can get this kind of error...

...because the the car radio(s) are trying to use the same port

EDIT: I actually fixed it already that you can have as many instances as you like (without copying the program) as long as you have different setups for them. I just would like to add some more stuff before releasing it.

Last edited by Aquilifer, Tue 13 Feb 2007, 23:19.