torcs-devel Mailing List for TORCS - The Open Racing Car Simulator

Hi all,
I had recently sent a mail regarding creating a storyboard for TORCS, my
team and I are currently working on a 3D game based on the TORCS engine. So
we had a few doubts about TORCS and how things work in it. I hope we get
your support. Thanks again

So here i go with the questions:

1) Is there any difference between running the IDE Visual C++ Express
(Visual Studio 2005) and Visual C++ 6.0.
- Im currently using VC++ Express and i cant compile with it.

2) I tried compiling TORCS from source 1.3.0 in VC++ Express but it said no
windows.h.
A lot of header files in wtorcs project is missing other than osspec.h and
windowsspec.h. The other header files like driverconfig.h, exitmenu.h,
mainmenu.h, optionmenu.h, raceman.h, splash.h, torcs.h etc are not there in
the zip. When i load up the torcs.dsw, this is wat happens. So just wanted
to know if this had something to do with the new sources 1.3.0 maybe missing
out on a few files or maybe the headers werent required.

3) I needed some help creating a track:
- I tried creating many a track, i used trackeditor to create a track and
then trackgen to make the ac file of it. Now i wanted to put 3D objects into
the map to give a real feel to it. Like make the track based on a real place
wit the real surrounding and elements. So i load up ac3d and try to edit the
ac file but im still having difficulties wit the file. Also how do i create
the accc file, im using a guide available onlien called "build your TORCS
track in 20mins." Im finding it slightly hard to understand but i think
maybe if some1 could help me a bit maybe i can get it done. Is it possible
to see what all are the different images available in trackeditor which are
given to the track.

4) Videos in TORCS:
- Im planning on introducing a few videos to the game which we are producing
ourselves. The storyboard which i had mentioned in the previous mail is
about a group of college kids who create a racing gang (sortof like
underground racing league or something) so we are going to be featuring
videos of us etc. BTW i hope thats ok wit u guys.

Dear Sir,
I am working on my final year conputer science project which
is a 3D Race simulation and as such my team and I are
developing the game based on the game engine of TORCS. We
were planning on introducing a storyboard to the game along
with introduction of many new tracks and new cars to the
game. We were wondering if we could expect your support in
the same. In case we run into any trouble on the way. I hope
the storyboard can later on be incorporated to the TORCS
game itself.

Thanking you
Regards

Antony F. M.
+91-989-517-3456

Hi Christos

>> GTR as client IIRC (UDP is for many games just required if you run it as server,
>> maybe that the central server which announces all the currently running games is
>> unloaded, no TCP retransmissions for lost packets etc.). Loads of games (at
>> least clients) use TCP only today (quite easy to recognise if you have not to
>> open any ports/protocols on the firewall). All other things scream for problems
>> with todays network infrastructure IMO.
>>
> 
> Really?  I guess the TCP downside is the delay when packets need to be
> retransmitted..  How can we get around that?  By discarding 'old' packets?

We cannot get around that with TCP.

Bye, Bernhard.

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

Bernhard Wymann wrote:
> Hi Brian
> 
>>> Wrong, for internet play you cannot use UDP, you can
>>> not pass any realistic
>>> infrastructure with this. My implementation will
>>> rely on TCP, like most
>>> commercial games on this genre.
>> Sorry, I disagree as you can see at this website many
>> games use UDP ports especially racing games such as
>> NASCAR 4.  http://www.gameconfig.co.uk/  
>>
>> Bernie what would be an example that uses only TCP?
> 
> GTR as client IIRC (UDP is for many games just required if you run it as server,
> maybe that the central server which announces all the currently running games is
> unloaded, no TCP retransmissions for lost packets etc.). Loads of games (at
> least clients) use TCP only today (quite easy to recognise if you have not to
> open any ports/protocols on the firewall). All other things scream for problems
> with todays network infrastructure IMO.
> 

Really?  I guess the TCP downside is the delay when packets need to be
retransmitted..  How can we get around that?  By discarding 'old' packets?

Bernhard Wymann wrote:
> Hi Donnar
> 
>> *How many bytes have to be send between the players and the server ? 
>> *How many bytes have to be send between the players only ?
>> *What is the maximum timeframe ? 
> 
> The implementation which I plan has different sort of messages, but just the
> very frequent car updates are interesting in terms of performance, I need at least:
> - Player Identification (you need to know to who the following data belongs) (1
> byte)
> - The car position in space (max 12 bytes).
> - The car orientation in space (max 12 bytes).
> - The car velocity in space (for inter/extrapolation) (max 12 bytes).
> - The steer angle (for proper graphics and interpolation) (max 4 bytes).
> - The engine rpm for proper sound (max 4 bytes).
> - Vertical wheel position relative to the car body (max 16 bytes).
> - The ping (2 bytes)
> - Maybe I missed something
> 
Perhaps break/accel and rotational inertia of the car.  If we want the
sound for other cars' tyres and collisions then we need to pass some of
the relevant variables too.

On Fri, 23 Mar 2007 21:32:42 +0100
Bernhard Wymann <be...@bl...> wrote:

> To be honest I do not care, people can always create cars and put it
> on the net if they are not happy with this, then it is up to them to
> take the legal risk for their content. I am just responsible for the
> content which I have created, so I try to reduce the risk of legal
> trouble on all artwork I touch:
> - Removing company names/trademarks.
> - Not following the design of real cars.
> - Not abusing product names.

Ahh, now i understand. This is the same way that rFactor
developers used. rFactor ships with some fictional manufacturers and
cars. They even describe this fictional story in game manual. Community
modders create real cars and take any risk of legal trouble.

Also i think that you should state this explicitly somewhere on the
website or in documentation. For ex. i didn't know that some cars just
look like the real ones and there's is no aspiration to make them
behave like the real ones.

-- 
Alexander Tsamutali

Hi Brian

>> Wrong, for internet play you cannot use UDP, you can
>> not pass any realistic
>> infrastructure with this. My implementation will
>> rely on TCP, like most
>> commercial games on this genre.
> Sorry, I disagree as you can see at this website many
> games use UDP ports especially racing games such as
> NASCAR 4.  http://www.gameconfig.co.uk/  
> 
> Bernie what would be an example that uses only TCP?

GTR as client IIRC (UDP is for many games just required if you run it as server,
maybe that the central server which announces all the currently running games is
unloaded, no TCP retransmissions for lost packets etc.). Loads of games (at
least clients) use TCP only today (quite easy to recognise if you have not to
open any ports/protocols on the firewall). All other things scream for problems
with todays network infrastructure IMO.

Pure technically I agree that UDP is the "right" solution, because you are not
interested in retransmission/reliability of outdated packets.

Bye, Bernhard.

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

I send the CarControls and CarPhysics structs for each
car plus the time of the packet(2 bytes), the type of
packet(1 byte) and the number of cars in the packet(1
byte)  About 78 bytes per car.  I think I can reduce
this number by using fixed point for some of the
values in the tDynPt structure.  The player just needs
to send the carcontrols structure to other players and
to the server.  

struct CarControls
{
	char	index;     /* Car index */
	char steering;//Fixed point between -1,1
	char throttle;//Fixed point between -1,1
	char brake;//Fixed point between -1,1
	char clutch;//Fixed point between -1,1
	char gear;
};

struct CarPhysics
{
    tDynPt	DynGCg;		/* GC global data */
};

I have found that AI robots require the most frequent
updating because they adjust their steering, throttle,
and brakes much more often then a human player does,
and it is very difficult to predict their position
especially after a collision.  In a race with only
human players you can get away with a slower update
rate.

> Wrong, for internet play you cannot use UDP, you can
> not pass any realistic
> infrastructure with this. My implementation will
> rely on TCP, like most
> commercial games on this genre.
Sorry, I disagree as you can see at this website many
games use UDP ports especially racing games such as
NASCAR 4.  http://www.gameconfig.co.uk/  

Bernie what would be an example that uses only TCP?

The weakness in my design would be that it requires
the server to send alot of data, but I would rather
have smooth playing game with fewer players. I do not
know how many players can play on my implementation,
but I do have actual cars racing each other on
different computers.  If I can play against 10 of my
friends on the Internet I would be happy.

Brian



 
____________________________________________________________________________________
The fish are biting. 
Get more visitors on your site using Yahoo! Search Marketing.
http://searchmarketing.yahoo.com/arp/sponsoredsearch_v2.php

Am Freitag, 23. M=E4rz 2007 21:32 schrieb Bernhard Wymann:
> Hi Alexander
snip
>
> To be honest I do not care, people can always create cars and put it on t=
he
> net if they are not happy with this, then it is up to them to take the
> legal risk for their content. I am just responsible for the content which=
 I
> have created, so I try to reduce the risk of legal trouble on all artwork=
 I
> touch: - Removing company names/trademarks.
> - Not following the design of real cars.
> - Not abusing product names.
>
> Bye, Bernhard.

I guess that people in other parts of the world than Europe don't know: A=20
manufacturer of toy cars was recently sued by the OPEL AG ( a subsidiary of=
=20
General Motors) for the use of the brand name 'OPEL' on it's toy cars.=20
Altough I consider it a stupid behaviour of this company (getting future=20
consumers associated to a brand for free), it's a fact. So Bernhards approa=
ch=20
is (unfortunately) right...

BTW: When I was a kid I played with a toy Volkswagen and nowadays I own 

Donnar

Hi Alexander

> I found the following here:
> http://sourceforge.net/mailarchive/forum.php?thread_id=30478647&forum_id=3774
> 
>> Ah, a word about that. I do not care much about such data, because these cars
>> exist(ed) in very different configurations. My goals are the following over the
>> long term:
>> - Anonymizing the car names/textures (no logos/names anymore, just "car7-trb1").
>> - Make them fun to drive, so every car should feel a bit different and have its
>> strength and weaknesses).
> 
> Are you sure that people will like anonymized cars which will look
> like the real cars and will have different physic parameters than the
> real cars? I think modelling real cars is a very important for TORCS.

To be honest I do not care, people can always create cars and put it on the net
if they are not happy with this, then it is up to them to take the legal risk
for their content. I am just responsible for the content which I have created,
so I try to reduce the risk of legal trouble on all artwork I touch:
- Removing company names/trademarks.
- Not following the design of real cars.
- Not abusing product names.

Bye, Bernhard.

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

Hi Donnar

> *How many bytes have to be send between the players and the server ? 
> *How many bytes have to be send between the players only ?
> *What is the maximum timeframe ? 

The implementation which I plan has different sort of messages, but just the
very frequent car updates are interesting in terms of performance, I need at least:
- Player Identification (you need to know to who the following data belongs) (1
byte)
- The car position in space (max 12 bytes).
- The car orientation in space (max 12 bytes).
- The car velocity in space (for inter/extrapolation) (max 12 bytes).
- The steer angle (for proper graphics and interpolation) (max 4 bytes).
- The engine rpm for proper sound (max 4 bytes).
- Vertical wheel position relative to the car body (max 16 bytes).
- The ping (2 bytes)
- Maybe I missed something

So for the first shot I think between 30 and 100 bytes (without TCP packet
header etc.).

>> I don't think Croquet would fit with TORCS.  TORCS
>> really just needs a simple UDP base network library.

Wrong, for internet play you cannot use UDP, you can not pass any realistic
infrastructure with this. My implementation will rely on TCP, like most
commercial games on this genre.

>> I would guess the largest game would involve about 10
>> players.    Alot of the network libraries are made for

Hmm, I disagree as well with that, I would like have online races with up to 50
players as goal (I know that this will not happen, but the implementation should
be able to handle this).

>> sdl_net is a logical choice for TORCS because it works
>> on well on MacOS,Linux, and Windows platforms.  It

Yes, I agree with that.

Bye, Bernhard.

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

Hi Brian,

thanks a lot for the reply. I was already quite sure that something like=20
'croquet' cannot be used for games like TORCS, namely because of the high=20
latency of croquet (or other distributed p2p frameworks).

Anyway, I would be interested in the functional specification a network=20
functionality to be used in a future release of TORCS:

*How many bytes have to be send between the players and the server ?=20
*How many bytes have to be send between the players only ?
*What is the maximum timeframe ?=20

I would be glad if you (or somebody else) could answer my question...

Regards,
Donnar

Am Freitag, 23. M=E4rz 2007 00:27 schrieben Sie:
> Hi Donnar,
> Sorry to take so long.  Don't worry about offending
> anybody.  There is more then one correct way to solve
> a programming problem, and my way isn't the only way
> either.
>
> I don't think Croquet would fit with TORCS.  TORCS
> really just needs a simple UDP base network library.
> I would guess the largest game would involve about 10
> players.    Alot of the network libraries are made for
> World of Warcraft type games with huge numbers of
> players, but do not require fast responses to things
> such as collisions or spinouts.  Different kinds of
> games need different types of libraries.
>
> sdl_net is a logical choice for TORCS because it works
> on well on MacOS,Linux, and Windows platforms.  It
> has been used by a lot of games doesn't have been bugs
> because of this.  The first version of TORCS
> supporting network play probably won't a Game lobby or
> a chat window unless you add it of course :)  The user
> would probably use a IRC/chat program to find out the
> IP address and then would just hand type in the IP
> address to connect to the server.
>
> I would suggest you take a look at library such as
> raknet (http://www.rakkarsoft.com).  It is what I am
> currently using and is a good way to learn a little
> about network protocols for games.  Unfortunately the
> license isn't GPL friendly so I plan on replacing it
> with sdl_net once I get the code working.
>
> If you are interested I could send you a zip file of
> my code, but it is still pretty rough and only works
> on Windows MSVC right now.
>
> Brian

Hi, all

I found the following here:
http://sourceforge.net/mailarchive/forum.php?thread_id=30478647&forum_id=3774

> Ah, a word about that. I do not care much about such data, because these cars
> exist(ed) in very different configurations. My goals are the following over the
> long term:
> - Anonymizing the car names/textures (no logos/names anymore, just "car7-trb1").
> - Make them fun to drive, so every car should feel a bit different and have its
> strength and weaknesses).

Are you sure that people will like anonymized cars which will look
like the real cars and will have different physic parameters than the
real cars? I think modelling real cars is a very important for TORCS.
Look at RFactor, they have lots of mods and many modders try to make
simulation as close to real word as possible!

-- 
Alexander Tsamutali

Hi Brian,

may I can ask you about your opinion about=20
'Croquet' (http://www.croquetconsortium.org/index.php/Main_Page) ?

Would it suit TORCS multiplayer gaming ?

Only a suggestive question btw., I don't want to offend anybody... ;)

Donnar

Am Samstag, 17. M=E4rz 2007 00:55 schrieb brian gr:
> Hi Donnar and Bernhard,
> I am not the official voice of TORCS but I will give
> my opinion.  I have experimenting with the TORCS code
> and I have been able to get a client server race to
> work on my LAN, and it is fairly smooth. The code is
> still pretty rough and lots of things left to do.  The
> biggest challenge I found was when a collision occurs
> between two cars.
>
> Here is my implementation
> 1. The server does all of the physics calculations and
> so do the clients.  Client assumes controls(steering,
> brakes, and etc) remain the same until the server
> passes all of the cars positions and control settings
> to all clients.
> 2.  Clients(human players) send data to the server
> telling its control settings.
> 3.  All Robot AI cars are done on the server.
> 4.  Server periodically(~5-10 seconds) updates car
> times, damage, fuel, and etc.
>
> Little trick
> Use Fixed Point integers and chars to pass car
> information to save reduce packet size.
>
> Brian

Hi Donnar and Bernhard,
I am not the official voice of TORCS but I will give
my opinion.  I have experimenting with the TORCS code
and I have been able to get a client server race to
work on my LAN, and it is fairly smooth. The code is
still pretty rough and lots of things left to do.  The
biggest challenge I found was when a collision occurs
between two cars.  

Here is my implementation
1. The server does all of the physics calculations and
so do the clients.  Client assumes controls(steering,
brakes, and etc) remain the same until the server
passes all of the cars positions and control settings
to all clients.  
2.  Clients(human players) send data to the server
telling its control settings.
3.  All Robot AI cars are done on the server.
4.  Server periodically(~5-10 seconds) updates car
times, damage, fuel, and etc.

Little trick
Use Fixed Point integers and chars to pass car
information to save reduce packet size.

Brian




 
____________________________________________________________________________________
Never miss an email again!
Yahoo! Toolbar alerts you the instant new Mail arrives.
http://tools.search.yahoo.com/toolbar/features/mail/

Hi Donnar

> I'm sorry to ask the same question again, but I would like to know if there 
> was already some 'brainstorming' about how to implement multiplayer 
> networking ? (Yes, I had a look at the 'devel' archive).

For me it is actually quite clear how I will implement it. To keep it short:
-There will be a server process to "host" the game, of course on the same
machine a client can run.
-The server is just responsible for managing the game state and to distribute
information between the clients, it will not simulate anything.
-Every client is responsible to simulate the "local" car(s)
-The client sends it data to the server
-The server distributes the data to the other clients.
-On the clients there will be interpolation/extrapolation to smooth the network
opponents ride.
-Collision detection happens locally on every client.
-I don't even try to avoid/detect cheating, I consider this as waste of time,
somebody will always work around it.

> Do the main developers like to use an already existing networking library, 
> i.e. 'libgrapple' (http://opensource.linuxgamepublishing.com/grapple.php) ?

I will use SDL for portability, that's it, I do not see any requirement for
using such a library.

> Or do they plan to implement a feature like i.e. p2p gaming without any 
> central servers at all (http://www.ieee-infocom.org/2004/Papers/03_2.PDF) ?

No. Bye, have fun,

Bernhard.

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

I'm sorry to ask the same question again, but I would like to know if there 
was already some 'brainstorming' about how to implement multiplayer 
networking ? (Yes, I had a look at the 'devel' archive).

Do the main developers like to use an already existing networking library, 
i.e. 'libgrapple' (http://opensource.linuxgamepublishing.com/grapple.php) ?

Or do they plan to implement a feature like i.e. p2p gaming without any 
central servers at all (http://www.ieee-infocom.org/2004/Papers/03_2.PDF) ?

Unfortunately, the last postings on the 'devel' mailinglist about this matter 
is over a year old... Are there some guidelines for people who like to 
volunteer ?

BTW: TORCS is a quite impressing work. Unbelievable, that it's OSS and 
downloadable for free... ;)

Donnar