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

Compilation under vc6 is very smooth, but same job with vc8 is tough enough.
Finally compilation and running is OK with vc8(2005), and my notes is
following, hope to be helpful to others may concern:

1. modify winnt.h line 222
#define POINTER_64 __ptr64
typedef void *PVOID;
typedef void * POINTER_64 PVOID64;

2. In project client and track, use code generation option
"multithread", don't use "multithread dll"

3. add link option "/force:multiple" in case symbol redefinition warning

4. when compiling client project, there will be error LNK2001:
unresolved external symbol _errno
create a new cpp to sovle it
extern "C"
{
int __mb_cur_max;
unsigned short* _pctype;
int errno;
}

5. fix errors caused by type-casting check and windows API change
(like sqrt(..) )


Cheers,
Jackie Ju

Hello Everyone,

 I am a newbie to TORCS. I am very interested in TORCS game and trying
make a new track for my city.
 I'v read the pdf of "Buiilld yourr ttrrack fforr TORCS in 20
miinuttes" and other very useful documents. But still some problems
are not sovled.

1. Can I create two crossed straight ways, one is normal ground
streeet and the other is hight way? I tried it, but the trackgen.exe
crashed.
2. Can I add NPC in the way? e.g. girl, man, dog. police car...
3. Is it possible to place river, tower, stadium, and things like that
nearby the tracke ?
4. The start line and end line will not be same in my track, is it posssible?

Any comments will be appreciated, thanks !

Best Regards,
Jackie Ju

Wolf-Dieter Beelitz a écrit :
> Hello,
>
> you have to help the compiler with manual casts:
>
> const char * constC;
> char* c = (char *) (constC);
>
> float f = sqrt((float)(value))
>
> and so on.
>
> The postbuild errors are from copying the result to the place where it
> should stay for later use. 
> The path you uses doesn't contain the targets path. So the result is not
> found later for linking.
>
> Cheers
>
> wdbee 
>
>
>
> -------------------------------------------------------------------------
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services for
> just about anything Open Source.
> http://sourceforge.net/services/buy/index.php
> _______________________________________________
> Torcs-devel mailing list
> Tor...@li...
> https://lists.sourceforge.net/lists/listinfo/torcs-devel
>
>
>   
Using 1.3.1 CVS branch helps a lot, but it is not yet polished (to say 
the less) so be patient and read the REAME.vc2005 file.

Eric.

Hello,

you have to help the compiler with manual casts:

const char * constC;
char* c = (char *) (constC);

float f = sqrt((float)(value))

and so on.

The postbuild errors are from copying the result to the place where it
should stay for later use. 
The path you uses doesn't contain the targets path. So the result is not
found later for linking.

Cheers

wdbee 

Hi,
   if torcs developers are interested, i can possibly contribute some 
code for a network module based on ACE framework 
(http://www.cse.wustl.edu/~schmidt/ACE.html). Later other functionality 
like a central torcs game server and client to server protocol must be 
implemented.

If yes, i have to find out what should i take care of as developer to 
code the module according to torcs coding guide lines...? To accelerate 
my work also some hints about code integratoin in torcs, requirements of 
this network module.. If my team approves it, only then i will start....

A question of mine beeing newbie about torcs is: what is the best 
location in source code to extract data from game for logging / export / 
analysis? Method SimUpdate in simu.cpp or SimCarUpdate2 in car.cpp? What 
does the telemetry module do and any doc about it?


best regards,
ali

Hi,
   i have compiled torcs 1.3 successfully on Linux. Compiling the same 
sources with Visual Studio on M$ Windows has following errors about some 
conversions, ambiguous calls and some missing libs..

Error	1	error C2440: '=' : cannot convert from 'const char *' to 
'char *' 
d:\dateien\development\externalprojects\torcs-1.3.0\src\modules\graphic\ssggraph\grtexture.cpp 
51
Error	2	error C2668: 'sqrt' : ambiguous call to overloaded function 
d:\dateien\development\externalprojects\torcs-1.3.0\src\modules\graphic\ssggraph\grcam.cpp 
1752
Error	3	error C2668: 'sqrt' : ambiguous call to overloaded function 
d:\dateien\development\externalprojects\torcs-1.3.0\src\modules\graphic\ssggraph\grcam.cpp 
1775
Error	4	error C2668: 'sqrt' : ambiguous call to overloaded function 
d:\dateien\development\externalprojects\torcs-1.3.0\src\modules\graphic\ssggraph\grcam.cpp 
1798
Error	5	error C2668: 'sqrt' : ambiguous call to overloaded function 
d:\dateien\development\externalprojects\torcs-1.3.0\src\modules\graphic\ssggraph\grcam.cpp 
1821
Error	6	error C2668: 'fabs' : ambiguous call to overloaded function 
d:\dateien\development\externalprojects\torcs-1.3.0\src\libs\raceengineclient\raceengine.cpp 
60
Error	13	error PRJ0019: A tool returned an error code from "Performing 
Post-Build Event..."	track
Error	14	error C2440: 'initializing' : cannot convert from 'const char 
*' to 'char *' 
d:\dateien\development\externalprojects\torcs-1.3.0\src\tools\trackgen\objects.cpp 
276
Error	15	error C2440: '=' : cannot convert from 'const char *' to 
'char *' 
d:\dateien\development\externalprojects\torcs-1.3.0\src\tools\trackgen\objects.cpp 
279
Error	19	error PRJ0019: A tool returned an error code from "Performing 
Post-Build Event..."	simuv2
Error	22	error PRJ0019: A tool returned an error code from "Performing 
Post-Build Event..."	tanhoj
Error	25	error PRJ0019: A tool returned an error code from "Performing 
Post-Build Event..."	tita
Error	51	error PRJ0019: A tool returned an error code from "Performing 
Post-Build Event..."	lliaw
Error	73	error C2666: 'pow' : 6 overloads have similar conversions 
d:\dateien\development\externalprojects\torcs-1.3.0\src\libs\learning\mathfunctions.cpp 
206
Error	99	error PRJ0019: A tool returned an error code from "Performing 
Post-Build Event..."	inferno2
Error	102	error PRJ0019: A tool returned an error code from "Performing 
Post-Build Event..."	inferno
Error	103	fatal error LNK1104: cannot open file 'client.lib'	human
Error	106	error PRJ0019: A tool returned an error code from "Performing 
Post-Build Event..."	damned
Error	109	error PRJ0019: A tool returned an error code from "Performing 
Post-Build Event..."	cylos1
Error	110	fatal error LNK1104: cannot open file 'client.lib'	wtorcs
Error	113	error PRJ0019: A tool returned an error code from "Performing 
Post-Build Event..."	bt
Error	114	error C2668: 'sqrt' : ambiguous call to overloaded function 
d:\dateien\development\externalprojects\torcs-1.3.0\src\drivers\berniw3\pathfinder.cpp 
667
Error	115	error C2668: 'sqrt' : ambiguous call to overloaded function 
d:\dateien\development\externalprojects\torcs-1.3.0\src\drivers\berniw2\pathfinder.cpp 
667
Error	116	error C2668: 'sqrt' : ambiguous call to overloaded function 
d:\dateien\development\externalprojects\torcs-1.3.0\src\drivers\berniw\pathfinder.cpp 
667

Any hints would be very helpfull..

Best Regards
Ali

Hi Daniele

>> I know Valgrind and already used it for debugging some of my personal
>> projects.
>> I would be glad to help in debugging the game.
>> I did some run with valgrind but I  was not able to resolve symbols... I
>> suppose this is due to dynamic modules.
> 
> If you compile torcs with the --enable-debug configure option, then the 
> symbols should be available.
> 
>> Do you know how to solve this problems (or there is any documentation
>> about that...).

To be able to gather info about the modules I just commented out dlclose, look 
at linuxspec.cpp, search for "valgrind". Of course you cannot sport this way 
problems which happen on unloading the modules, but you can track their 
behaviour. Maybe this is not necessary anymore (with newer valgind versions), I 
do not know.

Hope this helps.

Bernhard

Hi Daniele (and others),

Op Thursday 19 June 2008 12:23:18 schreef Daniele Loiacono:
> Hi Bernhard,
>
> thank you for the reply.
>
> I know Valgrind and already used it for debugging some of my personal
> projects.
> I would be glad to help in debugging the game.
> I did some run with valgrind but I  was not able to resolve symbols... I
> suppose this is due to dynamic modules.

If you compile torcs with the --enable-debug configure option, then the 
symbols should be available.

> Do you know how to solve this problems (or there is any documentation
> about that...).
>
> Thank you very much.
> Daniele

Regards,

Mart

Hi Bernhard,

thank you for the reply.

I know Valgrind and already used it for debugging some of my personal
projects.
I would be glad to help in debugging the game.
I did some run with valgrind but I  was not able to resolve symbols... I
suppose this is due to dynamic modules.

Do you know how to solve this problems (or there is any documentation
about that...).

Thank you very much.
Daniele

Bernhard Wymann wrote:
> Hi Daniele
>
>   
>> my name is Daniele Loiacono and I am a Postdoc at Politecnico di Milano
>> (Italy).
>> In the last months I am working on applying Machine Learning techniques
>> to TORCS as I think it is a very nice testbed for many techniques.
>>
>> I discovered a memory leak in TORCS when the race is restarted. I am
>> using TORCS 1.3.0 on Linux (Open Suse) and each time the race is
>> restarted the memory used increase approximately of 500KBytes. As I
>> extended TORCS with an automatic restart procedure to apply Machine
>> Learning to it, I quickly get a memory overflow.
>>
>> I tried to debug TORCS to locate the leak and it seems to be related
>> with the loading of Track information, but I was not able to understand
>> where exactly the problem is?
>>     
>
> Then you have to dig deeper. I hunted down loads of leaks already, but a few I 
> could not fix back then (at least not quickly, and under "normal use" it was 
> acceptable to leave those). IIRC the following are open:
> - plib is not intended for dynamic linking, so this leaves some garbage everytime.
> - TORCS isself has some open points, I think there was a problem with releasing 
> the cameras as example.
> - The environment (X11/OpenGL) did leak some resources as well back then.
>
> If you do not already use it, install valgrind. It makes execution about 50 
> times slower, but is great to find leaks etc. I you find some fixes or if you 
> can identify to problem, let us know.
>
> Hope this helps
>
> Bernhard
>
> -------------------------------------------------------------------------
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services for
> just about anything Open Source.
> http://sourceforge.net/services/buy/index.php
> _______________________________________________
> Torcs-devel mailing list
> Tor...@li...
> https://lists.sourceforge.net/lists/listinfo/torcs-devel
>   


-- 
Daniele Loiacono, PhD Student
AI & Robotics Lab
Politecnico di Milano (Italy)
http://home.dei.polimi.it/loiacono
Phone  +39 02 2399 3540

Wed, 18 Jun 2008 00:33:29 +0200, Bernhard Wymann <be...@bl...> :

> I am working on the finishing touches of release 1.3.1, you can check
> out the branch r1-3-1 to get a preview and to test it. To the
> developers: please do not commit to this branch!
In 1.3.0, there's a hardcoded installation suffix of %s/games/. To
package Torcs for Pardus Linux (http://www.pardus.org.tr/eng/) I had to
manually sed out the /games/ suffix in more than a dozen files [1].

Can this possibly be a variable that can be set in the configure
script in the new version?

[1]
http://svn.pardus.org.tr/pardus/devel/applications/games/torcs/actions.py

                   Doruk

--
FISEK INSTITUTE - http://www.fisek.org.tr

Hi Andrew (and others),

Op Tuesday 17 June 2008 07:41:36 schreef Andrew Sumner:
> I've had a bit of difficulty testing this under Windows.  The DLL compiled
> by the "Release" build in Visual Studio 2005 is some 508kb in size,
> compared with the 164kb version distributed with 1.3.0.
>
> When I run a race of, say, 6 cars using the new DLL performance is
> painfully
>
> jerky and slow - I have a fast new laptop & TORCS is usually very smooth.
>
> I'm not sure what the problem is - could it be changes in the code that are
> causing the slowdown (in which case it should be noticeable on linux too)
> or is it something to do with VS 2005?

I didn't noticed any slowdown in Linux. I also checked the size of the simuv2 
shared object without debug symbols (in Linux). It is 182 KB here (and did 
grow with only a couple of bytes after updating CVS).

> cheers
> Andrew

Regards,

Mart

Hi all

I am working on the finishing touches of release 1.3.1, you can check out the 
branch r1-3-1 to get a preview and to test it. To the developers: please do not 
commit to this branch!

Missing bits (I am currently working on this):
- Nice textures for trb1 cars
- Windows bat files
- Installer
- Packaging
- Cleanup

So I guess the 1.3.1-test1 will come in a few weeks.

Enjoy

Bernhard

I've had a bit of difficulty testing this under Windows.  The DLL compiled
by the "Release" build in Visual Studio 2005 is some 508kb in size, compared
with the 164kb version distributed with 1.3.0.  

When I run a race of, say, 6 cars using the new DLL performance is painfully

jerky and slow - I have a fast new laptop & TORCS is usually very smooth.

I'm not sure what the problem is - could it be changes in the code that are
causing the slowdown (in which case it should be noticeable on linux too) or
is it something to do with VS 2005?

cheers
Andrew


-----Original Message-----
From: Christos Dimitrakakis [mailto:ole...@fa...]
Sent: Thursday, 5 June 2008 5:20 PM
To: tor...@li...
Subject: [Torcs-devel] CVS simuv2 transmission/inertia/differential
fixes


Committed the previously discussed patch fixing the inertia of the whole
drivetrain and the differentials...

Hi Daniele

> my name is Daniele Loiacono and I am a Postdoc at Politecnico di Milano
> (Italy).
> In the last months I am working on applying Machine Learning techniques
> to TORCS as I think it is a very nice testbed for many techniques.
> 
> I discovered a memory leak in TORCS when the race is restarted. I am
> using TORCS 1.3.0 on Linux (Open Suse) and each time the race is
> restarted the memory used increase approximately of 500KBytes. As I
> extended TORCS with an automatic restart procedure to apply Machine
> Learning to it, I quickly get a memory overflow.
> 
> I tried to debug TORCS to locate the leak and it seems to be related
> with the loading of Track information, but I was not able to understand
> where exactly the problem is?

Then you have to dig deeper. I hunted down loads of leaks already, but a few I 
could not fix back then (at least not quickly, and under "normal use" it was 
acceptable to leave those). IIRC the following are open:
- plib is not intended for dynamic linking, so this leaves some garbage everytime.
- TORCS isself has some open points, I think there was a problem with releasing 
the cameras as example.
- The environment (X11/OpenGL) did leak some resources as well back then.

If you do not already use it, install valgrind. It makes execution about 50 
times slower, but is great to find leaks etc. I you find some fixes or if you 
can identify to problem, let us know.

Hope this helps

Bernhard

Hi all,

my name is Daniele Loiacono and I am a Postdoc at Politecnico di Milano
(Italy).
In the last months I am working on applying Machine Learning techniques
to TORCS as I think it is a very nice testbed for many techniques.

I discovered a memory leak in TORCS when the race is restarted. I am
using TORCS 1.3.0 on Linux (Open Suse) and each time the race is
restarted the memory used increase approximately of 500KBytes. As I
extended TORCS with an automatic restart procedure to apply Machine
Learning to it, I quickly get a memory overflow.

I tried to debug TORCS to locate the leak and it seems to be related
with the loading of Track information, but I was not able to understand
where exactly the problem is?

Does someone notice the same problem or find any solution?

Best regards,
Daniele Loiacono

-- 
Daniele Loiacono, Postdoc
AI & Robotics Lab
Politecnico di Milano (Italy)
http://home.dei.polimi.it/loiacono
Phone  +39 02 2399 3540

Hi,

I am doing some changes in the source code torcs for monitoring some actions
of the driver, in specific actions that make more economic direction and
fuel economy.
To conduct the tests I need a runway that has crosses to see the reaction
through these situations.
Currently only need a runway with crosses.
Could you tell me where can I find a new version of trackeditor changed?

Thank you

Hello,
The internal representation of tracks does not allow any crossroads, 
since it's not necessary for racing tracks, so there's no tool that 
can allow you to create them.

But, we at the Robotics Institute, at the University of Valencia 
(Spain) are working on an urban simulator based on Torcs, so we 
modified the internal structure of tracks, as well as the track editor 
in order to allow the user to design tracks with crossroads and dead 
ends. We also modified the trackgen tool for generation of the 3D 
model. So we can now create urban tracks (with crossroads and dead 
ends) in a visual way with our modified trackeditor and trackgen, in a 
special track format.
We are currently working on modifying the simuv collision detection 
and other stuff to make torcs (and robots) understand what a crossroad 
mean, so you 
can't actually play with a 'special' track, by the moment, you can 
generate the XML and AC3D files with our tools.

Let me know if you are interested in source or compiled tools.

Greetings


> Message: 1
> Date: Mon, 9 Jun 2008 11:46:09 -0300
> From: "Tales Bogoni" <tal...@gm...>
> Subject: [Torcs-devel] tracks
> To: tor...@li...
> Message-ID:
> 	<177...@ma...>
> Content-Type: text/plain; charset="iso-8859-1"
> 
> Hallo,
> 
> I need to create a track that has crosses, but with the trackeditor 
not
> getting.
> Is there any other tool to create avenues for TORCS?
> 
> Thank you
> -------------- next part --------------
> An HTML attachment was scrubbed...
> 
> ------------------------------
> 

--
/*********************************
* Miguel Angel Expósito Sánchez *
*********************************/

Hi, Christos, and all.

> Committed the previously discussed patch fixing the inertia of the whole
> drivetrain and the differentials.
> 
> Main Visible Changes:
> 
> Inertia: When power-spinning, it takes longer for the wheels to spin _and_ to
> spin down.
 > Differential: The Free and Limited Slip differentials have changed
 > to be closer to reality.

After a few tries, I feel this really makes car behaviour more reallistic
as far as "oversteering at end of turns because of too heavy right foot"
is concerned ... thus making Simu V2 get closer to Simu V3 ...
(don't know which of the Inertia/Differential changes makes this real
  difference, may the two ones ...)

Cheers,

Jean-Philippe.

Hi;

My first post to this alias:

I would like to build a Torcs driving robot in C#. However, I'm just getting
up to speed on Torcs and have many other coding tasks ahead of me.

I was therefore wondering if there are people out there who are looking to
make money, have time, and could wrap the necessary Torcs APIs.

I realize that there are actually 2 levels:

   - the set of things the Torcs world provides the driving robot
   - the complete set of random and cool functions a driving robot written
   in C# could possibly want to do its job.

I am not really familiar with the difference between those two tasks but am
willing to work with someone to define this.

If you are interested, please tell me:
Your thoughts on this task
What do you do? How many hours per week could you work on this? (Working
part-time is totally fine if you can do an hour or 2 a day!) When can you
start?
Would you have time for further work after or are you "between jobs?"
What is your familiarity with Torcs and Mono/.Net?

Note that I have considered doing a Java API, but I like C# more, and once
we have a C# API, it is easy for anyone else to build a Java API. Also, Mono
has compilers for more platforms than Java:
x86<http://en.wikipedia.org/wiki/X86>,
SPARC <http://en.wikipedia.org/wiki/SPARC>,
PowerPC<http://en.wikipedia.org/wiki/PowerPC>,
ARM <http://en.wikipedia.org/wiki/ARM_architecture>,
S390<http://en.wikipedia.org/wiki/S390>(in 32-bit and 64-bit mode),
and
x86-64 <http://en.wikipedia.org/wiki/AMD64>,
IA64<http://en.wikipedia.org/wiki/IA64>and SPARC for 64-bit modes.

I only use Linux and it would need to work on Mono on Linux.

If you are interested please reply to me!

I will send follow-up status e-mails about this project to this alias if
people are interested. This will be free software, so maybe it could even
make it back into the official torcs codebase one day?

Thanks!

-Keith

Hallo,

I need to create a track that has crosses, but with the trackeditor not
getting.
Is there any other tool to create avenues for TORCS?

Thank you

Committed the previously discussed patch fixing the inertia of the whole
drivetrain and the differentials.

Main Visible Changes:

Inertia
-------
When power-spinning, it takes longer for the wheels to spin _and_ to
spin down.

Inertia is maintained when changing gears and using the clutch, so you
can make even under-powered cars spin in this way if you really want to.

NOTE: The increased inertia may require some robots to increase their
braking settings somewhat. So check your robots.


Differential
------------
The Free and Limited Slip differentials have changed to be closer to
reality.  The Free differential is 100% OK.  However, the Limited Slip
differential is only one of many possible choices.  (The old limited
slip differential was in fact a Locking differential, which I have now
moved to a new category).

Minor changes
-------------
There was a double 'rolling resistance' line in car.cpp - I removed one
of them.


PLEASE TEST!

Hi, Ziomek.

>  Im new to Torcs, and was wondering if there is documentation of the source?
>    All i need from Torcs, while im driving, is the relative position of all
>  cars near me and then write over a TCP socket.
>    Seems pretty simple but im unsure of how to start. Thanks!

First, if you are developping under some Unix, and have Doxygen installed,
you can generate the API doc like this :
- add the following line on top of the main Makefile :
   SRCDOC = 1
- make info

That way, you'll get the HTML doc into doc/manual/api (index.html).

Then, I've found some weeks ago a pre-compiled API doc for the 1.2.2 release
on the net, ... but can't remember where ... may be on Bernhardt web site,
but can't find it now ...

Cheers,

Jean-Philippe.

Hi,

 Im new to Torcs, and was wondering if there is documentation of the
 source?
   All i need from Torcs, while im driving, is the relative position of all
 cars near me and then write over a TCP socket.

   Seems pretty simple but im unsure of how to start. Thanks!

Ziomek