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

This should work with older octave versions....


-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

I made a small tool in octave that plots graphs for cars, including
engine Tq/power plots, torque output per gear, by rpm and by Km, wind
resistance plot, acceleration curves and calculation of optimal
differential and even gear ratios for a given wing setting... the top
of the file includes a lot of examples of cars, though that could be
removed.

I thought it'd be useful, but I didn't know where to put it on the CVS
tree...


-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

On Wed, 15 Sep 2004 19:30:48 +0200 Eric Espie =
<eri...@fr...> wrote:

> I have set the binary flag on all the binary files of CVS
> so could you check that it's ok now (for that problem at =
least) ?

I see that many files now have the binary flag, but I suppose =
that the
following files should be binary too:

- torcs/data/cars/models/cg-nascar-rwd/.xvpics/cg-nascar-rwd.rgb
- torcs/data/cars/models/mclaren-f1/mclaren-f1.rgb.ok (?)
- torcs/data/cars/models/torcs/*.rgb
- torcs/data/data/fonts/*.glf
- torcs/data/data/sound/corolla_l.wav
- torcs/data/data/sound/locked.wav
- torcs/src/drivers/berniw/10/buggy.rgb
- torcs/src/drivers/*/*/*.rgb
- torcs/src/drivers/*/*/*.xcf
- torcs/src/drivers/*/*.rgb
- torcs/src/drivers/*/*.xcf
- torcs/src/modules/simu/simuv2/SOLID-2.0/solid2.dvi

Thanks.

--=20
Xavier Nodet
"They that can give up essential liberty to obtain a little =
temporary
safety deserve neither liberty nor safety." - Benjamin =
Franklin, 1759.

Hi,
When I try to execute the torcs from my bin directory I get this message.
 
 relocation error: /home/lib/torcs/lib/libconfscreens.so: undefined symbol: _Z18GfctrlGetNameByRefii
 
I am able to compile torcs without any errors.
I just added some code in the screen.cpp file which creates another window.
 
I will appreciate any help.
 
Thanks!

 

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

On Wed, 15 Sep 2004 19:37:52 +0200 Eric Espie =
<eri...@fr...> wrote:

> TORCS is not multithreaded, but there are some design issues =
about
> sharing static variables between DLLs. May be the PLIB static=
 lib is
> linked several times by different DLLs, that could be the =
problem

I just re-installed MSVC 6, and recompiled everything using =
unmodified
settings. But the 'client' module did not link correctly: it =
was missing
the ssgInit() function, called from TorcsEntry() in entry.cpp.

So I first tried to add the ssg.lib to the link step for =
'client', but
of course this leads to PLIB being linked twice, in two =
different DLLs
(client and wtorcs).

So I reversed those changes, and moved the ssgInit call from =
TorcsEntry
to main() (in wtorcs), just before calling TorcsEntry. This =
gave good
results: PLIB is linked just once, from wtorcs, and all the =
tracks I
tried so far load correctly.

--=20
Xavier Nodet
"They that can give up essential liberty to obtain a little =
temporary
safety deserve neither liberty nor safety." - Benjamin =
Franklin, 1759.

On Sep 16, 2004, at 9:20 AM, SBBG wrote:

> Hello,
>     I just visited TORCS site, and I think it's really impressive!! :P
>     I got a PowerPC machine running Mac OS X, I would like to try to
> compile it for Mac OS X, I'm eager to know if anyone ever tried this
> before and if any kind of hinders are met. Any reminds or suggestions
> are welcome.
>     I've just edit few lines of configure script in order to make it
> find the neccessary headers. However, I still got stuck at libm in
> configure process.
>

I had TORCS 1.2.1 and all the robots running at the end of last year. 
I've been busy and working on other projects, so I haven't tried the 
current source base. It wasn't too hard to get working however. Maybe 
I'll try to update my sources this week and submit the changes this 
time.

Kevin

Hi SBBG

Ok, make and gcc look ok then... can you try to compile the attached 
code snippet with "c++ -Wall gaga.cpp" to look what happens?

bye, Bernhard.


-- 

visit my homepage http://www.berniw.org
coming soon: The TORCS Racing Board, http://www.berniw.org/trb

On Wed, 15 Sep 2004 19:30:48 +0200 Eric Espie =
<eri...@fr...> wrote:

> I have set the binary flag on all the binary files of CVS
> so could you check that it's ok now (for that problem at =
least) ?

I'll check when Sourceforge's CVS servers are back... Thanks.

--=20
Xavier Nodet
"They that can give up essential liberty to obtain a little =
temporary
safety deserve neither liberty nor safety." - Benjamin =
Franklin, 1759.

Hi SBBG

> here is the error message , almost 1000 lines...
> I'm really sorry that I'm not familiar with C++, but if anyone could
> give any suggestion. I would be really glad to try it out. And I would
> also do any tweaking I can here.

Ok. The good news is: I think the problem should be trivial to fix, 
because it is actually the same problem repeated. Now the bad news: I 
have no idea why it happens. In the include files is a guard to avoid 
multiple includes, so this cannot be the problem. Look at this, that is 
actually weird on a first glimpse:
...
ulEndianSwap(short unsigned int*)' conflicts with
...
ulEndianSwap(unsigned int*)' here
...

Now I have some questions:
- What toolchain do you use to compile the stuff (gcc?)? I ask because 
it could be that "short unsigned int" is the same like "unsigned int" 
for your compiler.
- Do you use really a c++ compiler, or is the c++ just a link to a c 
compiler?

What do the following comands say:
cpp --version
c++ --version
make -v

bye, Bernhard.


-- 

visit my homepage http://www.berniw.org
coming soon: The TORCS Racing Board, http://www.berniw.org/trb

http://mail.cs.nchu.edu.tw/~s9256033/makelog.txt

here is the error message , almost 1000 lines...
I'm really sorry that I'm not familiar with C++, but if anyone could
give any suggestion. I would be really glad to try it out. And I would
also do any tweaking I can here.

Thanks again.

Hi SBBG

>   So does anyone can share some experience if someone ever built plib
> on Mac OS X successfully? Or can I just disable the joystick part in
> control config?
>    Thanks!!

No. But what does it complain about, please post some output, perhaps we 
can help you...

bye, Bernhard.

-- 

visit my homepage http://www.berniw.org
coming soon: The TORCS Racing Board, http://www.berniw.org/trb

TORCS does not use the JS part of PLIB. I copied (I know that's bad)
the js.h from PLIB into TORCS.

Eric.

SBBG wrote:
> Hello,
>    I just fed the torcs source the libglut from Fink, which is built
> by debian's apt. And the configure is passed. However, I still can't
> find any way to build a proper plib for Mac OS X. The JS part of plib
> just can't be compiled and caused that the controlconfig.cpp can't be
> compiled either.
>   So does anyone can share some experience if someone ever built plib
> on Mac OS X successfully? Or can I just disable the joystick part in
> control config?
>    Thanks!!
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
> Project Admins to receive an Apple iPod Mini FREE for your judgement on
> who ports your project to Linux PPC the best. Sponsored by IBM.
> Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php
> _______________________________________________
> Torcs-devel mailing list
> Tor...@li...
> https://lists.sourceforge.net/lists/listinfo/torcs-devel
> 
> 


-- 
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=
               TORCS - http://torcs.org
             The Open Racing Car Simulator
         AKA The Other Release Coming Soon (Skin'r)
            How soon is soon ? (RaceBlizter)
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=

Hello,
   I just fed the torcs source the libglut from Fink, which is built
by debian's apt. And the configure is passed. However, I still can't
find any way to build a proper plib for Mac OS X. The JS part of plib
just can't be compiled and caused that the controlconfig.cpp can't be
compiled either.
  So does anyone can share some experience if someone ever built plib
on Mac OS X successfully? Or can I just disable the joystick part in
control config?
   Thanks!!

SBBG wrote:
> Greeting, 
>    Thanks for Eric& Christos's help, after the ieee-fp option is
> removed, the libm part is passed.  And somehow I found that the
> configure script  need a CPPFLAGS="-I/usr/X11R6/include" to find
> glx.h.
>    Here's the new problem (I'm really sorry that I'm not that talented...):
> configure script can't find libglut. those guys using GLUT in Mac OS X
> usually compile with argument to gcc like "-framework GLUT" instead of
> "-lglut"  . And GLUT is NOT located in /usr/X11R6/include and
> /usr/X11R6/lib. It's placed in queer path
> "/Developer/SDKs/MacOSX10.2.7.sdk/System/Library/Frameworks/
> GLUT.framework/Versions/A" ( But the confiigure script is still able
> to find glut.h after change <GL/glut.h> to <GLUT/glut.h> )
>    I tried to replace all "-lglut" in configure script  by "-framework
> GLUT",but it seems not working.  So does anyone have any suggestion?
> Thanks.
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
> Project Admins to receive an Apple iPod Mini FREE for your judgement on
> who ports your project to Linux PPC the best. Sponsored by IBM.
> Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php
> _______________________________________________
> Torcs-devel mailing list
> Tor...@li...
> https://lists.sourceforge.net/lists/listinfo/torcs-devel
> 
> 
maybe you can set

LDFLAGS=-L/Developer/SDKs/MacOSX10.2.7.sdk/System/Library/Frameworks/GLUT.framework/Versions/A

and then run configure, I don't know if this works on Mac OS X.

Eric.
-- 
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=
               TORCS - http://torcs.org
             The Open Racing Car Simulator
         AKA The Other Release Coming Soon (Skin'r)
            How soon is soon ? (RaceBlizter)
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=

Greeting, 
   Thanks for Eric& Christos's help, after the ieee-fp option is
removed, the libm part is passed.  And somehow I found that the
configure script  need a CPPFLAGS="-I/usr/X11R6/include" to find
glx.h.
   Here's the new problem (I'm really sorry that I'm not that talented...):
configure script can't find libglut. those guys using GLUT in Mac OS X
usually compile with argument to gcc like "-framework GLUT" instead of
"-lglut"  . And GLUT is NOT located in /usr/X11R6/include and
/usr/X11R6/lib. It's placed in queer path
"/Developer/SDKs/MacOSX10.2.7.sdk/System/Library/Frameworks/
GLUT.framework/Versions/A" ( But the confiigure script is still able
to find glut.h after change <GL/glut.h> to <GLUT/glut.h> )
   I tried to replace all "-lglut" in configure script  by "-framework
GLUT",but it seems not working.  So does anyone have any suggestion?
Thanks.

 I am trying to figure out how to display one screen in each window , instead of
displaying mulitple screens in one window.The screen.cpp (under src/libs/tgfclient/) file creates a window when the full screen 
is not selected. 
The grmain.cpp (under src/modules/graphic/ssggraph) file has a grAdaptScreenSize() function that activates the number of screens, activates these screens in one window, 
I wanted to know how can one create these screens each one in a different window
The activate and desactivate are found in the grscreen.h that activates and deactivates the screen.
I would appreciate if you could help me.
-Mustafa


		
---------------------------------
Do you Yahoo!?
vote.yahoo.com - Register online to vote today!

Stefanie Tellex wrote:
> Hello,
> 
> How hard would it be to modify torcs to support objects on the track
> that the car can hit?  

not so hard I think.

> 
> I am thinking that you somehow specify them in the track XML file, and
> then add them to the SOLID simulation just like cars.  Collisions with
> them could be handled in the same way that car collisions are handled.  

The current design tries to separate the description (in XML) for the simulation
and the 3D representation for the graphic module.
All the data used in the simu come from the XML description.
I understand that describing the objects twice (XML + 3D) and maintaining
the coherency between the 2 representations is not easy, but the XML
description of the objects can be much more simple than the 3D one
(just a bounding box for example).

Using SOLID to manage the collision is the good way to do it.

If you want the robots to be aware of the objects on the track, then
the objects should be added to the "Situation" structure instead of the track
structure, because only the situation structure is given to the robots
during the race.

The biggest problem I see, is the lack of tools to create the tracks.
Currently, you have to describe the track first in XML, then generate
the 3D file from the XML file, then enhance the 3D file without moving
the track itself.

> 
> If I implemented this, would the patch be likely to be merged into the
> main source stream?

Of course, you're welcome.

> 
> Stefanie
> 

Eric.
-- 
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=
               TORCS - http://torcs.org
             The Open Racing Car Simulator
         AKA The Other Release Coming Soon (Skin'r)
            How soon is soon ? (RaceBlizter)
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=

I removed the ieee-fp option in the CVS version, so give it a try.

Eric.

Christos Dimitrakakis wrote:
>>configure:2268: checking for sin in -lm
>>configure:2290: c++ -o conftest -g -O2 -Wall -Wstrict-prototypes -O3 -mieee-fp \
>>-I/usr/X11R6/include  -L/usr/X11R6/lib conftest.C -lm   1>&5
>>cc1plus: error: invalid option `ieee-fp'
>>configure: failed program was:
>>#line 2276 "configure"
>>#include "confdefs.h"
>>============== End of message ====================
>>   If everyone somehow believe that porting to Mac OS X is possible,
>>would you please explain what's all these about? Thanks.
>>
> 
> 
> It seems that your c++ compiler does not support the ieee-fp option.
> 
> According to gcc info:
> `-mieee-fp'
> `-mno-ieee-fp'
>      Control whether or not the compiler uses IEEE floating point
>      comparisons.  These handle correctly the case where the result of
>      a comparison is unordered.
> 
> It is usually a good idea to specify this flag. However, this is a
> machine-dependent option: in some
> architectures this option might not make any sense. I understand that
> OSX uses the PPC architecture, which might not need this flag.
> 
> Check out the info for your compiler.
> 
> 


-- 
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=
               TORCS - http://torcs.org
             The Open Racing Car Simulator
         AKA The Other Release Coming Soon (Skin'r)
            How soon is soon ? (RaceBlizter)
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=

> configure:2268: checking for sin in -lm
> configure:2290: c++ -o conftest -g -O2 -Wall -Wstrict-prototypes -O3 -mieee-fp \
> -I/usr/X11R6/include  -L/usr/X11R6/lib conftest.C -lm   1>&5
> cc1plus: error: invalid option `ieee-fp'
> configure: failed program was:
> #line 2276 "configure"
> #include "confdefs.h"
> ============== End of message ====================
>    If everyone somehow believe that porting to Mac OS X is possible,
> would you please explain what's all these about? Thanks.
>

It seems that your c++ compiler does not support the ieee-fp option.

According to gcc info:
`-mieee-fp'
`-mno-ieee-fp'
     Control whether or not the compiler uses IEEE floating point
     comparisons.  These handle correctly the case where the result of
     a comparison is unordered.

It is usually a good idea to specify this flag. However, this is a
machine-dependent option: in some
architectures this option might not make any sense. I understand that
OSX uses the PPC architecture, which might not need this flag.

Check out the info for your compiler.


-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

Hello,
    I just visited TORCS site, and I think it's really impressive!! :P
    I got a PowerPC machine running Mac OS X, I would like to try to
compile it for Mac OS X, I'm eager to know if anyone ever tried this
before and if any kind of hinders are met. Any reminds or suggestions
are welcome.
    I've just edit few lines of configure script in order to make it
find the neccessary headers. However, I still got stuck at libm in
configure process.

===============message in config.log===================
configure:2268: checking for sin in -lm
configure:2290: c++ -o conftest -g -O2 -Wall -Wstrict-prototypes -O3 -mieee-fp \
-I/usr/X11R6/include  -L/usr/X11R6/lib conftest.C -lm   1>&5
cc1plus: error: invalid option `ieee-fp'
configure: failed program was:
#line 2276 "configure"
#include "confdefs.h"
/* Override any gcc2 internal prototype to avoid an error.  */
#ifdef __cplusplus
extern "C"
#endif
/* We use char because int might match the return type of a gcc2
    builtin and then its argument prototype would still apply.  */
char sin();
int main() {
sin()
; return 0; }

============== End of message ====================
   If everyone somehow believe that porting to Mac OS X is possible, 
would you please explain what's all these about? Thanks.

Hi there Stefanie

> How hard would it be to modify torcs to support objects on the track
> that the car can hit?
>
Not so hard, we have to add them to the simu module and put interfaces
between that and the graphics engine.

> I am thinking that you somehow specify them in the track XML file, and
> then add them to the SOLID simulation just like cars.  Collisions with
> them could be handled in the same way that car collisions are handled.
>

Unfortunately, SOLID is not a simulation - I have to implement
everything myself.

> If I implemented this, would the patch be likely to be merged into the
> main source stream?
>

Hm... currently I am re-working simuv3, the sequel to the current sim
engine, which extends the simulation from 2d+ to 3d. Basically
the only thing missing from that is proper 3d collisions.
Unfortunately I have not had a lot of time to work on it during the
summer, but I hope to continue after I some thesis-related deadlines
pass (should be a bit more relaxed next week)

So, currently the simuv3 code is in flux.. however I could stabilise
it soon. Any input you have is welcome.. I was planning to insert the
car wheels as a separate object in the code soon, because of a lot of
problems with the current implementation when the angle between the
road and the car is very high..

So, what you can do:

Currently I was thinking of implementing the
actual collision code as a simple spring-damper system. I am sure that
there are much better techniques, but this has the advantage you don't
need a closed form solution afai can understand. If you have any
suggestions, or even better, some code that we can integrate, it will
be cool.

You could try and work on the xml side of things instead - probably it
will be easier.


Ciao
-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

Hello,

How hard would it be to modify torcs to support objects on the track
that the car can hit?  

I am thinking that you somehow specify them in the track XML file, and
then add them to the SOLID simulation just like cars.  Collisions with
them could be handled in the same way that car collisions are handled.  

If I implemented this, would the patch be likely to be merged into the
main source stream?

Stefanie

Xavier Nodet wrote:
> Hi all,
> 
> Using the latest CVS version of Tracs on Windows XP-SP2 with MSVC 7.1, I
> have the following problem: almost all tracks will crash Tracs when
> starting a race. The only road tracks that do work for me are 'CG
> Track' 2 and 3, and 'E Track' 2 and 6.
> 
> The last traces from the program are: 
> 
> ++++++++++++ Track ++++++++++++
> name     = Alpine 1
> author   = E. Espie
> filename = tracks/road/alpine-1/alpine-1.xml
> nseg     = 1179
> version  = 4
> length   = 6355.651367
> width    = 12.000000
> XSize    = 949.537292
> YSize    = 1064.563477
> ZSize    = 150.001266
> Pits     = present on track side
> Loading Track 3D Description...
> GfParmReadFile: Openning "tracks/road/alpine-1/alpine-1.xml" (0B7EDC7C)
> Loading tracks/road/alpine-1/background.png
> Loading Environment Mapping Image env.png
> Loading data/textures/env.png
> Loading data/textures/env.png
> Loading data/textures/envshadow.png
> Loading tracks/road/alpine-1/shadow2.rgb
> LoadAC3D loading alpine-1.ac
> WARNING: ssgLoad: Unrecognised file type '.ac'
> 
> and then it crashes.
> 
> To be able to debug this problem, I started using plib-1.8.3 instead of
> the one that is included into Tracs' CVS. This did not change anything,
> and the problem is the same between the two versions. It turns out that
> the problem is that the ssgLoad function does not find any suitable
> loader for the '.ac' file format.  
> 
> The situation is that format informations are stored in some static
> variables ('formats' and 'num_formats', in ssgLoad.cpp), but when Tracs
> tries to load the track file, the 'num_format' variable is equal to 0,
> although I verified that many formats (including '.ac') were included at
> the beginning of the execution.
> 
> Would it be the case that those two actions ('registering formats at the
> beginning', and 'loading the track') are done in separate threads that
> would not share the same copy of the static variables?
> 
> Thanks for any idea.
> 
> 
> Note: the reason why a few tracks are still working is the following.
> Just before loading those tracks, the '.acc' format is registered (and
> thus is the only format registered for this thread).  As those tracks
> actually use this format, this works fine...
> 

TORCS is not multithreaded, but there are some design issues
about sharing static variables between DLLs.
May be the PLIB static lib is linked several times by different
DLLs, that could be the problem, if only one has run the ssgInit()
function, the instances of PLIB in other DLLs are not initialized.
Didn't you changed some project setting in order to have TORCS
compiled in your environment ?

Eric.
-- 
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=
               TORCS - http://torcs.org
             The Open Racing Car Simulator
         AKA The Other Release Coming Soon (Skin'r)
            How soon is soon ? (RaceBlizter)
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=

Xavier Nodet wrote:
> On Tue, 14 Sep 2004 16:19:47 +0200 (Romance Daylight Time) Xavier Nodet <xav...@fr...> wrote:
> 
> 
>>  Could this problem be related to the same kind of errors as for the
>>  PNG files?
> 
> 
> It turns out that many other binary files (glf, xcf, rgb, wav) in the
> data were checked-in as text instead of binary. Re-getting them properly
> fixed my problem.
> 
> It would of course still be nice to correct this problem, so as to avoid
> it in the future.
> 
> Thanks.
> 

I have set the binary flag on all the binary files of CVS
so could you check that it's ok now (for that problem at least) ?

Eric.
-- 
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=
               TORCS - http://torcs.org
             The Open Racing Car Simulator
         AKA The Other Release Coming Soon (Skin'r)
            How soon is soon ? (RaceBlizter)
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=

On Wed, 15 Sep 2004 09:24:58 +0200 (Romance Daylight Time) =
Xavier Nodet <xav...@fr...> wrote:

> Using the latest CVS version of Tracs [...]

Oops. I must have missed some cups of coffee, for not even =
remembering
the name of the project!  Hope you were not (too) offended...

--=20
Xavier Nodet
"They that can give up essential liberty to obtain a little =
temporary
safety deserve neither liberty nor safety." - Benjamin =
Franklin, 1759.