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

El dc 31 de 08 del 2005 a les 12:12 +0200, en/na Bernhard Wymann va
escriure:
> Hi Albert
> 
> Hmm, no idea... can you run "torcs -l"? What happens if you disable the sound or
> change the backend to plib (just because of this "loopback")?

torcs -l
                libracescreens.so
=> /usr/local/games/torcs/lib/lib/libracescreens.so (0xb7fdb000)
        librobottools.so
=> /usr/local/games/torcs/lib/lib/librobottools.so (0xb7fd7000)
        libclient.so => /usr/local/games/torcs/lib/lib/libclient.so
(0xb7cce000)
        libconfscreens.so
=> /usr/local/games/torcs/lib/lib/libconfscreens.so (0xb7cc0000)
        libtgf.so => /usr/local/games/torcs/lib/lib/libtgf.so
(0xb7cb5000)
        libtgfclient.so
=> /usr/local/games/torcs/lib/lib/libtgfclient.so (0xb7c9d000)
        libtxml.so => /usr/local/games/torcs/lib/lib/libtxml.so
(0xb7c84000)
        libraceengine.so
=> /usr/local/games/torcs/lib/lib/libraceengine.so (0xb7c71000)
        liblearning.so => /usr/local/games/torcs/lib/lib/liblearning.so
(0xb7c5d000)
        libglut.so.3 => /usr/local/games/torcs/lib/lib/libglut.so.3
(0xb7c26000)
        libGLU.so.1 => /usr/X11R6/lib/libGLU.so.1 (0xb7ba5000)
        libGL.so.1 => /usr/X11R6/lib/libGL.so.1 (0xb7b45000)
        libpng.so.3 => /usr/local/games/torcs/lib/lib/libpng.so.3
(0xb7b19000)
        libz.so.1 => /usr/lib/libz.so.1 (0xb7b08000)
        libdl.so.2 => /lib/tls/i686/cmov/libdl.so.2 (0xb7b04000)
        libopenal.so.0 => /usr/local/games/torcs/lib/lib/libopenal.so.0
(0xb7a7a000)
        libXrandr.so.2 => /usr/X11R6/lib/libXrandr.so.2 (0xb7a76000)
        libXrender.so.1 => /usr/lib/libXrender.so.1 (0xb7a6e000)
        libXmu.so.6 => /usr/X11R6/lib/libXmu.so.6 (0xb7a59000)
        libXi.so.6 => /usr/X11R6/lib/libXi.so.6 (0xb7a51000)
        libXt.so.6 => /usr/X11R6/lib/libXt.so.6 (0xb7a01000)
        libSM.so.6 => /usr/X11R6/lib/libSM.so.6 (0xb79f8000)
        libICE.so.6 => /usr/X11R6/lib/libICE.so.6 (0xb79e0000)
        libXext.so.6 => /usr/X11R6/lib/libXext.so.6 (0xb79d3000)
        libX11.so.6 => /usr/X11R6/lib/libX11.so.6 (0xb790e000)
        libstdc++.so.5 => /usr/lib/libstdc++.so.5 (0xb7854000)
        libm.so.6 => /lib/tls/i686/cmov/libm.so.6 (0xb7832000)
        libgcc_s.so.1 => /lib/libgcc_s.so.1 (0xb7827000)
        libc.so.6 => /lib/tls/i686/cmov/libc.so.6 (0xb76fa000)
        libpthread.so.0 => /lib/tls/i686/cmov/libpthread.so.0
(0xb76ea000)
        /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0xb7feb000)

Serendipitously, if I run:

torcs 1>/dev/null 2>/dev/null

then it runs ok (with sound disabled). So the error seems to be flooding
the console, and it takes a while to start.

> 
> If you find that the shipped openal version is the problem, you can try to point
> with the symbolic link in the lib dir to the libraries installed on you system,
> perhaps that helps (try OpenAL first).
> 
> Bye, Bernhard.
> 
> 
> > Dears,
> > 
> > I get this error when starting a race with torcs installed via
> > torcs-1.2.4-linux-glibc-2.3-i686.bz2.run:
> > 
> > Just before the graphics appear, it gets stucks in a black screen with
> > the "Ready" prompt, and the console reporting:
> > 
> > Visual Properties Report
> > ------------------------
> > Compatibility mode, properties unknown.
> > [_ae_loopback_array_elt,759] aa->offset = 57
> > [_ae_loopback_array_elt,759] aa->offset = 137
> > [_ae_loopback_array_elt,759] aa->offset = 57
> > [_ae_loopback_array_elt,759] aa->offset = 137
> > [_ae_loopback_array_elt,759] aa->offset = 57
> > [_ae_loopback_array_elt,759] aa->offset = 137
> > [_ae_loopback_array_elt,759] aa->offset = 57
> > [_ae_loopback_array_elt,759] aa->offset = 137
> > [...]
> > 
> > Any hint?
> > 
> > This is under Ubuntu Hoary,
> > 
> >     Albert.
> > 
> > 
> > 
> > -------------------------------------------------------
> > SF.Net email is Sponsored by the Better Software Conference & EXPO
> > September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
> > Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
> > Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
> > _______________________________________________
> > Torcs-users mailing list
> > Tor...@li...
> > https://lists.sourceforge.net/lists/listinfo/torcs-users
> > 
> 
> 

Hi Albert

Hmm, no idea... can you run "torcs -l"? What happens if you disable the sound or
change the backend to plib (just because of this "loopback")?

If you find that the shipped openal version is the problem, you can try to point
with the symbolic link in the lib dir to the libraries installed on you system,
perhaps that helps (try OpenAL first).

Bye, Bernhard.


> Dears,
> 
> I get this error when starting a race with torcs installed via
> torcs-1.2.4-linux-glibc-2.3-i686.bz2.run:
> 
> Just before the graphics appear, it gets stucks in a black screen with
> the "Ready" prompt, and the console reporting:
> 
> Visual Properties Report
> ------------------------
> Compatibility mode, properties unknown.
> [_ae_loopback_array_elt,759] aa->offset = 57
> [_ae_loopback_array_elt,759] aa->offset = 137
> [_ae_loopback_array_elt,759] aa->offset = 57
> [_ae_loopback_array_elt,759] aa->offset = 137
> [_ae_loopback_array_elt,759] aa->offset = 57
> [_ae_loopback_array_elt,759] aa->offset = 137
> [_ae_loopback_array_elt,759] aa->offset = 57
> [_ae_loopback_array_elt,759] aa->offset = 137
> [...]
> 
> Any hint?
> 
> This is under Ubuntu Hoary,
> 
>     Albert.
> 
> 
> 
> -------------------------------------------------------
> SF.Net email is Sponsored by the Better Software Conference & EXPO
> September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
> Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
> Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
> _______________________________________________
> Torcs-users mailing list
> Tor...@li...
> https://lists.sourceforge.net/lists/listinfo/torcs-users
> 


-- 

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

Dears,

I get this error when starting a race with torcs installed via
torcs-1.2.4-linux-glibc-2.3-i686.bz2.run:

Just before the graphics appear, it gets stucks in a black screen with
the "Ready" prompt, and the console reporting:

Visual Properties Report
------------------------
Compatibility mode, properties unknown.
[_ae_loopback_array_elt,759] aa->offset = 57
[_ae_loopback_array_elt,759] aa->offset = 137
[_ae_loopback_array_elt,759] aa->offset = 57
[_ae_loopback_array_elt,759] aa->offset = 137
[_ae_loopback_array_elt,759] aa->offset = 57
[_ae_loopback_array_elt,759] aa->offset = 137
[_ae_loopback_array_elt,759] aa->offset = 57
[_ae_loopback_array_elt,759] aa->offset = 137
[...]

Any hint?

This is under Ubuntu Hoary,

    Albert.

Hi Christos

> Hm.. it seems this was already there but commented out. Weird. Anyway, 

Yes, it was there from the very beginning of the CVS, weird indeed...

> I'll replace the simuv3 one with this one. I looked at it graphically 
> and it behaves well in the steer range [-1, 1]. Difference between the 
> two calculations is in the order of 2^-16, as expected.
> 
> Why is the arctan used in the first place? Maybe we can add some 
> documentation at the top of steer.cpp to say what we are trying to 
> simulate.

I think the intension is that the outer wheel angle fits the radius it drives 
(on tight turns the outer wheel drives on a significant larger radius), good to 
see with the F1 from top. Real cars are doing that as well. Perhaps one could 
simplify the whole expression/code snippet even more.

Bye, thank you for the analysis,

Bernhard.


-- 

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

>
> I have another attempt to fix it (simuv2), could you try/analyse it, does it 
> work for you?
>

Ah, yes, replacing arctan(x,y/z) with arctan(x*z,y) is a wise move. 
Great fix.

Hm.. it seems this was already there but commented out. Weird. 
Anyway, I'll replace the simuv3 one with this one. I looked at it 
graphically and it behaves well in the steer range [-1, 1]. Difference 
between the two calculations is in the order of 2^-16, as expected.

Why is the arctan used in the first place? Maybe we can add some 
documentation at the top of steer.cpp to say what we are trying to 
simulate.

Hm.. also note potential interaction between this and 
maximum steering speed of each wheel. Furthermore, I have observed 
that in most cars the wheels also tilt when you steer, perhaps we can 
mark steering for future review.

> Bye, Bernhard.
>

Thanks
>
>

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

Hi all

I have another attempt to fix it (simuv2), could you try/analyse it, does it 
work for you?

Bye, Bernhard.


-- 

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

Tested for a bit and committed it to CVS in simuv3/steer.cpp

The bug is really not very visible, but I definitely affects my 
steering with low sensitivity (say < 0.5) and 0 dead zone, especially 
in positioning manouevers on long straights, where you must steer only 
a little bit to get your car to the required position.

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

> Hi Bernhard,
>
> Finally i found the problem with the steering.
> Look in simuv2/steer.cpp. There you can see, if steer is between 0.0 and -0.01, the steer value for right wheel is set to steer and the value for left wheel is set to -steer. This means, steering a little to right points the right wheel to right and the left wheel to left.
>

Actually, I looked a bit into that and I plotted the relationship 
steer and steer2. It seems that the (fabs(steer)>0.01) check makes 
things assymetric in [-0.01, 0.01]. The correct behaviour should be:

if (fabs(steer) > 0.01f) {
 	tanSteer = tan(steer);
 	steer2 = atan2(car->wheelbase , fabs(car->wheelbase / tanSteer) - car->wheeltrack);
} else if (steer < 0.0f) {
 	steer2 = -steer;
} else {
 	steer2 = steer;
}

with the rest of the code unchanged


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

Hi Eugen

> Hi Bernhard,
> 
> Finally i found the problem with the steering.
> Look in simuv2/steer.cpp. There you can see, if steer is between 0.0 and -0.01, the steer value for right wheel is set to steer and the value for left wheel is set to -steer. This means, steering a little to right points the right wheel to right and the left wheel to left.

I will have a look at this. Thank you for your help, bye

Bernhard.

-- 

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

Hi Bernhard,

Finally i found the problem with the steering.
Look in simuv2/steer.cpp. There you can see, if steer is between 0.0 and -=
0.01, the steer value for right wheel is set to steer and the value for le=
ft wheel is set to -steer. This means, steering a little to right points t=
he right wheel to right and the left wheel to left.

bye,
Eugen


be...@bl... schrieb am 23.08.05 14:19:33:
>=20
> Hi Eugen
>=20
> > You can see the reported problem after making some modifications to th=
e source code.
> > I have inserted the following code lines into human.cpp at line about =
642, where steer command is set.
> >=20
> > 	if( fabs(steerValue) > steerLimit )
> > 		steerDelta *=3D -1;
> >=20
> > 	steerValue +=3D steerDelta;
> >=20
> > 	car->=5FsteerCmd =3D steerValue;
> >=20
> > You also have to define the variables:
> >=20
> > static float steerValue =3D 0;
> > static float steerDelta =3D 0.0005;
> > static float steerLimit =3D 0.010;
> >=20
> > Now compile and start a practice race. The car should steer (accelerat=
e first) to left and to right alternating. I only see it steer to left. Wh=
en i set steerLimit to 0.015, the car steers to right, but not so smooth a=
s to left.
>=20
> Here it steers exactly as smooth in both directions when I output the va=
lue of=20
> steer.cpp, simuv2... add printf's there for steer and steer2, run the ti=
me at=20
> times "0.02" and watch the values. To clarify, it goes not straight here=
 as=20
> well, but that can have a lot of different reasons, steering is perfect =
here, it=20
> goes smooth from left to right and so on.
>=20
> What could cause going non straight is "logigal" physics (because we acc=
elerate=20
> we do not (!) add the same amount of angular momentum per timestep, depe=
nds on=20
> actual steer value and speed) or problems in e.g. the tire model, differ=
ential=20
> code, etc, actually what differential/car are you using for this test (d=
o you=20
> use a custom setup)=3F Perhaps your car setup is broken (e.g. one suspensi=
on/ride=20
> height different)=3F
>=20
> If you use a "free" differential and 1.2.3 has not this bug, then it mig=
ht be=20
> the new differential code... could you check:
> - Do you have the same problem with the same car/setup on 1.2.3=3F
> - What happens if you change the differential to limited slip=3F
>=20
> Additional info needed if you really believe in a bug:
> - Car
> - Track
> - setup xml if not the default
> - preferences.xml
>=20
>=20
> Bye, Bernhard.
>=20
>=20
> --=20
>=20
> Visit my homepage http://www.berniw.org
> Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb=

>=20
=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F
Mit der Gruppen-SMS von WEB.DE FreeMail k=F6nnen Sie eine SMS an alle=20
Freunde gleichzeitig schicken: http://freemail.web.de/features/=3Fmc=3D021179

On Sunday 28 August 2005 19:10, David Haggett wrote:
> I'm following the instructions on the web site and in the INSTALL file
> downloaded with the "all-in-one" source tarball.
>
> I've checked I have all the dependencies installed.
>
> ./configure completes without errors, but when I issue `make` I get
>   "make: `Make-config' is up to date."
> instead of the usual compilation output.
>
> What am I missing?

Oops - found it in FAQ 

4.2.5 TORCS does not compile with make: 'Make-config' is up to date.

You have some environment variables set like $TORCS_BASE and $MAKE_DEFAULT (to 
follow the robot tutorial), they are pointing now to the wrong directory. 
Unset or update the variables suitable for the new directories. 
-- 

David Haggett

I'm following the instructions on the web site and in the INSTALL file 
downloaded with the "all-in-one" source tarball.

I've checked I have all the dependencies installed.

./configure completes without errors, but when I issue `make` I get
  "make: `Make-config' is up to date."
instead of the usual compilation output.

What am I missing?
-- 

David Haggett

Hi all

TORCS, The Open Racing Car Simulator is a highly portable multi platform car 
racing simulation. It is used as ordinary car racing game, as AI racing game and 
as research platform. It runs on Linux (x86, AMD64 and PPC), FreeBSD and Windows.

The 1.2.4 release highlights are reworked tracks and cars, updated sound and 
added OpenAL support, more clever and new opponents, support for texture 
compression and downscaling, heavily improved collision detection and response, 
a Windows debug project, lots of little improvements and bug fixes, and an 
updated track editor.

Full release notes: http://www.torcs.org
Recent screenshots: http://www.berniw.org/trb/gallery/gallery.php
Older screenshots: http://torcs.sourceforge.net/sections.php?op=listarticles&secid=3

Bye, have fun,

Bernhard.


-- 

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

Hi Bernhard,

I used Mercedes CLK on different tracks. I did not change the setup of the=
 car, and i have seen this problem in 1.2.3 and 1.2.4-test3.
I will try to do some test this weekend.

Thanks,
Eugen

be...@bl... schrieb am 23.08.05 14:19:33:
>=20
> Hi Eugen
>=20
> > You can see the reported problem after making some modifications to th=
e source code.
> > I have inserted the following code lines into human.cpp at line about =
642, where steer command is set.
> >=20
> > 	if( fabs(steerValue) > steerLimit )
> > 		steerDelta *=3D -1;
> >=20
> > 	steerValue +=3D steerDelta;
> >=20
> > 	car->=5FsteerCmd =3D steerValue;
> >=20
> > You also have to define the variables:
> >=20
> > static float steerValue =3D 0;
> > static float steerDelta =3D 0.0005;
> > static float steerLimit =3D 0.010;
> >=20
> > Now compile and start a practice race. The car should steer (accelerat=
e first) to left and to right alternating. I only see it steer to left. Wh=
en i set steerLimit to 0.015, the car steers to right, but not so smooth a=
s to left.
>=20
> Here it steers exactly as smooth in both directions when I output the va=
lue of=20
> steer.cpp, simuv2... add printf's there for steer and steer2, run the ti=
me at=20
> times "0.02" and watch the values. To clarify, it goes not straight here=
 as=20
> well, but that can have a lot of different reasons, steering is perfect =
here, it=20
> goes smooth from left to right and so on.
>=20
> What could cause going non straight is "logigal" physics (because we acc=
elerate=20
> we do not (!) add the same amount of angular momentum per timestep, depe=
nds on=20
> actual steer value and speed) or problems in e.g. the tire model, differ=
ential=20
> code, etc, actually what differential/car are you using for this test (d=
o you=20
> use a custom setup)=3F Perhaps your car setup is broken (e.g. one suspensi=
on/ride=20
> height different)=3F
>=20
> If you use a "free" differential and 1.2.3 has not this bug, then it mig=
ht be=20
> the new differential code... could you check:
> - Do you have the same problem with the same car/setup on 1.2.3=3F
> - What happens if you change the differential to limited slip=3F
>=20
> Additional info needed if you really believe in a bug:
> - Car
> - Track
> - setup xml if not the default
> - preferences.xml
>=20
>=20
> Bye, Bernhard.
>=20
>=20

=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F
Mit der Gruppen-SMS von WEB.DE FreeMail k=F6nnen Sie eine SMS an alle=20
Freunde gleichzeitig schicken: http://freemail.web.de/features/=3Fmc=3D021179

Hi Eugen

> You can see the reported problem after making some modifications to the source code.
> I have inserted the following code lines into human.cpp at line about 642, where steer command is set.
> 
> 	if( fabs(steerValue) > steerLimit )
> 		steerDelta *= -1;
> 
> 	steerValue += steerDelta;
> 
> 	car->_steerCmd = steerValue;
> 
> You also have to define the variables:
> 
> static float steerValue = 0;
> static float steerDelta = 0.0005;
> static float steerLimit = 0.010;
> 
> Now compile and start a practice race. The car should steer (accelerate first) to left and to right alternating. I only see it steer to left. When i set steerLimit to 0.015, the car steers to right, but not so smooth as to left.

Here it steers exactly as smooth in both directions when I output the value of 
steer.cpp, simuv2... add printf's there for steer and steer2, run the time at 
times "0.02" and watch the values. To clarify, it goes not straight here as 
well, but that can have a lot of different reasons, steering is perfect here, it 
goes smooth from left to right and so on.

What could cause going non straight is "logigal" physics (because we accelerate 
we do not (!) add the same amount of angular momentum per timestep, depends on 
actual steer value and speed) or problems in e.g. the tire model, differential 
code, etc, actually what differential/car are you using for this test (do you 
use a custom setup)? Perhaps your car setup is broken (e.g. one suspension/ride 
height different)?

If you use a "free" differential and 1.2.3 has not this bug, then it might be 
the new differential code... could you check:
- Do you have the same problem with the same car/setup on 1.2.3?
- What happens if you change the differential to limited slip?

Additional info needed if you really believe in a bug:
- Car
- Track
- setup xml if not the default
- preferences.xml


Bye, Bernhard.


-- 

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

Hi

You can see the reported problem after making some modifications to the so=
urce code.
I have inserted the following code lines into human.cpp at line about 642,=
 where steer command is set.

	if( fabs(steerValue) > steerLimit )
		steerDelta *=3D -1;

	steerValue +=3D steerDelta;

	car->=5FsteerCmd =3D steerValue;

You also have to define the variables:

static float steerValue =3D 0;
static float steerDelta =3D 0.0005;
static float steerLimit =3D 0.010;

Now compile and start a practice race. The car should steer (accelerate fi=
rst) to left and to right alternating. I only see it steer to left. When i=
 set steerLimit to 0.015, the car steers to right, but not so smooth as to=
 left.

bye,
Eugen

be...@bl... schrieb am 22.08.05 22:47:29:
>=20
> Hi all
>=20
> >>> There is something wrong with car steering. I used a mouse. When i=20
> >>> steer to left, car starts steering to left immediately as expected.=20
> >>> Steering to right does nothing first. After moving the mouse further=
=20
> >>> the car steer angle is changed immediately to a certain value and=20
> >>> then steering is normal again. It seems there is a difference betwee=
n=20
> >>> steering left and right at small angles.
> >>
> >>
> >=20
> > I have posted a message regarding that also on the developer list. It =
is=20
> > the same thing with the joystick control. I am not sure what is causin=
g=20
> > it either. It could be a problem with bad center/left/right values,=20
> > coupled with the nonlinearity of the control transfer function..
> >=20
> > Aha! My guess is this:
> > right now we are passing the raw left-right values through some=20
> > nonlinear function f() and the final output is
> >=20
> > steer =3D f(right) - f(-left);
> >=20
> > However, becaus measuring the center is always a bit wrong, it can cau=
se=20
>=20
> Hmm, is it in a relevant way=3F At least I have never seen a case myself w=
here it=20
> did matter (tested with 2 Joysticks, 3 wheels and some mice).
>=20
> > many troubles at the switch over point..
> >=20
> > So, for me a better way would be
> >=20
> > steer =3D f(right-left);
>=20
> Depends on what f actually does, if it fixes the non symmetry of some=20
> controllers this will not work, but we can have a look at this for 1.3.0=
, there=20
> will be a solution for sure. I will put it on the todo list after the 1.=
2.4 release.
>=20
> Bye, Bernhard.
>=20
>=20
> --=20
>=20
> Visit my homepage http://www.berniw.org
> Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb=

>=20
>=20
>=20
> -------------------------------------------------------
> SF.Net email is Sponsored by the Better Software Conference & EXPO
> September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practi=
ces
> Agile & Plan-Driven Development * Managing Projects & Teams * Testing & =
QA
> Security * Process Improvement & Measurement * http://www.sqe.com/bsce5s=
f
> =5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F
> Torcs-users mailing list
> Tor...@li...
> https://lists.sourceforge.net/lists/listinfo/torcs-users


=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F
Mit der Gruppen-SMS von WEB.DE FreeMail k=F6nnen Sie eine SMS an alle=20
Freunde gleichzeitig schicken: http://freemail.web.de/features/=3Fmc=3D021179

Hi all

>>> There is something wrong with car steering. I used a mouse. When i 
>>> steer to left, car starts steering to left immediately as expected. 
>>> Steering to right does nothing first. After moving the mouse further 
>>> the car steer angle is changed immediately to a certain value and 
>>> then steering is normal again. It seems there is a difference between 
>>> steering left and right at small angles.
>>
>>
> 
> I have posted a message regarding that also on the developer list. It is 
> the same thing with the joystick control. I am not sure what is causing 
> it either. It could be a problem with bad center/left/right values, 
> coupled with the nonlinearity of the control transfer function..
> 
> Aha! My guess is this:
> right now we are passing the raw left-right values through some 
> nonlinear function f() and the final output is
> 
> steer = f(right) - f(-left);
> 
> However, becaus measuring the center is always a bit wrong, it can cause 

Hmm, is it in a relevant way? At least I have never seen a case myself where it 
did matter (tested with 2 Joysticks, 3 wheels and some mice).

> many troubles at the switch over point..
> 
> So, for me a better way would be
> 
> steer = f(right-left);

Depends on what f actually does, if it fixes the non symmetry of some 
controllers this will not work, but we can have a look at this for 1.3.0, there 
will be a solution for sure. I will put it on the todo list after the 1.2.4 release.

Bye, Bernhard.


-- 

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

>
> First thank you for testing and your very detailed and valuable reports:-)
>
>> i update my report with some new informations after some tests with 
>> 1.2.4-test3.
>> 
>> There is something wrong with car steering. I used a mouse. When i steer to 
>> left, car starts steering to left immediately as expected. Steering to 
>> right does nothing first. After moving the mouse further the car steer 
>> angle is changed immediately to a certain value and then steering is normal 
>> again. It seems there is a difference between steering left and right at 
>> small angles.
>

I have posted a message regarding that also on the developer list. It 
is the same thing with the joystick control. I am not sure what is 
causing it either. It could be a problem with bad center/left/right 
values, coupled with the nonlinearity of the control transfer 
function..

Aha! My guess is this:
right now we are passing the raw left-right values through some 
nonlinear function f() and the final output is

steer = f(right) - f(-left);

However, becaus measuring the center is always a bit wrong, it can 
cause many troubles at the switch over point..

So, for me a better way would be

steer = f(right-left);

to make sure that the transition between left and right is smooth.

I will test it in 1 week if Bernhard doesn't have time until then.

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

Hi Eugen

> There is something wrong with car steering. I used a mouse. When i steer to left, car starts steering to left immediately as expected. Steering to right does nothing first. After moving the mouse further the car steer angle is changed immediately to a certain value and then steering is normal again. It seems there is a difference between steering left and right at small angles.

Not confirmed, works fine here. Could there be any other reason for that (dirt 
in the mouse, mouse driver glitch, wireless mouse problem, mouse "calibration" 
in TORCS), is it really a TORCS issue?
If you think it is a TORCS issue, could you send me your preferences.xml such 
that I can have a ride with your settings (I did now steer on mouse, accel and 
brake on mouse buttons)?

Bye, thank you,

Bernhard.


-- 

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

Hi Eugen

First thank you for testing and your very detailed and valuable reports:-)

> i update my report with some new informations after some tests with 1.2.4-test3.
> 
> There is something wrong with car steering. I used a mouse. When i steer to left, car starts steering to left immediately as expected. Steering to right does nothing first. After moving the mouse further the car steer angle is changed immediately to a certain value and then steering is normal again. It seems there is a difference between steering left and right at small angles.

I will have a look at this, but it will not be fixed for 1.2.4 even if it is a 
problem. The code regarding to mouse control is that same like in 1.2.3, so it 
is quite likely equally bad or good (if not a weird side effect causes your 
problem).

> Engine sound
> The new engine sounds are sometimes annoing, when driving a long time at a high engine speed.
> Specially with the Porsche GT1 there are high sounds starting with 7000 rev/minute.
> 
> In the engine sound of Acura, Nascar and McLaren F1 there are soft clicks each 3-5 seconds.

I agree, but will not be fixed for 1.2.4. There is as well the question if the 
samples cause the trouble or the code. Does it happen as well with the plib backend?

> 
> Torcs exited after starting a race with the following trace log (i think it was Viper on B-Speedway, starting the race once again did not lead to the fault).
> This happened only 2 times while i was playing for several hours.

Ok, that is exactly 2 times too often then...

The backtrace below is quite esotheric:-) I guess you reached the per process 
memory limit due to the remaining texture state memory/plib leaks in TORCS, is 
that possible (that is definitively very hard to fix, I will not go into details 
here, but in a nutshell the problem are some TORCS issues and that plib is 
simply not suitable for dynamic loading -> this will remain a problem for quite 
some time... say 2-3 years or so)?

Bye, thank you very much,

Bernhard


-- 

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

Hi all,

i update my report with some new informations after some tests with 1.2.4-=
test3.

There is something wrong with car steering. I used a mouse. When i steer t=
o left, car starts steering to left immediately as expected. Steering to r=
ight does nothing first. After moving the mouse further the car steer angl=
e is changed immediately to a certain value and then steering is normal ag=
ain. It seems there is a difference between steering left and right at sma=
ll angles.

Engine sound
The new engine sounds are sometimes annoing, when driving a long time at a=
 high engine speed.
Specially with the Porsche GT1 there are high sounds starting with 7000 re=
v/minute.

In the engine sound of Acura, Nascar and McLaren F1 there are soft clicks =
each 3-5 seconds.

Torcs exited after starting a race with the following trace log (i think i=
t was Viper on B-Speedway, starting the race once again did not lead to th=
e fault).
This happened only 2 times while i was playing for several hours.

Thanks,
Eugen

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 16384 (LWP 1839)]
0x4095e21b in strlen () from /lib/i686/libc.so.6
(gdb) #0  0x4095e21b in strlen () from /lib/i686/libc.so.6
#1  0x400c774e in ulStrEqual(char const*, char const*) (
    s1=3D0x4ac09661 <Address 0x4ac09661 out of bounds>, s2=3D0x625fd013 ".png"=
)
    at ul.cxx:475
#2  0x4007f720 in ssgAddTextureFormat(char const*, bool (*)(char const*, s=
sgTextureInfo*)) (extension=3D0x625fd013 ".png",
    loadfunc=3D0x62566c10 <grLoadPngTexture(char const*, ssgTextureInfo*)>)
    at ssgLoadTexture.cxx:206
#3  0x62560f80 in grLoadScene(tTrack*) (track=3D0xbab1450) at grscene.cpp:19=
2
#4  0x62551301 in initTrack(tTrack*) (track=3D0xbab1450) at grmain.cpp:449
---Type <return> to continue, or q <return> to quit---#5  0x403eed82 in Re=
RaceEventInit() () at racemain.cpp:116
#6  0x403f0fbd in ReStateManage() () at racestate.cpp:74
#7  0x403ed1bb in ReStartNewRace(void*) () at raceinit.cpp:129
#8  0x403c70d6 in gfuiButtonAction(int) (action=3D1) at guibutton.cpp:577
#9  0x403c5f57 in gfuiMouseAction(void*) (vaction=3D0xff666d4e)
    at guiobject.cpp:458
#10 0x403c3db7 in gfuiMouse (button=3D0, state=3D1, x=3D351360, y=3D147) at gui.cp=
p:390
#11 0x4042d995 in processEventsAndTimeouts () from /usr/lib/libglut.so.3
#12 0x4042e039 in idleWait () from /usr/lib/libglut.so.3
#13 0x4042e76d in glutMainLoop () from /usr/lib/libglut.so.3
#14 0x08048f2f in main (argc=3D7, argv=3D0xbffff264) at main.cpp:110



Christos Dimitrakakis <dim...@id...> schrieb am 16.08.05 17:08:13:
>=20
> Thank you for your report. Let me try and reply:
>=20
> > I found the following issues in the version 1.2.4-test1 with linux,=20
> > maybe they are still in the test2 version, i couldn't test it.
> >
> > Fuel is refilled when you don't accelerate (seen with Mercedes CLK=20
> > DTM). To reproduce this: accelerate until fuel display changes, then=20
> > stop accelerating, the displayed fuel ammount increases by 0.1.
> >
>=20
> This is a known bug, fixed in simuv3.
>=20
> > When a control is not configured, then it is accessible via '-' key.=20
> > To see this: unconfigure "light command" start a race and press '-'=20
> > to slow down the time, the lights go on (with the right car model).
> >
>=20
> Thanks, noted. Maybe this is related to other UI bugs.
>=20
> > With the most cars the engine sound had strange noise. Maybe this is=20
> > my system, have to check it on windows with the test2 version. I=20
> > have used the modified OpenAL from the torcs page.
> >
>=20
> Could you perhaps elaborate=3F Do you mean that the engine sound is not=20
> good, or=3F What is the difference between 'most cars' and the rest=3F
>=20
> > Sometimes torcs just exits when starting a race with the following=20
> > error: line 86: 1816 (or other number) Speicherzugriffsfehler=20
> > (=3DMemory access error) $LIBDIR/torcs-bin (and so on). I will try to=20
> > reproduce this.
> >
>=20
> This could be one of the UI bugs I mentioned: I think running torcs -d=20
> will enable debugging for 1.2.4-test1 - could you return the output of=20
> this=3F
>=20
> > Thanks,
> > Eugen
> >
>=20
> Thanks for your input,
>=20
> Christos
>=20
>=20
> > =5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=
=5F=5F=5F
> > Mit der Gruppen-SMS von WEB.DE FreeMail k=F6nnen Sie eine SMS an alle
> > Freunde gleichzeitig schicken: http://freemail.web.de/features/=3Fmc=3D021=
179
> >
> >
> >
> >
> >
> > -------------------------------------------------------
> > SF.Net email is Sponsored by the Better Software Conference & EXPO
> > September 19-22, 2005 * San Francisco, CA * Development Lifecycle Prac=
tices
> > Agile & Plan-Driven Development * Managing Projects & Teams * Testing =
& QA
> > Security * Process Improvement & Measurement * http://www.sqe.com/bsce=
5sf
> > =5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F
> > Torcs-users mailing list
> > Tor...@li...
> > https://lists.sourceforge.net/lists/listinfo/torcs-users
> >
>=20
> --=20
> Christos Dimitrakakis
> IDIAP (http://www.idiap.ch/~dimitrak/main.html)


=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F
Verschicken Sie romantische, coole und witzige Bilder per SMS!
Jetzt bei WEB.DE FreeMail: http://f.web.de/=3Fmc=3D021193

On Sun, Aug 21, 2005 at 02:55:17AM +0200, Bernhard Wymann wrote:
> Hi Bob
> 
> Hmm, are this all errors, no report about a missing zlib.h? Do you have 
> zlib.h
> installed (in SuSE it is in package zlib-devel. perhaps similar for Fedora)?

Yes, I have both zlib and zlib-devel installed.  I did a "locate
zlib.h" and discovered I also had a /usr/local/include/zlib.h dated
1997 and indicating version 1.0.4.  It was picking up that one.  I
deleted it, ran "make clean" and "./configure", and now make completes
satisfactorily.

Now I have to figure out how to get my joystick recognized.  Back to
the docs and FAQ.

Thanks for the clue.

> Bye, Bernhard.

Cheers,
-- 
Bob McClure, Jr.             Bobcat Open Systems, Inc.
rob...@ea...  http://www.bobcatos.com
Peace at any price is inflationary.

Hi Bob

Hmm, are this all errors, no report about a missing zlib.h? Do you have zlib.h
installed (in SuSE it is in package zlib-devel. perhaps similar for Fedora)?

Bye, Bernhard.

> I'm having a problem compiling torcs (v1.2.3).  I checked the FAQs, I
> don't find any archives to search, and a Google search turned up dry.
> 
> System is Fedora Core 3.  I installed freeglut-2.2.0-14 by RPM, and
> the plib-1.8.4 tarball.  I did the default ./configure, but make
> reports:
> 
> g++ -I/u/proj/torcs-1.2.3/export/include -I/u/proj/torcs-1.2.3 -g -O2 -Wall -fPIC -O2 -DUSE_RANDR_EXT -Wall -fPIC -O2 -DUSE_RANDR_EXT  -D_SVID_SOURCE -D_BSD_SOURCE -DSHM -DHAVE_CONFIG_H  -c grloadac.cpp
> grloadac.cpp: In function `int do_object(char*)':
> grloadac.cpp:420: error: `gzgets' undeclared (first use this function)
> grloadac.cpp:420: error: (Each undeclared identifier is reported only once for each function it appears in.)
> grloadac.cpp: In function `int do_data(char*)':
> grloadac.cpp:482: error: `gzgetc' undeclared (first use this function)
> grloadac.cpp: In function `int do_numvert(char*)':
> grloadac.cpp:683: error: `gzgets' undeclared (first use this function)
> grloadac.cpp: In function `int do_numsurf(char*)':
> grloadac.cpp:730: error: `gzgets' undeclared (first use this function)
> grloadac.cpp: In function `int do_surf(char*)':
> grloadac.cpp:743: error: `gzgets' undeclared (first use this function)
> grloadac.cpp: In function `int do_refs(char*)':
> grloadac.cpp:790: error: `gzgets' undeclared (first use this function)
> grloadac.cpp: In function `ssgEntity* myssgLoadAC(const char*, const ssgLoaderOptions*)':
> grloadac.cpp:1212: error: `gzgets' undeclared (first use this function)
> make[5]: *** [grloadac.o] Error 1
> 
> What am I missing, besides gzgets() and gzgetc()?
> 
> Cheers,


-- 

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

I'm having a problem compiling torcs (v1.2.3).  I checked the FAQs, I
don't find any archives to search, and a Google search turned up dry.

System is Fedora Core 3.  I installed freeglut-2.2.0-14 by RPM, and
the plib-1.8.4 tarball.  I did the default ./configure, but make
reports:

g++ -I/u/proj/torcs-1.2.3/export/include -I/u/proj/torcs-1.2.3 -g -O2 -Wall -fPIC -O2 -DUSE_RANDR_EXT -Wall -fPIC -O2 -DUSE_RANDR_EXT  -D_SVID_SOURCE -D_BSD_SOURCE -DSHM -DHAVE_CONFIG_H  -c grloadac.cpp
grloadac.cpp: In function `int do_object(char*)':
grloadac.cpp:420: error: `gzgets' undeclared (first use this function)
grloadac.cpp:420: error: (Each undeclared identifier is reported only once for each function it appears in.)
grloadac.cpp: In function `int do_data(char*)':
grloadac.cpp:482: error: `gzgetc' undeclared (first use this function)
grloadac.cpp: In function `int do_numvert(char*)':
grloadac.cpp:683: error: `gzgets' undeclared (first use this function)
grloadac.cpp: In function `int do_numsurf(char*)':
grloadac.cpp:730: error: `gzgets' undeclared (first use this function)
grloadac.cpp: In function `int do_surf(char*)':
grloadac.cpp:743: error: `gzgets' undeclared (first use this function)
grloadac.cpp: In function `int do_refs(char*)':
grloadac.cpp:790: error: `gzgets' undeclared (first use this function)
grloadac.cpp: In function `ssgEntity* myssgLoadAC(const char*, const ssgLoaderOptions*)':
grloadac.cpp:1212: error: `gzgets' undeclared (first use this function)
make[5]: *** [grloadac.o] Error 1

What am I missing, besides gzgets() and gzgetc()?

Cheers,
-- 
Bob McClure, Jr.             Bobcat Open Systems, Inc.
rob...@ea...  http://www.bobcatos.com
God doesn't have (or need) a Plan B.

Hi T.P.

> I agree that the job SHOULD be done adequately by the OS, but it doesn't 
> appear to be the case with Linux. Gemini doesn't produce a driver for 
> Linux like
> they do for Windows. In this case, what do users of Linux do? Just like 
> FlightGear,
> users of Linux HAVE to rely on the programs to COMPENSATE for the 
> shortcomings of the
> OS, i.e. inadequate calibration of joysticks. Generally, TORCS does a 
> pretty good job as
> it is, but my case clearly presents a problem. I'm not even sure whether 
> a generic
> method to compensate for controller problems such as mine could be 
> implemented on an
> application basis. <sigh>

Perhaps it could be done, but IMHO it should not be done, the right way is to:
- Write proper device drivers instead of patching around the problem in hundrets 
of applications (nonsense, waste of time, no).
- Buy just devices from vendors which provide drivers or release documentation 
about the device and free drivers are already available (you cannot be sure that 
somebody writes a driver even if the docs/specs are available).

Bye, Bernhard.


-- 

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