[ODE] test program problems
vmelkon at yahoo.com
Mon Aug 22 19:30:07 MST 2005
I had some trouble myself in the begining. It was due
to ode.dll that had some issues.
The current ODE version is working fine for me except
the demoes had some issues with multihtreading, so I
got rid of that.
--- Martijn Buijs <buijs512 at planet.nl> wrote:
> I'm new to ODE. I am planning to use ODE in one of
> my projects, however I'm having a bit of trouble
> getting things to work.
> First, all of the test programs are running to fast
> on my machine (Windows XP), aren't the demos
> supposed to be framerate independent? As workaround
> I have enabled v-sync so things look better, but
> there are other (mayor) problems still;
> Many of the provided pre-compiled demo programs
> crash, or don't work properly on any of my machines.
> I've re-compiled all of the test apps of most
> releases but they all exhibit the same problems,
> though all user programs (those linked on the ODE
> site) I tried work flawlessly.
> Am I doing something wrong?
> Anyway, here are some descriptions of the problems:
> test_boxstack.exe - Runs fine.
> test_buggy.exe - No problems.
> test_chain1.exe - No problems.
> test_chain2.exe - No problems besides occasionally
> intersecting boxes.
> test_collision.exe - Instantly displays an error
> message titled "ODE INTERNAL ERROR 0" and the
> message reads "bad". Clicking "OK" closes the app
> and console window. The console displays a single
> line "testing batch 1 (1000 reps)..."
> The DLL release spits out an additional "abnormal
> program termination" MSVC++ warning.
> test_crash.exe - No problems.
> test_friction.exe - No probs.
> test_hinge.exe - No probs.
> test_I.exe - Console reports 2 errors about every
> second (when v-sync enabled, otherwise console
> floods even more quickly):
> w-error = 2.3304e-004 (0.45,-0.19,0.65) and (0.32,
> 0.54, -0.45)
> q-error = 4.3434e-005
> These numbers/coordinates vary with each message,
> but note that the errors are always reported in
> pairs (always a 'q' and 'w' error). The viewport
> window is visible, the cubes spin for a second,
> then snap to a different, seemingly random, position
> (however, the yellow cubes do stay in relation
> to the white object until it snaps).
> test_joints.exe - Some test pass, some fail. The
> error "error not computed" occurs a few times and
> the test number jumps from 3 to 200 (not sure its
> supposed to). At some point the app exits.
> test_moving_trimesh.exe - When a single spawned
> bunny drops it intersects with the static two, and
> will jitter. Newly spawned bunnies do the same, but
> there's no trouble with any spawned primitives.
> Performance does drops considerably with only a few
> (4-5) bunnies and at some point the app crashes
> (generic Windows crash report warning).
> test_ode.exe - Console shows, most tests pass, some
> fail. Then a messagebox is show titled "ODE
> INTERNAL ERROR 0" with the message "A*x = b+w,
> maximum difference = 1.525879e-005" (the number is
> always the same). The last line on the console reads
> "dTestSolveLCP". Clicking "OK" closes the app.
> The DLL version also reports an "abnormal program
> termination" warning.
> test_slider.exe - When v-sync is on the yellow box
> slides all the way towards the 'ends' of the blue
> box, when v-sync is off it slides a considerably
> shorter distance (nowhere near the ends of the blue
> box). No other probs.
> test_space.exe - The boxes appear/disappear at a
> different location seemingly every frame, even with
> v-sync enabled. Can't pause the simulation from the
> test_space_stress.exe - Not unexpectedly, it takes a
> while to load and runs slow. Spheres don't move
> at all (should they?). Seems to run fine otherwise.
> test_step.exe - Same as with test_space.exe, but can
> be paused.
> test_trimesh.exe - Seems to run ok, when ~30 objects
> (any of the primitives) are dropped they
> roll down as expected, but the performance suddenly
> drops considerably and the app crashes without
> warning, before or shortly before the objects
> All in all, pretty serious I'd say.
> I was hoping to play around with the code of these
> test programs to get more familiar with ODE, but
> obviously, they have to work in order for me to
> learn from them. :-)
> ODE mailing list
> ODE at q12.org
Start your day with Yahoo! - make it your home page
More information about the ODE