IrrODE - an Irrlicht - ODE wrapper (now with SVN)

[Brainsaw]

A new update. This update does mainly fix an issue reported in the Irrlicht forum, the length of (dynamic) capsules is now defined by the height of the object, not the ODE way (depth of the object). Additionally the IrrOdeCar demo is now mostly event driven, a huge step in it's design.

@ulao: I think I have fixed your issue (at least I hope so). Take a look at the "IrrOdeTestbed" application, there is a "capsule" test which shows the changed version. I could reproduce the issue, although my "tuned" gcc version of ODE didn't crash, but it didn't behave corretly either.

[ulao]

Hey didnt work first go but there are lots of variables here for me. Need to know at least these...

> Did you copy this to the zip from your main page, or is it only in svn?
> would using irredit make any difference?
> where is this test bead I dont see it?

I browsed the svn and I see the commit comments so I download the cpp and it looks like the change may be in the h file. (<tests/CTestCapsule.h>). so I browsed to the test folder but its all cpp file no h file. Ok I managed to get the svn down, i'll see if I can do it from there.

AHH: "sufferin succotash" ! well it compiled but now I have that check error back. Maybe best to wait for you to upload the zip, no sure why I get that building from source.

[Brainsaw]

I uploaded a complete package to my homepage, best idea is to download the new version.

[ulao]

where? the download irrode.zip is identical to yesterdays and does not seem to have the changes in it?

for fun I tried to build against your latest libirrode.lib ( in irrlib.zip ) and now that range check is back. I wonder what makes that happen?

[Brainsaw]

I'll do a check, but some of the directories should not be in the IrrOde.zip.

[ulao]

I guess as long as the lib is built in VS2008 with the capsule change( and head if applicable ). It should do the job.

[Brainsaw]

I just checked ... there are some directories in your structure that I removed some time ago (IrrOdeMarble, IrrOdePlayground, IrrOdeSurfaceEdit). I think the best idea would be to remove the folder and install it again from the latest downloaded package. I hope that it all works then (well ... it should at least).

[ulao]

there are some directories in your structure that I removed some time ago

-- um the screen shot was your structures not mine? One is the svn, and the other is a zip file. Both are from you?

Though. I did take this suggestion and downloaded the latest unzipped to a folder, and renamed the libirrode.lib to irrode.lib. I Dropped that in my lib folder for my project and pointed to the new headers. I then did a clean, rebuild. I still get this stupid run time check failure. All I know is its something in the way you compile the lib. I wish I knew more about it. Could it be the libirrode vs irrode lib thing? What ever you did on the post:
"by Brainsaw » Mon Feb 27, 2012 8:17 am"
Fixed the range check junk last time.

[Brainsaw]

Damnit ... you're right ... there are still some unneeded directories in the zip (I should check that more often). But I still got no idea what's wrong with the libs I compile. Should I provide the complete package, i.e. also include the ODE static libs I use?

[ulao]

at this point, anything can help. I think its always best to include that stuff for users that want to trouble shoot compile issues.

[Brainsaw]

OK, I'll upload a new version within the next days, but first I need to create a new branch on sourceforge, at the moment I'm doing quite heavy API changes (trying to get the whole thing multithreaded ... one thread ODE, on thread Irrlicht for the start)

[ulao]

Ahh, that would be a nice to have.

[Brainsaw]

I'm at the moment uploading a new version:

This time the update includes quite a huge API change: a lot of functions have been moved from the CIrrOdeManager class to CIrrOdeWorld. This is planned as a step towards multithreading. This version does also include everything needed to compile (also ODE libs), so the ODE is no longer necessary.

Last but not least: upgraded to ODE 0.12

[ulao]

Well the new download didnt fix my run time check failure, though I finaly found the sucker. It was an extra link search folder hidden in there. Finally!

So I rebuilt all and get this

.\source\phys\physmanager.cpp(37) : error C2039: 'removeWorld' : is not a member of 'irr::ode::CIrrOdeManager'

I'm guessing removeWorld has changed? Is it handled in the drop() now?

ODE internal error 2
Also if I skip that error I get that old bad argument(s) capsule.cpp:51

So I'm guessing capsule is not read then?

[Brainsaw]

Well, the ODE manager got some API changes, lots of stuff moved to CIrrOdeWorld, and the add/removeworld were removed. The worlds are now found when the physics is initialized by iterating the scene, tso removeWorld is no longer used. Did you do a complete rebuild? Maybe delete everything from your binary folder, I think VC keeps some files around with debug/linker information that might not get updated.

I think I'll try to build the IrrOdeTestbed with VC, there is also a capsule-test included.