IrrNewt irrlicht\newton framework >> SVN access

igorfk · Post by **igorfk** » Fri Mar 21, 2008 5:32 am

Does anynody compiled for irrlicht 1.4?

Halifax · Post by **Halifax** » Fri Mar 21, 2008 7:51 am

Wow, there's this crazy thing called it takes only a small amount of your valuable time to try it out instead of spamming. I suggest that.

igorfk · Post by **igorfk** » Fri Mar 21, 2008 6:28 pm

ops, sorry if sounds like that.
I'm asking this here because, when I'd tried to compile here, I had lots of errors. The round() function I could solve thanks to past posts here, but there is a camerabody_fps that's trying to instantiate an abstract class and I cant figure out how to fix it.

Halifax · Post by **Halifax** » Sat Mar 22, 2008 6:04 am

igorfk wrote:ops, sorry if sounds like that.
I'm asking this here because, when I'd tried to compile here, I had lots of errors. The round() function I could solve thanks to past posts here, but there is a camerabody_fps that's trying to instantiate an abstract class and I cant figure out how to fix it.

Haha, take my comments like that lightly. I was just being sarcastic, I'm not trying to be harsh.

But instead of having it use the IClassName it should use the CClassName. That's the problem with putting abstract classes in other classes. But that should fix it. (If that is how the code is structured.)

psyco001 · Post by **psyco001** » Fri Apr 25, 2008 3:28 pm

Halifax wrote:
igorfk wrote:ops, sorry if sounds like that.
I'm asking this here because, when I'd tried to compile here, I had lots of errors. The round() function I could solve thanks to past posts here, but there is a camerabody_fps that's trying to instantiate an abstract class and I cant figure out how to fix it.
Haha, take my comments like that lightly. I was just being sarcastic, I'm not trying to be harsh. Laughing

But instead of having it use the IClassName it should use the CClassName. That's the problem with putting abstract classes in other classes. But that should fix it. (If that is how the code is structured.)

jep there is a function named createCamera that want to instanciate ICameraBodyFPS and i've the same problem:
When i want to compile IrrNewt I've the error:

Code: Select all

error C2259: 'irr::newton::ICameraBodyFPS'Instance of abstract class can not be created because of member:
"bool irr::IEventReceiver::OnEvent(const irr::SEvent &)": is abstract

but the Camera-class itself was already compiled without problems!
has anyone a solution for that problem, or what am i making wrong?

Vsk · Post by **Vsk** » Fri Apr 25, 2008 4:57 pm

Red the FAQs, I think there is a problem with the signature of that operation (I think.).
If not. Let me think.

white tiger · Post by **white tiger** » Sat May 10, 2008 12:34 pm

i haven't tried to compile for 1.4 for now. it's strange, it semms there are not API changes to IEventReceiver or onevent functino from 1.3 to 1.4.
have you follow the instructions on how to compile for 1.3?

BTW

irrnewt sdk 0.4 is online again

i get the new newton 2.0 beta. it changes some function name, this can be fixed easily. but there are heavy changes wich will require some time. for example the vehicle interface and ragdool interface are completely removed from the core. instead the custom joint interface is improved and ragdool\vehicles are implemented externally in classes using custom joints.
the interface is basically the same for the ragdoll, just switch to class and call proper functions. not for the vehicle, wich now is basic vehicle (the classic implemented so far) and a new raycast vehicle.

for the new feature we have NewtonMesh to implement breakable mesh and soft bodies(this last not implemented in the beta)

i don't knwo at all how we could implement breakable meshes, because we need to reconstruct the mesh from scratch and removing the old one, wich implies remove the scene node the mesh is based, wich implement the user code that refer to this scene nodes will crash. irrlicht is not able to "redirect" the scene node to another mesh. any ideas?

white tiger · Post by **white tiger** » Sat May 10, 2008 12:39 pm

in camera_fps.cpp change

Code: Select all

bool irr::newton::ICameraBodyFPS::OnEvent(irr::SEvent event) {

to

Code: Select all

bool irr::newton::ICameraBodyFPS::OnEvent(const irr::SEvent& event) {

and in camera_fps.hpp

Code: Select all

virtual bool OnEvent(irr::SEvent event);

to

Code: Select all

virtual bool OnEvent(const irr::SEvent& event);

thanks for reporting

Vsk · Post by **Vsk** » Mon May 12, 2008 8:32 pm

white tiger wrote:i haven't tried to compile for 1.4 for now. it's strange, it semms there are not API changes to IEventReceiver or onevent functino from 1.3 to 1.4.
have you follow the instructions on how to compile for 1.3?

BTW

irrnewt sdk 0.4 is online again

i get the new newton 2.0 beta. it changes some function name, this can be fixed easily. but there are heavy changes wich will require some time. for example the vehicle interface and ragdool interface are completely removed from the core. instead the custom joint interface is improved and ragdool\vehicles are implemented externally in classes using custom joints.
the interface is basically the same for the ragdoll, just switch to class and call proper functions. not for the vehicle, wich now is basic vehicle (the classic implemented so far) and a new raycast vehicle.

for the new feature we have NewtonMesh to implement breakable mesh and soft bodies(this last not implemented in the beta)

i don't knwo at all how we could implement breakable meshes, because we need to reconstruct the mesh from scratch and removing the old one, wich implies remove the scene node the mesh is based, wich implement the user code that refer to this scene nodes will crash. irrlicht is not able to "redirect" the scene node to another mesh. any ideas?

Glad to have you back!.

I think is possible, first at all, the original mesh/node, should be destroy.
It would be conceptually wrong to mantain this one, cause it has been splited. But if we could, this would make easy all the isssue.
So in the first case, I think is the programmer (irrnewt user) which has to deal with this; think about it, this happen always in all kind of programs when we destroy a object and we must be sure that all other objects that references doesn't have any problem with it in future.
So the only thing to deal about is intra-irrnewt (but not what the user declares but what the user can not make any change). About this last one, you are the only one that know it perfectly

.

Could you tell me an specific intra irrnewt case?
Have you ask Julio?, Am sure he will find a answer

.

white tiger · Post by **white tiger** » Wed May 14, 2008 11:43 am

uh i agree. maybe we could make a callback that tell the user the pointer is no longer available and should be set to NULL manually by the user and submit the two bodies the original body was splitted into. i haven't asked julio yet, i will but i think yours could be a good solution

i think there should be no problem inside irrnewt. we should remove the body and create the new ones, i think newton submit the new mass for the bodies so we can recalculate the tensors of inertia and register the new bodies

Vsk · Post by **Vsk** » Wed May 14, 2008 1:43 pm

white tiger wrote:uh i agree. maybe we could make a callback that tell the user the pointer is no longer available and should be set to NULL manually by the user and submit the two bodies the original body was splitted into. i haven't asked julio yet, i will but i think yours could be a good solution
i think there should be no problem inside irrnewt. we should remove the body and create the new ones, i think newton submit the new mass for the bodies so we can recalculate the tensors of inertia and register the new bodies

Glad you like the idea.
Yes, in fact newton give you everything you need for the creation of new meshses/bodies, even more, the fracture pattern is full customizable!

Either way, make a question in here in the advanced forum, for this question, maybe there is a way to conserve (redirect) the old node, which will be an easy solution.

stevend · Post by **stevend** » Wed May 14, 2008 4:01 pm

This may be a stupid question, but,

I have compiled this for 1.4 (here is over 11 Warning in the includes, alot of them are converting F64 or F32 or something...) and I am using it in my project.

Someone is porting my engine to linux and i need to know if anybody has attempted to compile this for linux.

If no one has I will report our outcome, and ill post a download link if it is successful.

ps. Thank you white tiger for your work, it has helped me alot!

white tiger · Post by **white tiger** » Fri May 16, 2008 10:59 am

stevend wrote:This may be a stupid question, but,

I have compiled this for 1.4 (here is over 11 Warning in the includes, alot of them are converting F64 or F32 or something...) and I am using it in my project.

Someone is porting my engine to linux and i need to know if anybody has attempted to compile this for linux.

If no one has I will report our outcome, and ill post a download link if it is successful.

ps. Thank you white tiger for your work, it has helped me alot!

i have compiled on knoppix the hello world example of irrnewt 0.3 with irrlicht 1.2 some time ago. you don't need lots of changes because i use mingw under windows (not vc++) wich is a GNU compatible compiler, so you need only few changes or no changes at all to compile under linux.
BTW there is no config or make file for linux, you need to compile manually all the .cpp files under src/irrnewt_src
a download link for linux binaries is always welcome

and i'm glad you like my wrapper

Donner · Post by **Donner** » Mon May 19, 2008 5:20 pm

Hey,

is it possible to define the gravitation for the whole world or have I to define it for every object?

Thank you

D.

white tiger · Post by **white tiger** » Tue May 20, 2008 9:55 am

no you have to define it for every body