Control surfaces moving in wrong directions

[01010101lzy]

So now I am working around my B9 PWings Modified mod, and found that in 1.1 the direction of rotation of control surfaces is inverted in one side (See pic below, by @qromodynmc who is trying to pitch up)

Aero calculations are correct in the picture.

Do anyone have any idea about it?

[Crzyrndm]

The same issue is present in the base pack, my suspicion (haven't had a chance to test) is that it's the negative scaling of mirrored parts that's causing this and that is no longer supported in the stock implementation of control surfaces. Only other thing I can think of is that the rotation axis is wrong now (because they pitch for roll and roll for pitch...) which will require messing around in unity and exporting with the new Part tools.

PS
When I briefly tested this, it appeared to be visual only with the plane still pitching up with minimal roll (and rolling instead of pitching)
PPS
Unlikely to be caused by the plugin itself, there is nothing in there that deals with control inputs
 

Edited April 20, 2016 by Crzyrndm

[01010101lzy]

[blanked because of wrong reply]

Edited April 20, 2016 by 01010101lzy

[01010101lzy]

3 hours ago, Crzyrndm said:

The same issue is present in the base pack, my suspicion (haven't had a chance to test) is that it's the negative scaling of mirrored parts that's causing this and that is no longer supported in the stock implementation of control surfaces. Only other thing I can think of is that the rotation axis is wrong now (because they pitch for roll and roll for pitch...) which will require messing around in unity and exporting with the new Part tools.

PS
When I briefly tested this, it appeared to be visual only with the plane still pitching up with minimal roll (and rolling instead of pitching)
PPS
Unlikely to be caused by the plugin itself, there is nothing in there that deals with control inputs
 

Okay I've tested with the MirrorRefAxis deleted and the rotation worked well.

[qromodynmc]

So, any fixes now?

[01010101lzy]

15 hours ago, qromodynmc said:

So, any fixes now?

If you use the wing with symmetric texture and delete the "mirrorRefAxis" line in each wing_procedural_X.cfg, it should be behaving correctly. This is the only fix available now..

[NathanKell]

@Claw ping!

[Crzyrndm]

On 20/04/2016 at 11:32 PM, Crzyrndm said:

When I briefly tested this, it appeared to be visual only with the plane still pitching up with minimal roll (and rolling instead of pitching)

Further testing confirms this is a visual only effect. Forces applied are correctly oriented (looks very wierd...)

[Claw]

So nothing that I know of changed with mirror axis, and the only thing that changed with control surfaces was with the deploy option. So I'm not quite sure here.

So I'm at a loss for why it's working differently than in 1.0.5, though control surfaces shouldn't need the mirror axis anyway.

[Crzyrndm]

These need to be turned inside out (-ve scaling) for mirroring because the top and bottom are different. With stock control surfaces that don't matter which way up they face, you don't have to use the -ve scaling factor, they just get flipped. I would be looking at the code that moves the surface to see if it checks the mesh scaling is -ve and inverts it's direction if so, not anything to do with mirror axis itself

I can write something to determine which side should be the top without too much trouble (and need to anyway for various reasons), but functionality definitely changed here

Edited April 22, 2016 by Crzyrndm

[inkwell]

just would like to express my gratitude! having mostly functional pwings this soon after release is quite nice!

[Crzyrndm]

On 22/04/2016 at 5:49 PM, NathanKell said:

@Claw ping!

On 22/04/2016 at 7:23 PM, Claw said:

So nothing that I know of changed with mirror axis, and the only thing that changed with control surfaces was with the deploy option. So I'm not quite sure here.

So I'm at a loss for why it's working differently than in 1.0.5, though control surfaces shouldn't need the mirror axis anyway.

I believe I've resolved this now so no worries about the regression (I believe PWings were the only mods using the -ve scaling in the first place, certainly the only one where control inputs would matter). It's actually kinda nice having it more standard since it has caused a few issues with FAR in the past. Would be nice if the part.mirrored property wasn't always false though...

@01010101lzy

See: https://github.com/Crzyrndm/B9-PWings-Fork/commit/801b3703e78c6d9a23b3dc684d6a33af6c0a4a42
Ignore the KSPField attribute at the top, it gets forced into symmetery counterparts and so is useless and has to be saved/loaded manually (I thought I removed the attribute...)

Edited May 1, 2016 by Crzyrndm