Set camera mask for new elements

[jonyrock]

Hi

It is not obvious that I should write every time such thing

auto sprite = //<some new 3d spite>
sprite->setCameraMask(getCameraMask());
addChild(sprite);

I think this behaviour should be by default

[zilongshanren]

@super626
Could you take a look at this issue?

[stevetranby]

We just had this discussion come up again. We think that the sprite should get added with the parent node's camera mask. Maybe have a flag as whether to cascade or not. I would guess there's only a few, or none, instances where someone would want to addChild(node) while setting a different camera mask than it's parent.

I can submit a PR, but wanted to see if there's any discussion first.

// in method => void Node::addChildHelper(...) { 
child->setParent(this);
child->setCameraMask(getCameraMask()); // ADDED

http://discuss.cocos2d-x.org/t/solved-camera-not-applied-to-newly-created-object-sprite/29323/

[minggo]

I think we need it. @ricardoquesada what's your opinion?

[ricardoquesada]

yes, sounds good.

[minggo]

@stevetranby will you please send a PR?

[stevetranby]

Sure, been out on vacation. Will submit by Monday.

[minggo]

thanks @stevetranby

[stevetranby]

#16094

[ricardoquesada]

@stevetranby this patch broke the Scene3dTest. No nodes are shown on that test (cpp-tests -> Scene3dTest)

Any idea how improve this patch so that it works with the Scene3dtest ?

Perhaps this patch if fixing a real bug, but I'm worried that perhaps some games are relying on this bug... like Scene3dtest.

@minggo any idea what should we do?

Thanks.

[stevetranby]

@ricardoquesada ah, interesting, I wonder if just testing if the child's mask is currently set? I presume it's camera default at construction, so maybe if default then set as parent? I'm sure that's still wrong, with respect to anything similar to Scene3DTest where default camera may be used in the same way as the other cameras.

Maybe it's the wrong idea for a default as well, though we've found ourselves setting it explicitly in many places until we figured out we really just want addChild and setParent to set the same mask.

I'll take a look at that test and see if it sheds any light on this.

Prob smart to just revert for now?

[stevetranby]

Yeah, being reminded that the set mask cascades it should probably at minimum send in false for the optional cascade param. I'll continue to look through that test to understand why child is setting different camera mask than its parent.