Visportal not closing from one direction

[Springheel]

Anyone have any thoughts about what might make a visportal close properly when the door is shut, but only from one side?

 

When I'm on the west side the door, the visportal functions properly, closing when the door is closed.

 

When I'm on the east side of the door, the visportal stays open when the door is closed.

 

The visportal is properly buried in the door, as far as I can tell.

 

I don't think I've ever encountered a door visportal that works only from one side.

 

[stumpy]

something else goes through visportal on side where it stays open, something you can't visible see. brush split when dmap makes proc file.

[Goldwell]

Another thing it could be is that a visportal elsewhere is opening because of the angle you're looking at (which closes when you're facing away from it on the otherside of the door).

[Obsttorte]

@Goldwell: In that case the visportal would be displayed yellow.

 

@Springheel: Did you check for overdraw using r_showtris 2 or 3? Maybe the vp's status is just shown wrong.

[demagogue]

My idea is that the side it's ok is internally sealed and the side that's bugged has an internal leak (presumably to a third leaf). So I'd test for an internal leak.

 

Edit. Hmm, sorry, the leak would have to be across the vp boundary to kill the vp though. Now I'm thinking something in the rendering is opening an internal leak just from one side to the other but not vice versa, eg, in via a leaky area not rendered at all when you're on the good side. Not sure if that's possible, but still a reason to test for an internal leak on the bad side.

 

Edit2. Ugh, sorry again, wires crossed. You're saying it's just not closing by the door but it's otherwise valid. That sounds like to do with a quirk in the code that checks the vp is inside a closed door, and I'd double check what that code is literally doing before speculating, because it's hard to imagine what check like that would notice which way you're looking at it.

 

But for the record, strictly speaking I bet each side is its own vp for each leaf made during the dmap, so they aren't strictly the "same" vp. So that opens up possibilities like a rounding quirk may have been applied for its construction on one side but not the other.