Unit-facing value oddities

[bad6oy30]

I'm out-of-process, and I'm getting my target's facing value with UNIT_FIELD_R (0x8A8 in 4.0.3)

95% of the time it's accurate. But occasionally, in dungeons as far as I've observed, it seems to go off. Not by a wild amount... not in a float sense at least, in that it's still between 0 and 2pi. But degree-wise, it seems to be 180 degrees off.

It appears to stay like that until the mob moves, then it's accurate again. Does anyone use a rogue or feral druid, and encountered this?

[psyf4]

1) Clearly you have the right offset (you knew that or it wouldn't work at all).
2) I haven't observed this on mine and i'm just reading and writing to the R_OFFSET address (0x8A8 ).
3) "it's still between 0 and 2pi" -> also what I use, and I assume you confirmed that with the cases that fail? Otherwise of course there's the problem.
That being established, let's move on to things to check:

4) Are you using the correct atan function for your language? For example, in python I use something like "atan2(dx,dy)".
5) ~Works when the mob moves: did you check that there isn't some odd error in finding your / the mob's location and angle to mob? I've had problems in the past were I just zoned in and a mob (or myself) had (0,0,0). This hasn't happened to me since I redesigned my code but it could very well still happen if not handled correctly. Could it be that the mob or yourself are hidden (phased, stealthed, vanished, etc) and have a position of (0,0,0)? Given you're talking about a rogue or feral, you really should check this as they are the two classes that rely on stealth.
e.g. print your location, mob location, angle to mob and look at it for cases that fail. Or assert(location).
6) #5 is theoretical and untested; other than that we can't help you any more unless you post your code. That would explain why there has been no reply yet. In theory, there doesn't appear to be anything wrong with anything you posted and it doesn't look like anyone encountered this.

[bad6oy30]

It's not calculation results that are off... I mean, the functions are off obviously, but only because the actual value at 0x8a8 is off (I'm directly observing mine and the target mob's 0x8a.

Last night, Omor the Unscarred (Hellfire Ramparts) was doing it, and died without moving further. His corpse then continued to show the off-value, again appeared to be 180 degrees off... when I faced in his direction, my player 0x8a8 showed around 0.0, and his showed around 3.1. For many runs after that, I tested his visual orientation to values, and it appeared accurate... again, it's a relatively infrequent thing, and only seems to occur with mobs having a larger hitbox. I'm not sure if many people would have a reason to notice this, if it's common... testing a mob's facing value only really applies to rogues I think.