I had a version of this happen to me at my first job a few months out of school. I was supposed to write code to read IMU data from a flying sensor. The sensor was giving me a stream of pitch, roll, and yaw numbers, and I had to use this information to convert sensor measurements into absolute coordinates in lat, lon, and altitude. As is so often the case, there was no documentation on the data format coming from the IMU unit. I had no idea whether the numbers were in radians, degrees, gradians, or some other weird format. I also didn't know what order the pitch, roll, and yaw transformations should be applied in.<p>So in a team meeting I asked about the transformation order. Now, pitch, roll, and yaw transformations are not commutative. For the small angles that you're likely to see in typical flight, they're <i>almost</i> commutative, but of course we were trying to do better than "almost". But when I asked about the transformation order, someone answered that it didn't matter. These were actually some technical people who answered, so it wasn't even management talking ignorantly. I politely said that it's a mathematical fact that order does matter when applying pitch, roll, and yaw transformations. The guy responds, "Oh no, the IMU measurements represent a snapshot in time." I responded that in that case, the numbers carry with them an implied transformation order, and that I need to know what it is in order to write my software correctly. "No, I'm pretty sure it doesn't matter," he said. Now I'm getting desperate, so I bust out with the hand demonstrations. "Start here, pitch up 90 degrees, then yaw right 90 degrees. Now compare that to yaw right 90 degrees, then pitch up 90 degrees. See? We started from the same orientation, but we get different results." These guys weren't having any of it. They insisted that it didn't matter, so I dropped the issue. Later I brought it up again at another meeting and a PhD who wasn't in the first meeting says, "Oh, I know what you're talking about." Ahhhh, finally, vindication.<p>That didn't even help me all that much though, because nobody seemed to have information about the actual ordering. I think in the end I had to reverse engineer it by brute forcing every possible combination of units and orderings until I found one that made the data look right. And oh by the way, the units weren't in degrees, radians, OR gradians. They were in units of semicircles, where 1 = 180 degrees.