Closed jeanconn closed 11 months ago
This looks to me to be tiny differences in the archived and calculated NSM quaternion on day 2021:296. While this doesn't look to be a problem for every platform (and passes on ska3), perhaps the test should use an isclose for for that attitude anyway?
Can you run on kady? I am would rather not make this another blocker to the release.
OK. I set up a ska3-prime on kadi and this test did pass there. I'll update the top and set this low priority.
As an RFE it seems to me that NSM quaternions are a different thing from the rest of the quaternions and I don't know how I would pull those derived attitudes out easily. If you are also considering adding "informational" quaternions that would represent the estimated attitude during NSM recovery, perhaps different kinds of attitude "labels" would make sense anyway?
As an RFE it seems to me that NSM quaternions are a different thing from the rest of the quaternions
They are different only in that they may be inaccurate, mostly in rotation about the sun line (RASL). Given the way that these get used operationally, I'm not quite seeing the use case of your suggestions.
Fair. I suppose I was thinking of it mostly for this debug use case; if a mismatch comes up again in a non validation context it took a minute to figure out this could be a numeric issue because this quaternion was not just read from a file or table. And to make the regression go away it seems like an exception for the not accurate quaternions would be the thing?
Closed by #288
I'm seeing a test fail on ska3-prime with fido on test_commands_create_archive_regress.
This passes on ska3 on fido and on ska3-prime (2023.1rc9) on kady.