Lat/Lon from subpoint

[dieli]

In the function to get lat/lon of a subpoint, the reverse_terra function is called. I have looked at the code and think that this could lead wrong values in case the arctan2 function returns a negative value (for lon).
lon = (arctan2(y, x) - 15 * DEG2RAD * gast) % tau
Replacing above line with:
lon = (((arctan2(y, x) - 15 * DEG2RAD * gast) + tau / 2.0) % tau) - tau / 2.0
should fix the issue. Output will then be +/-pi

Am I wrong here?

see dieli#1

[dieli]

The conversion to lon is the same on the ellipsoid as for the sphere. For latitude the formula as stated in https://www.celestrak.com/columns/v02n03/ is implemented. I am more concerned about the modulo operation, when the arctan2 function returns negative values.

[davidmikolas]

Okay thanks. I'll delete my comment as it's tangential (no pun intended)

[AlanMilford]

Can confirm the fix from @dieli fixes erroneous longitude values I've been experiencing.

[brandon-rhodes]

I've merged the pull request — thanks!

[brandon-rhodes]

(I'm not sure why I thought % tau was good for longitude. I must be too used to Right Ascension!)