Closed jaycedowell closed 5 months ago
An interesting problem has popped up here. The version of astropy that work with Py3.6 works a little different from those for Py3.8/3.10/3.12 with regards to coordinate transforms to/from the ITRS frame. If I can't sort out how to fix things for Py3.6 I'll have to drop support for it.
I think what is happening is that for older astropy tranforming from ITRS -> AltAz has to go through CIRS, which introduces things like stellar aberration which shouldn't be there. In newer version of astropy there is a concept of a "topocentric ITRS" frame which can be directly transformed to AltAz without those corrections.
To fix this for Py3.6 we need to backport the relevant portions of the direct ITRS -> AltAz transform into LSL and use those when necessary. The thing that is holding me up is I'm not sure how many people are still on Py3.6. I was hoping telemetry would tell me that but it looks like it only records LSL version info, not Python version info. I should fix that.
The best thing is probably to drop support for Py3.6. Ubuntu 20.04 LTS ships with Py3.8 by default and that came out in April of 2020.
This PR drops Python2 support for #53. It also moves more of the internal astronomical calculations uses in LSL from pyephem to astropy.