According to testing @jskenyon has been doing on unrelated work the equatorial to local transformations in Skyfield is at least an order of magnitude faster. Curently this is one of the remaining bottlenecks in selecting sources (working out their az el etc. -- this is slower than the actual predict which is done via JIT code).
I feel strongly about ditching astropy so that we can make the codebase as fast as possible to stitch together many days of observation for scientific analysis and classification of the system SEFD once we can pick up astronomical sources like fornax.
According to testing @jskenyon has been doing on unrelated work the equatorial to local transformations in Skyfield is at least an order of magnitude faster. Curently this is one of the remaining bottlenecks in selecting sources (working out their az el etc. -- this is slower than the actual predict which is done via JIT code).
I feel strongly about ditching astropy so that we can make the codebase as fast as possible to stitch together many days of observation for scientific analysis and classification of the system SEFD once we can pick up astronomical sources like fornax.