[OLD] Parabolic orbits: final decision - poliastro/poliastro GitHub Wiki

This didn't age well! We ended up going back to p and adding a .parabolic classmethod to Orbit

Three criteria:

• Convenient in the majority of use cases
• Fail proof
• Easy to implement

Solutions:

• If using only $a$, it's convenient in the majority of cases and easy to implement, but needs a ValueError when the eccentricity is one and the parabolic method does not yield a == np.inf * u.km.
• If using only $p$, it's fail proof and easy to implement, but it's a pain in the neck to use in the majority of cases where $a$ is given and one needs a explicit conversion.
• If allowing any of those, it's fail proof and convenient in all cases, but the implementation is harder and more difficult to maintain.

As true parabolic orbits are extremely rare, I'm favoring the first option and this is final. Fixed in https://github.com/poliastro/poliastro/pull/106.