skyfield.api.Angle arithmetic ?

[captcha1]

Shouldn't skyfield.api.Angle have arithmetic methods (addition, subtraction) ...?

Eg, I'm computing altitude w/ altaz, and want to subtract semi-diameter. Is there an easy way to do this ?

[brandon-rhodes]

We are not currently planning to add math operations to Angle, Distance, or Velocity, because full math support involves dozens of methods having to be added to every one of those classes. Instead, we encourage programmers to use specific units like .degrees, which return Python floating point objects (or NumPy vectors), both of which already support the full range of Python math operations.

Can you do something like alt.degrees - semi_diameter_degrees?

[captcha1]

Mmm. That works.

Is there an easy way to format an angle with "dd mm.m" (common in celestial navigation) ?

I'm no python wizard, but couldn't you define arithmetic in "class Unit" since Angle, etc inherits from Unit ...?

[brandon-rhodes]

Is there an easy way to format an angle with "dd mm.m" (common in celestial navigation) ?

Yes! Here's the class documentation:

https://rhodesmill.org/skyfield/api-units.html

(Let me know if the descriptions are too obscure and I can try expanding them.)

I'm no python wizard, but couldn't you define arithmetic in "class Unit" since Angle, etc inherits from Unit ...?

Yes, if a level of indirection was introduced: the arithmetic would have to access a different attribute (au vs au_per_d versus radians) for every subclass of Unit. All of the resulting complexity would not only laden Skyfield with extra code, but would make something as basic as arithmetic even slower than with per-class dunder methods. So I'm still at the moment convinced that it's better for folks to do their floating-point and NumPy vector operations manually, rather than hiding them behind Skyfield operators.

But, yes — if the code were going to be written, adding it once to Unit would be far cleaner than adding it n times, once to each subclass! Good point.

[captcha1]

I saw the docs. I see "dd mm ss.s" format but I don't see "dd mm.m" ...

[brandon-rhodes]

Oh — I didn't read your question carefully enough, did I? No: there isn't any built-in way to stop at decimal minutes. The next time I'm in that code, I'll think about whether there's an easy way to add support for that use case.