timescale object works with .UTC but not with .UT1

[aendie]

I would like to confirm if this is a skyfield issue or am I doing something fundamentally illegal?

In both Python 2 and 3, I can reproduce an error on line 14 "t3 = ts.ut1(.......)" although line 12 "t2 = ts.utc(.......)" works.

Python 2 or 3 code (just change the first line to .../env python3):

#!/usr/bin/env python2
# -*- coding: UTF-8 -*-

import datetime
from skyfield.api import load
from pytz import timezone
utc = timezone('UTC')

ts = load.timescale()   # timescale object
PreviousNewMoon = datetime.datetime(2022, 11, 23, 22, 57, 13, 903886, tzinfo=utc)

t2 = ts.utc(PreviousNewMoon + datetime.timedelta(days=28))
print("t2 = " + str(t2))
t3 = ts.ut1(PreviousNewMoon + datetime.timedelta(days=28))
print("t3 = " + str(t3))

Python 2 hang:

Python 3 hang:

P.S. I forgot to mention that the Python 2 test was with Skyfield 1.30 and the Python 3 test was with Skyfield 1.37.

[Bernmeister]

I took a quick look at the documentation and it looks as if you have to pass the date components for UT1.

[brandon-rhodes]

I took a quick look at the documentation and it looks as if you have to pass the date components for UT1.

Yes, that's correct, @Bernmeister — thanks for the helpful link to the documentation!

Python datetime objects don't support UT1, so far as I know, but only UTC-derived time zones, so the problem of folks submitting UT1 date components as a datetime has never come up. If you do need to use the components of a datetime as though they were UT1, then simply pass the components dt.year, dt.month, and so forth as the arguments to ts.ut1().

Let me know if you run into any snags.

[aendie]

Ah ... the golden rule "READ THE DOCS". Thanks - I naturally assumed that ts.utc() and ts.ut1() were interchangeable.

The following code works in Python 3 and with a minor exception in Python 2:

import datetime
from skyfield.api import load, utc

ts = load.timescale()   # timescale object
PreviousNewMoon = datetime.datetime(2022, 11, 23, 22, 57, 13, 903886, tzinfo=utc)

t2 = ts.utc(PreviousNewMoon + datetime.timedelta(days=28))
print("t2 = " + str(t2))

dt3 = PreviousNewMoon + datetime.timedelta(days=28)
sec = dt3.second + dt3.microsecond/1000000.
t3 = ts.ut1(dt3.year, dt3.month, dt3.day, dt3.hour, dt3.minute, sec)
print("t3 = " + str(t3))

Result in Python 2 with Skyfield 1.30:

Result in Python 3 with Skyfield 1.37:

The minor exception in Python 2 is that you have to use sec = dt3.second + int(dt3.microsecond)/1000000. because dt3.microsecond returns a string in Python 2(.7.18)

One does not have to apply .replace(tzinfo=None) to dt3. HOWEVER: This is only correct when the datetime is in timezone 'utc'. For example: if it is in timezone 'est' even .replace(tzinfo=None) does not give the correct answer (of course). A time in another timezone has to be converted to timezone 'utc' first.

And ts.ut1() does not require the last argument , jd=None apparently.

Remember, read the ...

[brandon-rhodes]

I'll expand the documentation of those methods in a few days to help make the various options clearer — in particular, to explain why there's an unused jd argument dangling at the end of ut1()!

[brandon-rhodes]

There, I've improved the docs to hopefully prevent ambiguity in the future. The online version should update the next time I do a release!