1.2RC1 - test failure in astropy.time.tests.test_ut1.TestTimeUT1

[saimn]

Not sure if it is related to #5098 so I opened another issue.

________________________________________ TestTimeUT1.test_utc_to_ut1 _________________________________________

self = <astropy.time.tests.test_ut1.TestTimeUT1 instance at 0x7ff28f8325f0>

    def test_utc_to_ut1(self):
        "Test conversion of UTC to UT1, making sure to include a leap second"""
        t = Time(['2012-06-30 12:00:00', '2012-06-30 23:59:59',
                  '2012-06-30 23:59:60', '2012-07-01 00:00:00',
                  '2012-07-01 12:00:00'], scale='utc')
        t_ut1_jd = t.ut1.jd
        t_comp = np.array([2456108.9999932079,
                           2456109.4999816339,
                           2456109.4999932083,
                           2456109.5000047823,
                           2456110.0000047833])
        assert allclose_jd(t_ut1_jd, t_comp)
        t_back = t.ut1.utc
        assert allclose_jd(t.jd, t_back.jd)

        tnow = Time.now()

        # With internet off (which is the default for testing) then this will
        # fall back to the bundled IERS-B table and raise an exception.  But when
        # testing with --remote-data the IERS_Auto class will get the latest IERS-A
        # and this works.
        if INTERNET_OFF:
            with pytest.raises(IndexError):
                tnow.ut1
        else:
>           tnow.ut1

astropy/time/tests/test_ut1.py:51: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
astropy/time/core.py:1138: in __getattr__
    tm._set_scale(attr)
astropy/time/core.py:425: in _set_scale
    args.append(get_dt(jd1, jd2))
astropy/time/core.py:1260: in _get_delta_ut1_utc
    delta = iers_table.ut1_utc(jd1, jd2)
astropy/utils/iers/iers.py:215: in ut1_utc
    self.ut1_utc_source if return_status else None)
astropy/utils/iers/iers.py:312: in _interpolate
    self._check_interpolate_indices(i1, i, np.max(mjd))
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

self = <IERS_B length=19840>
 year month  day    MJD   ... e_UT1_UTC  e_LOD   e_dX_20... 2.8e-05
 2016     4    26 57504.0 ...   6.1e-06 1.95e-05      3e-05      3e-05
indices_orig = array([19839]), indices_clipped = array([19840]), max_input_mjd = 57554.0

    def _check_interpolate_indices(self, indices_orig, indices_clipped, max_input_mjd):
        """
            Check that the indices from interpolation match those after clipping
            to the valid table range.  This method gets overridden in the IERS_Auto
            class because it has different requirements.
            """
        if np.any(indices_orig != indices_clipped):
>           raise IERSRangeError('(some) times are outside of range covered '
                                 'by IERS table.')
E           IERSRangeError: (some) times are outside of range covered by IERS table.

astropy/utils/iers/iers.py:255: IERSRangeError
__________________________________ TestTimeUT1_IERS_Auto.test_ut1_iers_auto __________________________________

self = <astropy.time.tests.test_ut1.TestTimeUT1_IERS_Auto instance at 0x7ff2797c3bd8>

    def test_ut1_iers_auto(self):
        tnow = Time.now()
        iers_a = iers.IERS_Auto.open()
        tnow.delta_ut1_utc, status = iers_a.ut1_utc(tnow, return_status=True)
>       assert status == iers.FROM_IERS_A_PREDICTION
E       assert -2 == 2
E        +  where 2 = iers.FROM_IERS_A_PREDICTION

astropy/time/tests/test_ut1.py:97: AssertionError
===================== 4 failed, 11215 passed, 74 skipped, 42 xfailed in 2584.95 seconds ======================

[taldcroft]

What configuration? --remote-data?

FWIW I cannot reproduce this or #5098 on mac with Py2.7 for remote and not remote data.

[astrofrog]

Could it be a cache issue with the previously downloaded tables?

[saimn]

@taldcroft - Sorry, forgot to put some context indeed. Python 3.5, remote-data, full log here:
https://gist.github.com/saimn/98e8ca025b9ac8d09a22e3ff71688598

[saimn]

Cache content (seems unrelated ?):

❯ l ~/.astropy/cache/download/*
-rw------- 1 simon 1,6M 23 nov.   2015 /home/simon/.astropy/cache/download/2c9202ae878ecfcb60878ceb63837f5f

/home/simon/.astropy/cache/download/py2:
total 1,6M
-rw------- 1 simon 1,6M  3 juin  10:28 2c9202ae878ecfcb60878ceb63837f5f
-rw-r--r-- 1 simon   72  3 juin  11:01 urlmap.bak
-rw-r--r-- 1 simon   82  3 juin  10:28 urlmap.dat
-rw-r--r-- 1 simon   72  3 juin  11:01 urlmap.dir

/home/simon/.astropy/cache/download/py3:
total 1,6M
-rw------- 1 simon 1,6M  3 juin  10:54 2c9202ae878ecfcb60878ceb63837f5f
-rw-r--r-- 1 simon   72  3 juin  10:54 urlmap.bak
-rw-r--r-- 1 simon   82  3 juin  10:54 urlmap.dat
-rw-r--r-- 1 simon   72  3 juin  10:54 urlmap.dir

~/dev/astropy tox
❯ cat ~/.astropy/cache/download/py3/urlmap.bak
'http://data.astropy.org/tutorials/FITS-images/HorseHead.fits', (0, 82)

[eteq]

In theory the cache shouldn't matter, because the tests run in an environment that ignores the local cached copy.

@saimn - have you tried boosting the timeout as per what the thought in #5098? If that fixes it we know it's a download issue...

[astrofrog]

@saimn - can you confirm that this is fixed by #5111, which is now merged?

[saimn]

Fixed with #5111 !