Open nchampollion opened 2 years ago
I've been observing those issues on Anaconda for a while now, and it seems like everytime I check if they are resolved, more tests fail. Meanwhile, an environment that was manually installed via pip, with all the same latest versions, works perfectly fine. As do our Docker images.
I'm once again at the end of my wits with getting Anaconda to behave, I have no idea what's wrong. It must be something Anaconda-Specific. But what specifically? I don't know.
Thanks Timo. I installed manually this morning with pip (and python 3.9.12), and it works perfectly fine without test errors.
I installed OGGM on an linux environment, and lot of tests failed. I had to make several changes especially about the tolerance in order that the tests success. I do not understand the reason why (changes in some data for example) and if I am the only one in that case.
Here is the test output
List of changes I had to do.
In test_models.py:
assert_allclose([lons[0], lats[0]], [10.7470, 46.8048], atol=1e-3)
assert_allclose([lons[-1], lats[-1]], [10.8551, 46.8376], atol=1e-3)
mu_star: 199.92
In test_prepro.py:
np.testing.assert_allclose(dfh['Zmed'], entity.Zmed, atol=25)
np.testing.assert_allclose(dfh['Zmax'], entity.Zmax, atol=25)
np.testing.assert_allclose(dfh['Zmin'], entity.Zmin, atol=25)
np.testing.assert_allclose(d['downstream_line'].length, 12, atol=2.5)
np.testing.assert_allclose(df.dx.sum(), entity['Lmax'], rtol=0.25)
self.assertTrue(utils.rmsd(h1*100*50, h2*100*50) < 2)
cfg.PARAMS['min_mu_star'] = 4
assert df['calving_rate_myr'] > 300
assert df['calving_mu_star'] >= mu_bef * frac np.testing.assert_allclose(df['calving_flux'], df['calving_law_flux'], rtol=0.5, atol=0.5)
In test_shop.py:
assert utils.rmsd(ref, mine) < 5
I did not do a pull request because I think some of my changes are not good but let me know.