search

timothydmorton / isochrones

Pythonic stellar model grid access; easy MCMC fitting of stellar properties
http://isochrones.readthedocs.org
MIT License
120 stars 62 forks source link

Installed in Fresh Environment and Nosetests Fail #129

Closed jacobhhamer closed 4 years ago

jacobhhamer commented 4 years ago

Hello,

I followed the instructions at https://isochrones.readthedocs.io/en/latest/install.html as given, and when running nosetests it fails. Additionally, when trying to fit using model.fit(), the same error repeats until the kernel breaks: "KeyError: '[(nan, -1.25, 361)] not in index'".

Attached is the output of nosetests, and a list of the packages in the environment.

nosetests_isochrones.txt environment_list.txt

timothydmorton commented 4 years ago

Thanks for reporting-- I'm guessing there is some old pandas usage in there that might cause breakages for >1.0. Do you mind checking for me if things work in a fresh env with pandas < 1.0?

jacobhhamer commented 4 years ago

Hey Tim,

With pandas 0.25.3, I get the same result from the nosetests.

timothydmorton commented 4 years ago

Strange! I'll check this out myself. You don't by any chance have any old ~/.isochrones directory lying around, do you? That's another thing that might mess things up.

timothydmorton commented 4 years ago

OK, I just made a new env, too. nosetests for me gave two errors, one expected one, and one "flaky" one. I reminded myself that in the CI I actually use pytest (https://github.com/timothydmorton/isochrones/blob/master/.travis.yml). And so I also tried running pytests, and while the flaky one still does fail for me locally (the flaky one), I do not see the errors you report. Try either deleting ~/.isochrones or creating a new env, e.g., export ISOCHRONES=~/.isochrones-test and trying again.

timothydmorton commented 4 years ago

Ah. I see you're running on Linux. I see that I had dropped the linux builds from the Travis CI, but I don't recall why. Let me try adding them back and see what I see.

timothydmorton commented 4 years ago

Oh, I recall why. They were dying because the available linux nodes didn't have sufficient RAM to run the tests.

jacobhhamer commented 4 years ago

Sorry I didn't let you know I was running Linux earlier. I have just resolved the issue. I did remove /.isochrones, but that alone did not fix it. In a fresh Anaconda install of the latest release, which was installed in a non-parallel directory as my main install, I was able to get isochrones working.

timothydmorton commented 4 years ago

Excellent, thanks for letting me know.