Update environment #14
Comments
|
Outside of more packages, the biggest differences between the two are that in the python-training environment:
I think we just go with python-training wholesale, but possibly leaving out python-awips (I think so). Also, should we explicitly include pint? @dopplershift - what's your feeling on pint? |
|
If you're installing MetPy, then it's going to pull in pint. Unless you're planning on doing stand-alone teaching on Pint, I'd just leave it as a transitive dependency of MetPy. |
|
Note that if we are going to update the conda environment, I would like to know about this, sooner rather than later -- or at least not right before the workshop. The material in this repo will get automagically propagated to the user via nbgitpuller. But that is not the case for the environment. I will actually have to build a new container, push it out, and refer to it in a configuration file. So heads up there. That said, any adjustments you wish to make can be made by a PR to the Edit: I would also try to keep the environment to what is essential. That might improve kernel start time issues. |
|
Julien - we will need to update the environment. I will get a PR in with that asap. |
|
Closed by #15 -- edit -- |
|
@julienchastang One environment-related question I had: are the current installed versions of the packages what will be present on Sunday for the workshop, or will they be updated beforehand? My notebook (#19) has some version dependence between the current MetPy v0.12 and upcoming v1.0. |
|
Below is the latest and greatest. Metpy is at |
|
Sounds obvious, but it is worth mentioning: make sure your committed notebooks work in http://pyaos-workshop.unidata.ucar.edu/. |
|
@jthielen 1.0 won’t be available for the workshop. If you have a critical fix I can cut 0.12.1, but that’s not planned at the moment. |
|
@julienchastang @dopplershift Thanks, that's what I needed to know! No critical fixes, I just wanted to know what level of "xarray compatibility" I should target in the notebook. I'll be sure to include the more manual broadcasting and type casting required with the current version. |
|
Although, @dopplershift if you don't mind me asking, when is the current target for 1.0? Knowing that may help re-prioritize my current work here and in MetPy (xarray compat in kinematics Unidata/MetPy#1260 / Unidata/MetPy#893 (comment), xarray compat in general functions Unidata/MetPy#1223, coordinate and projection helpers, etc.). |
|
Hard pencils down time is Monday night (ahem early Tuesday morning). |
We should update our environment to mimic what is in the python-training repo. See https://github.com/Unidata/python-training/blob/master/environment.yml
The text was updated successfully, but these errors were encountered: