Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

enable cftimeindex on pangeo docker images #257

Closed
rabernat opened this issue May 16, 2018 · 9 comments · Fixed by pangeo-data/helm-chart#47
Closed

enable cftimeindex on pangeo docker images #257

rabernat opened this issue May 16, 2018 · 9 comments · Fixed by pangeo-data/helm-chart#47

Comments

@rabernat
Copy link
Member

Xarray's new support for non-standard calendars is an awesome new feature.

We should update our pangeo images as soon as possible to support this. Besides the latest version of xarray, I think we also need to install cftime from conda to enable this feature.

@jhamman
Copy link
Member

jhamman commented May 16, 2018

I think we also need to install cftime from conda to enable this feature.

If we are installing the latest version of netcdf4-python from conda-forge, we should be getting cftime as well.

So perhaps the only change needed is using xarray@master.

@rabernat
Copy link
Member Author

cftime is not importable on the current pangeo.pydata.org image.

I was able to get it by running conda install -c conda-forge cftime from the terminal.

@jhamman
Copy link
Member

jhamman commented May 16, 2018

Looking at the dockerfiles, I think just rebuilding the worker/notebook images would be enough to fix this issue.

@ocefpaf
Copy link
Contributor

ocefpaf commented May 16, 2018

I would declare netcdf4>1.4 explicitly just to be on the safe side and to be more informative about that change.

@mrocklin
Copy link
Member

FWIW so far I've preferred using defaults over conda-forge due to concerns about space (conda-forge packages were consistently more than 30% larger at the time). If people switch over it would be good to be mindful about space constraints.

Alternatively, I know that there has been some recent collaboration between the Anaconda and conda-forge build stacks. Perhaps this issue has improved since then?

@ocefpaf
Copy link
Contributor

ocefpaf commented May 16, 2018

Alternatively, I know that there has been some recent collaboration between the Anaconda and conda-forge build stacks. Perhaps this issue has improved since then?

Still slowly evolving... We are moving to the new compilers and conda-build 3 and that will make our binaries more consistent defaults. (The recipes are already ~90% consistent.)

If people switch over it would be good to be mindful about space constraints.

We should have the same flags as soon as we move the to new compilers and our binaries will be virtually the same size then, but that will take another 1-2 months to happen.

(Sorry that was a little bit off topic but people here are users of that stack so I guess it would be nice to update you on what is going on.)

PS: @jjhelmus just mention to me that cftime and netcdf4 are available on defaults 🎉 for recipe collaboration!

@rabernat
Copy link
Member Author

So the summary is that if we install netcdf4 from defaults today, we will get cftime?

I would also like to include #231 and a fix for #259 in the new image.

@rabernat
Copy link
Member Author

Fixed by #261 / #264.

@rabernat
Copy link
Member Author

I am reopening this issue because I can no longer import cftime on pangeo.pydata.org:

import cftime
---------------------------------------------------------------------------
ModuleNotFoundError                       Traceback (most recent call last)
<ipython-input-20-b8dc93698efa> in <module>()
----> 1 import cftime

ModuleNotFoundError: No module named 'cftime'

For reference

import netCDF4
print(netCDF4.__version__)
1.3.1

Somehow the docker images must have dropped this version constraint.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging a pull request may close this issue.

4 participants