Skip to content

Conversation

@xylar
Copy link
Contributor

@xylar xylar commented Apr 13, 2018

Updates to many package versions for v. 1.2.0

Change to a single install script for all machines.

@xylar xylar self-assigned this Apr 13, 2018
@xylar
Copy link
Contributor Author

xylar commented Apr 13, 2018

This PR is a work in progress because this update will not be able to take place until a version of e3sm_diags and processflow are available that work with cdat v. 8.0.

@xylar xylar force-pushed the xylar/e3sm_unified_1.1.4 branch from 3426d54 to 08fb958 Compare April 13, 2018 12:00
@doutriaux1
Copy link

@xylar
Copy link
Contributor Author

xylar commented Apr 13, 2018

Sure. I need a package name, channel and version to pin (if any).

@xylar
Copy link
Contributor Author

xylar commented Apr 13, 2018

... which I can get from your link above. Sorry. Will add.

@doutriaux1
Copy link

@xylar it was in the anaconda link above 😉
pkg name: e3sm_nex
version: 0.0.1
channel: cdat

@xylar
Copy link
Contributor Author

xylar commented Apr 18, 2018

@doutriaux1, we just finally got libnetcdf 4.6.1 to build on conda-forge:
https://github.com/conda-forge/libnetcdf-feedstock
conda-forge/libnetcdf-feedstock#41

This version has an important bug fix specifically included to address big NetCDF files produced by MPAS-Ocean in E3SM at high res. But...

When I try to peg that version, I get:

  - libnetcdf==4.6.1
  - thermo==8.0 -> cdms2 -> libcf -> hdf5[version='1.8.17|1.8.17.*']
  - thermo==8.0 -> cdms2 -> libcf -> libnetcdf=4.4

I realize that libnetcdf ==4.4.* was until recently the official pinning for conda-forge and it's only just being updated to 4.5.* so maybe we're going to have trouble. But could you look into this and see if there's any way of relaxing the restrictions on cdat and its dependencies? Otherwise there's no way for us to take advantage of the bug fix :-(

cc @czender

@xylar
Copy link
Contributor Author

xylar commented Apr 24, 2018

This PR is waiting on:

It is looks unlikely that we will be able to update the version of libnetcdf to 4.6.1.

@pwolfram
Copy link

Thanks @xylar for getting in tabulate cc @vanroekel

@xylar xylar changed the title Work in progress: e3sm-unified 1.1.4 Work in progress: e3sm-unified 1.2.0 Apr 25, 2018
@xylar
Copy link
Contributor Author

xylar commented Apr 25, 2018

I decided to up the version number to 1.2.0 because we're seeing some relatively significant changes in this update. I will reserve 2.x.x for versions of e3sm-unified that are compatible with E3SM v2.

@xylar xylar force-pushed the xylar/e3sm_unified_1.1.4 branch from 7037cef to 3e7d797 Compare April 25, 2018 11:48
@xylar
Copy link
Contributor Author

xylar commented Apr 26, 2018

Okay, thanks for the help @zshaheen and @doutriaux1. I think the only thing were now waiting on is a version of processflow that is compatible with all the other changes here. (No pressure, @sterlingbaldwin.)

xylar added 5 commits May 14, 2018 12:48
Rather than a script per machine one script knows enough to handle
the machine-specific details.

Also, one base conda install can now handle multiple environments,
to be selected with "conda activate <env>" or a module file that
takes care of that step.
Also, take out some unnecessary "export"s.
@xylar xylar force-pushed the xylar/e3sm_unified_1.1.4 branch from 15889b7 to 317d531 Compare May 14, 2018 10:48
@xylar xylar changed the title Work in progress: e3sm-unified 1.2.0 Update to e3sm-unified 1.2.0 May 14, 2018
@xylar xylar merged commit 8006558 into master May 14, 2018
@xylar xylar deleted the xylar/e3sm_unified_1.1.4 branch May 14, 2018 18:35
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants