Update to v1.2.5 - Githubissues

E3SM-Project / e3sm-unified

A metapackage for a unified anaconda environment for analyzing results from the Energy Exascale Earth System Model (E3SM).

BSD 3-Clause "New" or "Revised" License

8 stars 8 forks source link

Update to v1.2.5 #39

Closed xylar closed 5 years ago

xylar commented 5 years ago

Switches to using a meta.yaml build, which allows different dependencies for python 2 and python 3, as well as for OSX vs. linux.

Update versions of main packages to:

cdat 8.1
e3sm_diags 1.6.1
mpas-analysis 1.2.3 (new name on conda-forge)
nco 4.7.9
ilamb 2.3.1 (py2.7), 2.4 (py3.7) (added)
processflow 2.2.0
zstash 0.2.0 (py2.7 only)

Updated dependencies to:

cdat_info 8.1.1
distarray 2.12.2
cdms2 3.1.2
cdtime 3.1.2
cdutil 8.1
genutil 8.1.1
vtk-cdat 8.2.0rc2.8.1
dv3d 8.1
vcs 8.1
vcsaddons 8.1
wk 8.1
thermo 8.1
e3sm_nex 1.0.0
xarray 0.11.3
dask 1.1.3
nb_conda_kernels (added)
libnetcdf 4.6.2
hdf5 1.10.4
eofs 1.3.1
pywavelets

Fixes an issue with the csh version of the conda environment on E3SM supported machines by explicitly setting an environment variable needed by basemap.

closes #36

xylar commented 5 years ago

Not ready to merge because

zstash and mpas_analysis versions are not yet released
can't build with currently pinned cdms2, see #38

xylar commented 5 years ago

Almost there. Waiting for https://github.com/E3SM-Project/e3sm_diags/pull/203 and the subsequent build to be uploaded to the e3sm channel.

xylar commented 5 years ago

For my reference, here's the build command I used:

conda build --python=2.7 -c conda-forge -c e3sm -c cdat/label/v81 -c defaults meta.yaml

doutriaux1 commented 5 years ago

@xylar I would remove the -c cdat I'm trying to get away from this and tag/label each release.

xylar commented 5 years ago

@doutriaux1, that sounds good! All I would need for me to be able to drop the cdat channel is for you to add cdat/label/v81 to e3sm_nex.

doutriaux1 commented 5 years ago

Done!

xylar commented 5 years ago

I appreciate it!

xylar commented 5 years ago

@doutriaux1, I'm getting a ClobberWarning from libuuid and ossuuid as well as jupyter and jupyter_core:

ClobberWarning: This transaction has incompatible packages due to a shared path.
  packages: conda-forge::libuuid-2.32.1-h14c3975_1000, conda-forge::ossuuid-1.6.2-hf484d3e_1000
  path: 'lib/libuuid.a'

ClobberWarning: This transaction has incompatible packages due to a shared path.
  packages: conda-forge::libuuid-2.32.1-h14c3975_1000, conda-forge::ossuuid-1.6.2-hf484d3e_1000
  path: 'lib/libuuid.la'

ClobberWarning: This transaction has incompatible packages due to a shared path.
  packages: conda-forge::libuuid-2.32.1-h14c3975_1000, conda-forge::ossuuid-1.6.2-hf484d3e_1000
  path: 'lib/libuuid.so'

ClobberWarning: This transaction has incompatible packages due to a shared path.
  packages: conda-forge::libuuid-2.32.1-h14c3975_1000, conda-forge::ossuuid-1.6.2-hf484d3e_1000
  path: 'lib/pkgconfig/uuid.pc'

ClobberWarning: This transaction has incompatible packages due to a shared path.
  packages: conda-forge::jupyter_core-4.4.0-py_0, conda-forge::jupyter-1.0.0-py_1
  path: 'lib/python2.7/site-packages/jupyter.py'

ClobberWarning: This transaction has incompatible packages due to a shared path.
  packages: conda-forge::jupyter_core-4.4.0-py_0, conda-forge::jupyter-1.0.0-py_1
  path: 'lib/python2.7/site-packages/jupyter.pyc'

Is that anything you know something about? I tried to find out more about ossuuid but its feedstock points to an FTP URL that is either down or doesn't exist. Could it be one of your packages that brings in ossuuid? If so, what about using libuuid instead or doing something to make sure the two don't conflict?

doutriaux1 commented 5 years ago

@xylar are you bringing in cmor? I think cmor needs ossuuid. @dnadeau4 is cdms/libcdms using ossuuid? Or is it just cmor. I think it's just CMOR. If it's just CMOR then @mauzey1 could you please take a look into switching to libuuid? Thanks!

xylar commented 5 years ago

We will be bringing in CMOR thought processflow but that's not currently the case. So it must be another package. I'll try to track it down.

xylar commented 5 years ago

@doutriaux1, the dependency on ossuuid comes through libcf. Is that one of yours?

xylar commented 5 years ago

https://github.com/conda-forge/libcf-feedstock/blob/master/recipe/meta.yaml#L32

xylar commented 5 years ago

The jupyter/jupyter_core has already been reported: https://github.com/conda-forge/jupyter-feedstock/issues/4

Maybe not a critical problem but could break jupyter. The proposed solution of installing jupyter from defaults isn't practical for us because it is a dependency of both cdat and an explicit dependency of e3sm-unified.

doutriaux1 commented 5 years ago

yes. we'll try to update this in nightly

xylar commented 5 years ago

The jupyter issue should be resolved by https://github.com/conda-forge/jupyter-feedstock/pull/8