Now that v19.0 has been released it is time I propose that we do our annual refresh of the conda base environment. This means:
- python 3.7.6
- astropy 4.0.x
- numpy 1.17.4
- scipy 1.3.1
We will also upgrade the other packages such as pytest as defined in the scipipe_conda_env bleed definition.
I have tested lsst_distrib with the new versions and everything works fine. In testing matplotlib 3.1 some problems were encountered on macOS and therefore I propose sticking with our current 3.0.x (see
Following this update lsst_sims will need to update their copy of sncosmo to one that supports astropy4.
I also propose we update the base conda package to the current version.
I will also add the following new conda packages:
- pytest-subtest (as approved previously in
In addition I would like to add:
- boto3 and moto (so that gen3 butler tests can run properly).
- is triggered by
DM-22125 Consider upgrading pyarrow to 0.15.1 due to memory leak in 0.13
DM-22793 Strange pyvo warning about votable version
- is triggering
DM-22817 Update conda base environment
DM-23349 Build new shared stack with updated conda env
DM-23350 Update eups third party stubs for new conda env
- relates to
RFC-647 Add pytest-subtests to our standard packages
RFC-584 New Conda environment for Science Pipelines
- mentioned in
In theory we could update pyarrow as part of this if we can sort out how.
Happy to update sncosmo (I think - will double check with the most interested parties).
We should likely also update healpy.
Is this as simple as updating the "upstream tarball" files in the repos (and checking that they work .. is there a jenkins setup I should use to test?)?
The sncosmo ticket is SIM-2621 and updating the tar ball is sufficient.
Definite yes from SQuaRE, on the basis of staying fresh and also Christine Banek needs astropy 4.x for
Just remembered this but it would be very helpful if boto3 and moto could be added so that butler gen3 tests can run properly with S3.
DM-CCB agrees as of 2020-02-05.
DM-22125be relevant in this update?