Deploying rubin-env 1.0.0

On Tuesday, 2022-01-25, we updated our build infrastructure to use version 1.0.0 of the rubin-env conda-forge metapackage. The d_2022_01_26 nightly and w_2022_05 weekly will be the first tagged releases to incorporate this change.

This new environment adds several packages to support Spectractor as requested in RFC-800 and prmon to support monitoring of PanDA production batch jobs. In addition, version pins on click, galsim, networkx, moto, pandas, scikit-image, and sqlite have been released, as the Science Pipelines code is now compatible with newer versions (see RFC-821). Finally, minimum versions have been required for certain iDDS components used by PanDA.

rubin-env 1.0.0 should be used with all Science Pipelines code on or after d_2022_01_26 and w_2022_05. As indicated by its major version number increment, this cannot be used with most older versions of the Science Pipelines; it requires DM-33347 and thus code after d_2022_01_21. For older code, please use rubin-env 0.7.0.

1 Like

For people using NCSA resources, I have updated the shared stack to use rubin-env 1.0.0. It currently has weekly 4 in it. If you need access to the old shared stack, it can be found at /software/lsstsw/stack_20210813/

It appears that the newest version of moto is not yet currently available at conda-forge. This means that for the time being, installations of rubin-env 1.0.0 for the time will enable older versions of the Science Pipelines to be installed successfully, hence w_2022_04 in the shared stack above. But eventually, when moto version 3 is available, new installations of rubin-env 1.0.0 will break older Science Pipelines versions.

There was a problem with the 1.0.0 environment that caused it to obtain outdated (7.5.0) C, C++, and Fortran compilers under some circumstances. This has been fixed in build 1 of the environment. No Science Pipelines tagged versions should have been affected by this.

A convenience package that provides user-friendly links to the C, C++, and Fortran compilers was inadvertently left out of build 1 and has now been added into build 2. Only d_2022_01_26 will have been affected by this.