Deploying rubin-env 7.0.0

Data Management DM Notifications
#1

On Wednesday, 2023-07-12, we updated our build infrastructure to use version 7.0.0 of the rubin-env conda-forge metapackage. The w_2023_28 weekly will be the first tagged release to incorporate this change.

The biggest change with this new environment is the transition to Python 3.11 (RFC-932). Other packages added include fitsverify (needed for DM-39952 and DM-39953), pytest-vcr (RFC-943), rucio-clients (needed for the merge jobs in DMTN-213), and ruff 0.0.278 (RFC-939). The rubin-env-rsp environment has had jupyter-dash, and nb_black (RFC-929) added. pytest-flake8 (RFC-939) and importlib_resources (no longer needed by RFC-923 since Python 3.11’s version is now used) have been removed. Notable version updates include cfitsio 4.2.0, flake8 6.0, numpy 1.24, parsl 2023.06.12, starlink_ast 9.2.10, and allowing any wcslib 7.* version. Maximum pins on doxygen, ginga, matplotlib-base, and photutils were removed. Finally, most dependencies are now available on the aarch64 architecture.

Source and binary packages that built under rubin-env 6.0.0 will not function under rubin-env 7.0.0, as indicated by the semantic versioning.

1 Like
#2

We have discovered that rubin-env 7.0.0 downgrades some important package dependencies. I apologize for not catching this in advance. While w_2023_28 is functional for most purposes, it is experiencing problems in the Rubin Science Platform deployments. We anticipate releasing a rubin-env 7.0.1 shortly. We may be able to rebuild w_2023_28 containers with that environment.

#3

rubin-env 7.0.1 has been released (with an additional pin of pybind11 to avoid problems with its new 2.11 version). w_2023_28 has been rebuilt with this environment, and DM and RSP containers are available, as is the “shared stack” at the USDF. Any work (Python or C++) that was done with rubin-env 7.0.0 should continue to function with 7.0.1, but 7.0.0 should not be used in the future.

1 Like