Free-form coding and data analysis within Data Platform Jupyter environments (as well as many Python-based Spark jobs?) must happen within Conda-Analytics-based Conda environments.

However, many of the included packages are pinned at specific versions required by out-of-environment dependencies (look for the "#pinned" comment in the base environment file). As a result, many packages are very outdated and can't be updated (some because they're directly pinned and others because they depend on a pinned package) by the individual user. Some of these limitations depend on major platform upgrades (e.g. T338057) but others might be easier to deal with (e.g. upgrading Jupyter/JupyterLab or Python).