apache/airflow

Hash Virtual Environment Cache Based on Actual Package Versions in `PythonVirtualEnvOperator`

Open

#41.328 geöffnet am 8. Aug. 2024

Auf GitHub ansehen
 (10 Kommentare) (0 Reaktionen) (0 zugewiesene Personen)Python (16.781 Forks)batch import
area:coregood first issuekind:featureprovider:standard

Repository-Metriken

Stars
 (44.809 Stars)
PR-Merge-Metriken
 (Durchschn. Merge 7T 18h) (834 gemergte PRs in 30 T)

Beschreibung

Description

Update the PythonVirtualEnvOperator to hash the virtual environment cache based on the actual versions of the installed packages rather than just the checksum of the requirements. This change would ensure that the cache reflects the true state of the environment, avoiding issues with packages tagged as "latest" or other dynamic versioning.

Use case/motivation

Currently, when using the PythonVirtualEnvOperator, if dependencies in the requirements use tags like "latest", the checksum used for caching remains unchanged even if the package versions are updated. This can lead to situations where outdated versions of packages are used from the cache, causing potential inconsistencies and issues in workflows. By hashing the cache based on the actual versions of installed packages, the virtual environment would be refreshed appropriately whenever package versions change, ensuring that the most current versions are used.

Related issues

No response

Are you willing to submit a PR?

  • Yes I am willing to submit a PR!

Code of Conduct

Contributor Guide