In the near future, we will update a number of the libraries and packages that are preloaded on notebook clusters on both FireCloud and Terra (beta). The updates are listed below. We expect all of these updates to be backwards compatible, with the exception of the upgrade from Hail 0.1 to 0.2 which is a breaking change. However it is possible that even the minor version updates could break dependencies in some of your notebooks, so we have developed a rollout plan to minimize disruption as follows.
- For FireCloud, we are taking a conservative approach and soliciting feedback from all users before scheduling the update. Please comment below if you think any of the updates listed below will be disruptive for your work. These updates only affect the creation of new clusters; existing clusters that are paused will not be affected by this change.
- For Terra (new beta UI for FireCloud), the updates listed below will be released next week (planned for Dec 4). Because these updates will affect all clusters on Terra, we recommend you save a version of your current notebook and results before that date. After the release, re-run your analysis results using the using the old vs new packages and compare results against your saved version. Let us know if you find any problematic discrepancies that you need help with.
- In the future, we plan to make it possible to enable access to prior versions of the notebook kernels to ensure long-term reproducibility and avoid this becoming a recurring issue when we roll out kernel upgrades.
Version changes for libraries and packages:
- hail 0.1 -> 0.2
- python 3.4 -> 3.6
- R 3.4 -> 3.5
- pandas 0.22.0 -> 0.23.4
- numpy 1.15.0 -> 1.15.2
- google-api-core 1.3.0 -> 1.5.0
- google-cloud-bigquery 1.4.0 -> 1.6.0
- bokeh 1.0.0 -> 1.0.1
Hail 0.2 has a new file format and cannot read Hail 0.1. You can learn more about Hail 0.1 -> 0.2 changes here.
If you have any concerns, please comment below as soon as possible. Our goal is to equip you with the tools you need to get your work done, so the more we know, the better we can serve you.
Thank you for your patience as we work to improve the platform!