|
|
|
|
|
|
by appplication
1006 days ago
|
|
|
Definitely good advice on pinning the upper range for major versions. Pandas 2.0 inadvertently broke spark’s toPandas for (at the time) all existing spark versions, which some of our users lean on. Pandas response to this was apparently to upgrade spark to a version that was not yet released. We lost about half a day’s work from a small handful of devs trying to identify why some of our users were seeing sudden failures. |
|
|