mirror of
https://github.com/discourse/discourse.git
synced 2025-02-06 11:28:18 +00:00
bee68bba2e
We use the `id` of the upload to calculate a `depth` partition in the filename. This test would fail if your database had a higher seed because the depth it was looking for was hard coded to 1. The solution was to not save the records (which is faster anyway) and specify the `id` of the upload to make the hash deterministic.