You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The belief horizon column takes 16 bytes, but could also store minutes/seconds as integer, which would reduce 12 bytes per belief, which certainly a fifth of all our data size.
We need to find out how to migrate this (in the code/ORM data model, which should lead to an alembic migration).
Then, we need to know how timely-beliefs can translate this from and into a timedelta:
in input, we need to translate timedelta to int when a BeliefsDataFramw is created and in querying - query logic should still work the same, we only need to translate timedeltas to integer if a belief horizon parameter is given
in its output, it would be best to only translate (to timedelta) when someone accesses this column in the BeliefsDataFrame
The text was updated successfully, but these errors were encountered:
nhoening
changed the title
Change belief horizon column type
Change belief horizon column type to int to save space
Jun 7, 2024
The belief horizon column takes 16 bytes, but could also store minutes/seconds as integer, which would reduce 12 bytes per belief, which certainly a fifth of all our data size.
We need to find out how to migrate this (in the code/ORM data model, which should lead to an alembic migration).
Then, we need to know how timely-beliefs can translate this from and into a timedelta:
The text was updated successfully, but these errors were encountered: