The sqlite versions of variability truth tables each contain a column MJD with double precision values but the schema declaration gives MJD a datatype FLOAT. When these tables were translated to parquet (via script) and ingested into PostgreSQL (partially manual procedure), FLOAT was translated to the equivalent 32-bit datatype, and so precision was lost in the translated values.
The simplest way to fix this is to regenerate the tables, using a corrected schema. In order to use the parquet script as is one would need to first fix the type declaration in the source sqlite file. This problem affects tables
The sqlite versions of variability truth tables each contain a column MJD with double precision values but the schema declaration gives MJD a datatype FLOAT. When these tables were translated to parquet (via script) and ingested into PostgreSQL (partially manual procedure), FLOAT was translated to the equivalent 32-bit datatype, and so precision was lost in the translated values.
The simplest way to fix this is to regenerate the tables, using a corrected schema. In order to use the parquet script as is one would need to first fix the type declaration in the source sqlite file. This problem affects tables