In arrow_array_scan.cpp we handle many common data types. But not enough to handle common cases when dealing with snowflake. One user report:
but problematic data types seem to be in Snowflake number format NUMBER(p,s) withp > 2
i.e. NUMBER(7,2), NUMBER(5,2), NUMBER(15,2) etc. If I recast those columns to FLOAT it actually works now
#641 added diagnostics so we can get info on what's missing.
One way to handle this is to look at DuckDB and ClickHouse arrow metadata handling.
In
arrow_array_scan.cppwe handle many common data types. But not enough to handle common cases when dealing with snowflake. One user report:#641 added diagnostics so we can get info on what's missing.
One way to handle this is to look at DuckDB and ClickHouse arrow metadata handling.