Closed jbergeskans closed 2 months ago
@jbergeskans why would you list the columns if you do not want to explicitly set them? You can leave the columns array blank and achieve the same result?
@jbergeskans why would you list the columns if you do not want to explicitly set them? You can leave the columns array blank and achieve the same result?
We want to use other documentation features such as description, tests, and constraints. Basically, this allows us to omit the data type field which, when you're using parquet files, isn't needed anyway.
We want to use other documentation features such as description, tests, and constraints. Basically, this allows us to omit the data type field which, when you're using parquet files, isn't needed anyway.
For me this feels very conflicting, you want to infer the schema automatically but do want to manually add the column names for documentation.
I see two problems with this setup:
Description & motivation
Fixes #249 to support schema inference on BigQuery.
Currently, if the column names are defined in the sources file but the data types are not, the following SQL will be generated:
In order for BigQuery to infer the schema, the column names and data types needs to be omitted:
This has been achieved by introducing the variable
infer_schema
. When set to true, it will not iterate over the list of columns. Example source fileChecklist