Allowable characters in attribute / field names in jsonschema should be compatible with allowable values in parquet file format
As time progresses, more and more of the representation of data will be in hierarchical, efficient and binary file formats such as parquet, orc and avro along with lakehouse formats like delta, iceberg and hudi.
There are several restrictions on allowable column / field names in these file formats. The json-schemas in ASM should align with allowable column names in these file format. Doing otherwise will require downstream consumers to create yet another mapping from ASM -> Parquet / ORC / AVRO.
These characters are disallowed in Parquet:
[,;{}()\n\t=] as also whitespaces which are particularly prevalent in ASM jsonschemas.