Specifying the SCANSTRINGS=TRUE option might not reduce the length of Snowflake character columns when the following are true:

• You are using a CASLIB defined with srctype='snowflake'.
• You are using the CASUTIL procedure, the FEDSQL procedure, or CAS procedure to read the Snowflake table into SAS Cloud Analytic Services (CAS).
• You have specified the SCANSTRINGS=TRUE option.
• Either the first column in the Snowflake table if you are reading in all columns or the first column that you have selected is not a character column.

The variables in the CAS table will be the same length as defined in the Snowflake database or ~16.8 million if the variable was created without a defined length.

The only workaround for this issue is to select the variables in a different order than how they are defined in the database so that a character column is selected first.