We are in the process of adding SQL Server 2019 data storage and retrieval to a legacy application that was coded for Oracle.
We use varchar columns that contain ascii text and sometimes also hold non-printable ascii chars, ASCII 3 is a prime example. We use the non-printable ascii chars as a separator between data and special configuration information. All characters are in the 1-127 ascii range.
In Oracle when we query data from these fields the sorting/indexing performs as expected, sorting fields that only have text first, followed by fields with identical text + ascii control codes last.
As an example, consider this column sort from Oracle:
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…