When you design string manipulation functions such as SUBSTR, there could be two alternatives for length and position parameters as follows:
This approach fits current computer systems since many things such as buffer length and field length are defined in byte-counting and physically limited by byte length. However, bisection of DBCS characters may occur and you should take care of the bisected characters as explained in the following sections.
The merits and demerits of this approach are just the opposite of the byte-counting.