unicode vs non-unicode ODBC driver

Consider a simple scenario.

CREATE DATABASE "TEST1"
WITH OWNER = postgres
ENCODING = 'WIN1252'
LC_COLLATE = 'English, United States'
LC_CTYPE = 'English, United States'
CONNECTION LIMIT = -1;

CREATE TABLE z
(
c1 character(6)
)

If you invoke sqlColumns the data type returned from the column changes
depending on if you used the UNICODE or non-UNICODE driver. This is with the
8.04.01 driver.

Get Data All:
"TABLE_QUALIFIER", "TABLE_OWNER", "TABLE_NAME", "COLUMN_NAME", "DATA_TYPE",
"TYPE_NAME", "PRECISION", "LENGTH", "SCALE", "RADIX", "NULLABLE", "REMARKS",
"COLUMN_DEF", "SQL_DATA_TYPE", "SQL_DATETIME_SUB", "CHAR_OCTET_LENGTH",
"ORDINAL_POSITION", "IS_NULLABLE", "DISPLAY_SIZE", "FIELD_TYPE",
"AUTO_INCREMENT", "PHYSICAL NUMBER", "TABLE OID", "BASE TYPEID"
"TEST1", "public", "z", "c1", 1, "bpchar", 6, 6, <Null>, <Null>, 1, "",
<Null>, 1, <Null>, 12, 1, <Null>, 6, 1042, 0, 1, 45819, 0
1 row fetched from 24 columns.

Get Data All:
"TABLE_QUALIFIER", "TABLE_OWNER", "TABLE_NAME", "COLUMN_NAME", "DATA_TYPE",
"TYPE_NAME", "PRECISION", "LENGTH", "SCALE", "RADIX", "NULLABLE", "REMARKS",
"COLUMN_DEF", "SQL_DATA_TYPE", "SQL_DATETIME_SUB", "CHAR_OCTET_LENGTH",
"ORDINAL_POSITION", "IS_NULLABLE", "DISPLAY_SIZE", "FIELD_TYPE",
"AUTO_INCREMENT", "PHYSICAL NUMBER", "TABLE OID", "BASE TYPEID"
"TEST1", "public", "z", "c1", -8, "bpchar", 6, 12, <Null>, <Null>, 1, "",
<Null>, -8, <Null>, 12, 1, <Null>, 6, 1042, 0, 1, 45819, 0
1 row fetched from 24 columns.