Improve float formatting to also handle SMALLMONEY as 32-bit

- Add SMALLMONEY to the list of types that use 32-bit precision
- Enhance comment to explain bitSize usage more clearly

Co-authored-by: dlevy-msft-sql <[email protected]>

Author: Copilot

pkg/sqlcmd/format.go

@@ -535,13 +535,15 @@ func (f *sqlCmdFormatterType) scanRow(rows *sql.Rows) ([]string, error) {
 				// Format float64 to match ODBC sqlcmd behavior
 				// Use 'f' format with -1 precision to avoid scientific notation for typical values
 				// Fall back to 'g' format if the result would exceed the column display width
-				
+
 				// Use appropriate bitSize based on the SQL type (REAL=32, FLOAT=64)
+				// REAL columns should use 32-bit precision even though the value is scanned as float64
 				bitSize := 64
-				if f.columnDetails[n].col.DatabaseTypeName() == "REAL" {
+				typeName := f.columnDetails[n].col.DatabaseTypeName()
+				if typeName == "REAL" || typeName == "SMALLMONEY" {
 					bitSize = 32
 				}
-				
+
 				formatted := strconv.FormatFloat(x, 'f', -1, bitSize)
 				displayWidth := f.columnDetails[n].displayWidth
 				if displayWidth > 0 && int64(len(formatted)) > displayWidth {
@@ -552,7 +554,7 @@ func (f *sqlCmdFormatterType) scanRow(rows *sql.Rows) ([]string, error) {
 			case float32:
 				// Format float32 to match ODBC sqlcmd behavior
 				// float32 values are rare (database/sql typically normalizes to float64)
-				// but handle them if they occur
+				// Use bitSize 32 to maintain precision appropriate for the original float32 value
 				formatted := strconv.FormatFloat(float64(x), 'f', -1, 32)
 				displayWidth := f.columnDetails[n].displayWidth
 				if displayWidth > 0 && int64(len(formatted)) > displayWidth {