Question

I’ve got a column in my Microsoft SQL CE 3.5 SP2 database called Type that needs to store a single character.

To do this, I have formatted the column as nchar(1), and storing the data works fine.

However, whenever I attempt to fill a DataTable using Visual Studio 2010, the Type column values are stored as string values.

This really only causes havoc whenever I try to use one of our company methods to test if a given data row has changed. The value stored in memory (a character) is never equal to the value read back from the database (since it is a different data type).

Rather than write a special routine that tests every column to see if it is a char data type and cast that, is there a different SQL data type that I should be assigning to the table’s Column?

The list of available columns appears to be {bigint, binary, bit, datetime, float, image, int, money, nchar, ntext, numeric, nvarchar, real, rowversion, smallint, tinyint, uniqueidentifier, varbinary}.

Solved

To determine if my In Memory data had changed from the SQL CE Table, I used the following routine:

// Class RowData
// private List<CellData> cellList;

public bool Changed(DataTable table) {
  int index = RowIndex(table);
  if (-1 < index) {
    DataRow row = table.Rows[index];
    foreach (DataColumn col in table.Columns) {
      if (!String.IsNullOrEmpty(col.ColumnName)) {
        bool found = false;
        foreach (var cell in cellList) {
          if (cell.ColumnName == col.ColumnName) {
            object o = row[cell.ColumnName];
            if (col.DataType != typeof(string)) {
              if (!cell.Value.Equals(o)) {
                return true; // this fails on a Char.value 'C' != "C"
              }
            } else if (col.MaxLength != 1) {
              if (!cell.Value.Equals(o)) {
                return true; // this fails on a Char.value 'C' != "C"
              }
            } else {
              string str = o.ToString();
              if (!cell.Value.Equals(str[0])) {
                return true;
              }
            }
            found = true;
          }
          if (found) {
            break;
          }
        }
      }
    }
    return false;
  }
  return true;
}

Feel free to critique anything in my code. I hate hearing I don’t do something the best way, but it always spurs growth in my abilities.

Answer 1

SQL doesn’t have a char data type. All it has are strings — char(n), varchar(n), nchar(n) and nvarchar(n). You’ll notice that char is conspicuous by its absence in the conversion chart at http://msdn.microsoft.com/en-us/library/cc716729.aspx

If you are using a SqlDataReader, you can retrieve your char value as an array of chars via the SQLDataReader’s GetChars() method:

SqlDataReader sdr = ...
...
int    ordinalColumnNumber = 3 ;
char[] buffer              = new char[1] ;
int    charCount           = sdr.GetChars( ordinalColumnNumber , 0 , buffer , 0 , buffer.Length ) ;
if ( charCount != buffer.Length ) throw new InvalidOperationException() ;
char value = buffer[0] ;

That’s probably as close as your going to get. The alternative would be to store your single char as a smallint. The C# language specification defines an implicit conversion from char to short or ushort, so a comparison should work as you expect:

public bool Test( char value , short expected )
{
  return ( value == expected ) ;
}