C＃中的小数到浮点转换问题

Question

Can anyone please help me with this problem?

I have in the database a nullable "real" column that Entity Framework represents as "float?"

The system does some calculations in decimal but when assigns the value in this column the precision is lost.

The system is using the next conversion:

dbObject.floatCol = (float) decimalVar;

decimalVar is = 123.12341

but dbObject.floatCol is = 123.1234

I been trying to do this convertion in other ways without luck.

NOTES:

1. In the database I can update the value to 123.12341 without problem.
2. I can't alter the database's column
3. I'm using Visual Studio 2010 with .NET Framework 4

EDIT:
If I debug the problem (with the real numbers of my issue) I have:

decimal decimalVar = new decimal(123.13561);
// decimalVar is 123.13561
float? floatVar = 0;
// floatVar is 0
floatVar = (float)decimalVar;
// floatVar is 123.135612 <- has appeared a 2 at the end

// with a different decimal number
decimalVar = new decimal(128.13561); // replaced the 123 for 128
// decimalVar is 128.13561
floatVar = (float)decimalVar;
// floatVar is 128.1356 <- has disappeared the 1 at the end

The behavior is really weird
I need to know if is possible to do a correct conversion
When I try to print the values to Console or Logfile they are different :s

Answer 1

You are getting about 7 digits precision. This is what you can expect from a C# float . If you need more precision, use a C# double value instead. It gives you about 15 to 16 digits precision.

double d = (double)decimalVar;

An also you must make a difference between the C# data types and the database data types. In C#, the best data type for reading in a db decimal column is decimal .

Depending on the database type, different names are used for the numeric column types. And those names are different from the C# types. A SQL-Server float corresponds to a C# double . A SQL-Server real corresponds to a C# float (.NET Single ).

See: SQL Server Data Type Mappings

Concerning your EDIT: float and double values have a binary representation as a floating point number internally which requires decimal numbers to be rounded to the closest possible matching binary value. This binary value is then converted back to a decimal representation for display. Depending on whether the value was rounded up or down this might add or remove decimals. This is what you experienced.

See Jon Skeet's article on Binary floating point and .NET