Convert.ToInt32行为字符串vs float / double literal

Question

Convert.ToInt32 behaves different when passed string vs float/double literal

var result = Convert.ToInt32(12.4);//returns 12
result = Convert.ToInt32(12.44);//returns 12
result = Convert.ToInt32(12.4444444);//returns 12       
result = Convert.ToInt32("12.4"); // Input string was not in a correct format.

I understand different overloads of Convert.ToInt32 are being called for string and float/double

The question is why this inconsistent behavior shouldn't single overload for Convert.ToInt32 throw an exception for loss of precision ?

Answer 1

The way I see it, when you start with an float / double and you convert to int you expect a loss of precision. When you have a string that you convert to int you don't expect the parsing to do any losing of data, you just want it to parse and fail if the string is not valid.

Answer 2

The question is why this inconsistent behavior shouldn't single overload for Convert.ToInt32 throw an exception for loss of precision ?

You can think of the utility methods you're currently using to convert from double to int as "casting" ie (int)12.4 , (int)12.44 etc. which in essence means you for sure know that there is high chance that you'll lose data precision, thus in short is like telling the compiler "go ahead and convert it as I don't mind data loss" , so, no exception will be thrown whereas the last example that converts from string to int should throw an exception because according to MSDN:

ToInt32(String) method is equivalent to passing value to the Int32.Parse(String).

and as we all know Int32.Parse(String) throws an exception if the specified string is not in the correct format.

Answer 3

You first have to convert your string into double, then cast it to int. or do another convert to int.

result = Convert.ToInt32(Convert.ToDouble("12.4")); 

From msdn Convert.ToInt32(string)

Converts the specified string representation of a number to an equivalent 32-bit signed integer .

https://msdn.microsoft.com/en-us/library/sf1aw27b(v=vs.110).aspx

In given examples you can see that converting from double representation to int gives format exception.

so this is clearly by design. you should do it right.