What's the C++ suffix for long double literals?

76
votes

In C++ (and C), a floating point literal without suffix defaults to double, while the suffix f implies a float. But what is the suffix to get a long double?

Without knowing, I would define, say,

const long double x = 3.14159265358979323846264338328;

But my worry is that the variable x contains fewer significant bits of 3.14159265358979323846264338328 than 64, because this is a double literal. Is this worry justified?

4
c++cliteralsfloating-point-precision
Just preprocess this file #include <float.h> LDBL_MAX and you see 1.18973149535723176502e+4932L which answers your question. - Marc Glisse
As to the question "is this worry justified" it really depends. Your PI value has a precision of 1e-28. Machine epsilon for double is ~2.22e-16, so it IS smaller than what you are looking for to capture this number precisely. Does it matter? Depends on whether or not you need to be that precise in your calculations... - IdeaHat
@EricPostpischil it obviously shows how to spell out a long double constant... (suffix L) - Marc Glisse

4 Answers

83
votes

From the C++ Standard

The type of a floating literal is double unless explicitly specified by a suffix. The suffixes f and F specify float, the suffixes l and L specify long double.

It is interesting to compare with corresponding paragraph of the C Standard. In C there is used term floating constant instead of floating literal in C++:

4 An unsuffixed floating constant has type double. If suffixed by the letter f or F, it has type float. If suffixed by the letter l or L, it has type long double

25
votes

The C suffix is L. I'd strongly suspect that it is the same for C++.

Your worry is justified. Your literal would first be converted to a double, and thus truncated, and then converted back to long double.

13
votes

Your concern is valid and you should use a L suffix for long double literal.

1
votes

I tried the l suffix:

#include <iostream>
#include <iomanip>
#include <string>

using namespace std;

int main()
{
  float t1 = 1.10000000000000000002l;
  double t2 = 1.10000000000000000002l;
  long double t3 = 1.10000000000000000002L;
  cout << setprecision(25);
  cout << t1 << endl;
  cout << t2 << endl;
  cout << t3 << endl;
}

Still I get this output indicating the lack of desired precision:

1.10000002384185791015625
1.100000000000000088817842
1.100000000000000088817842