Convert string to double and then double to string without losing precision

Question

I am using C++11. I have real world coordinates such as 38.0098662, 23.7805398 in a text file, meaning 2 double variables. I parse that file with C++ and I convert the strings to double with the stod() method. When I tried printing the values with count.precision(9);, the values seemed to be the same as in the text file.

Then, I want to convert those double values back to string format. For that I use to_string() method. However that seems to be changing the value slightly. For example:

38.0098662, 23.7805398  became
38.009866,  23.780540
// in fact, every number is xx.yyyyyy

I don't understand why this is so hard to do with C++11. Why do I lose precision for no reason ? How can I do this properly ? Every question I looked into pointed to to_string() method, which is what I am using.

_"I have done the exact same thing in python, and I didn't see this behaviour."_ Not sure what the relevance of Python is here. That's a different language entirely. — Lightness Races in Orbit, Apr 22 '16 at 14:44

score 1 · Accepted Answer · edited May 23 '17 at 12:31

1

Someone had the same problem as you had. Simply set the precision of the std::ostringstream to your desired size and return it as std::string.

Here is a reference to an older question which shows the implementation:

Look at this!

Hope this helps!

edited May 23 '17 at 12:31

Community

1
1

answered Apr 22 '16 at 14:43

tistCoder

309
3
12

score -3 · Answer 2 · answered Apr 22 '16 at 14:44

-3

It cause from compiler code optimization(for fastest run). Sometimes C++ compiler optimizes floating point number processing, and loses a bit precision.

answered Apr 22 '16 at 14:44

Michael Her

1

4

This has nothing to do with optimisations. Anyway, how is the compiler supposed to "optimise" the value read from a file at runtime? – Lightness Races in Orbit Apr 22 '16 at 14:44
https://msdn.microsoft.com/en-us/library/aa289157(v=vs.71).aspx – Michael Her Apr 22 '16 at 15:06
gcc also has its optimization for floating point model. This optimization apply when value assign in variables. – Michael Her Apr 22 '16 at 15:12
Sorry, you're confused. – Lightness Races in Orbit Apr 22 '16 at 15:17
The article you posted states some else. *An optimizing C++ compiler [...] arranges the machine instructions in such a way as to improve efficiency and/or reduce size.* The compiler could not possibly optimize a value that has not been part of the program at the time the compilation occurs. – dimitris93 Apr 22 '16 at 18:42

Convert string to double and then double to string without losing precision

2 Answers2