See #98.
  http://stackoverflow.com/questions/747470/what-is-the-meaning-of-numeric-limitsdoubledigits10/16941784#16941784

This reverts commit af77b5b5.

See discussion at
  https://github.com/open-source-parsers/jsoncpp/commit/32009b17e4d35a812575cfcf6bae9ec49fc67446#commitcomment-7827708

This reverts commit 32009b17.

Users can set their own preferences in their personal vimrc.

* Add JSONTEST_ASSERT_THROWS macro to test if an expression
  throws an exceptions.
* add JSONTEST_FIXTURE(ValueTest, typeChecksThrowExceptions)

    clang-format -i $(find . -name '*.h' -or -name '*.cpp')

https://sourceforge.net/p/jsoncpp/bugs/68/

While I was at it, corrected whitespace too.

This allows applications for interactively viewing or editing JSON to do
a better job of highlighting errors. Also added offset accessors to
Value, offering the same sort of functionality even for non-errors.

Thanks to Zach Clifford (zacharyc@google.com) for the patch.

http://vim.wikia.com/wiki/Modeline_magic

This way, assertions can produce exceptions.
https://sourceforge.net/p/jsoncpp/bugs/67/

The previous one was confusing and prone to buffer overflows, and didn't
work correctly with 16-decimal-digit numbers. The new one simply uses
snprintf with a standard format string.

The major change is that we don't always print a decimal point now.
Fortunately, JSON doesn't distinguish between integers and reals.

The first demonstrates a bug that I will soon fix.

This can be used when it's clear that the consumer is able to deal with
this, as web browsers are. Thanks to Yatin Chawathe for the patch.

Fixed unit tests execution on MSVC 6 by removing usage of std::numeric_limits. It was returning 0 value in some max cases. Fixed Value::asFloat() to use integerToDouble().

Fixed compilation issues with MSVC 6: replace usage of ostringstream with valueToString to support 64 bits integer and high precision floating point conversion to string. Replace usage of ULL and LL literal with UInt64(expr) and Int64(expr). Introduced helper function uint64ToDouble() to work-around missing conversion. Unit tests do not pass yet.

Fixed MSVS 2003, 2005 and 2008 tests execution by normalizing floating-point string representation using helper normalizeFloatingPointStr().

- Worked-around unit test failure with MSVS* by "forcing" all floating-point numbers to be loaded from memory instead of FPU registers.

 *  isFoo methods determine exact representability.
 *  asFoo methods cause casting when safe.
 *  isConvertibleTo indicates whether casting is safe.

See NEWS.txt for details.

capabilities (and simplifying its implementation) in the process.

NEWS.txt for more details.

comprehensive.