commit 1b5fa314800a0e68e2b5d00d17e87a5b1fa3ac5d
Author: Shane <sffc@sffc1.com>
Date:   Fri Mar 2 01:26:53 2018 -0800

    Clarify output charset in DoubleToAscii documentation (#61)
    
    * Clarify output charset in DoubleToAscii documentation
    
    * Fixing typo in charset docs.

diff --git a/double-conversion/double-conversion.h b/double-conversion/double-conversion.h
index 9978bde..1ccd7fc 100644
--- a/double-conversion/double-conversion.h
+++ b/double-conversion/double-conversion.h
@@ -294,13 +294,18 @@ class DoubleToStringConverter {
   // should be at least kBase10MaximalLength + 1 characters long.
   static const int kBase10MaximalLength = 17;
 
-  // Converts the given double 'v' to ascii. 'v' must not be NaN, +Infinity, or
-  // -Infinity. In SHORTEST_SINGLE-mode this restriction also applies to 'v'
-  // after it has been casted to a single-precision float. That is, in this
-  // mode static_cast<float>(v) must not be NaN, +Infinity or -Infinity.
+  // Converts the given double 'v' to digit characters. 'v' must not be NaN,
+  // +Infinity, or -Infinity. In SHORTEST_SINGLE-mode this restriction also
+  // applies to 'v' after it has been casted to a single-precision float. That
+  // is, in this mode static_cast<float>(v) must not be NaN, +Infinity or
+  // -Infinity.
   //
   // The result should be interpreted as buffer * 10^(point-length).
   //
+  // The digits are written to the buffer in the platform's charset, which is
+  // often UTF-8 (with ASCII-range digits) but may be another charset, such
+  // as EBCDIC.
+  //
   // The output depends on the given mode:
   //  - SHORTEST: produce the least amount of digits for which the internal
   //   identity requirement is still satisfied. If the digits are printed
