Fix test failure. Logic was using the platform charset for a request body if none was specified. Test ran fine within intellij as the default charset was macroman, however, when run from the commandline, the test would fail as the default would be UTF-8. ISO-8859-1 is now used if no encoding is specified.

Author: rlubke

providers/grizzly/src/main/java/com/ning/http/client/providers/grizzly/GrizzlyAsyncHttpProvider.java

@@ -1965,7 +1965,7 @@ public boolean doHandle(final FilterChainContext ctx,
 
             String charset = request.getBodyEncoding();
             if (charset == null) {
-                charset = Charsets.DEFAULT_CHARACTER_ENCODING;
+                charset = Charsets.ASCII_CHARSET.name();
             }
             final byte[] data = new String(request.getByteData(), charset).getBytes(charset);
             final MemoryManager mm = ctx.getMemoryManager();
@@ -2068,7 +2068,7 @@ public boolean doHandle(final FilterChainContext ctx,
             StringBuilder sb = null;
             String charset = request.getBodyEncoding();
             if (charset == null) {
-                charset = Charsets.DEFAULT_CHARACTER_ENCODING;
+                charset = Charsets.ASCII_CHARSET.name();
             }
             final FluentStringsMap params = request.getParams();
             if (!params.isEmpty()) {