Bug 1751102, part 4 - Make VTYPE_CHAR and _WCHAR act like normal int …

…conversion. r=xpcom-reviewers,nika

This is probably less bad than what it is doing. I think the only
difference is that for doubles we won't just do a straight cast,
but will instead do some bounds checking. Thanks to the prior patch,
it will also produce an error for NaN. This might make us start to
fail if you try to convert, say, 3.3 to a char but hopefully that
doesn't matter.

Differential Revision: https://phabricator.services.mozilla.com/D137100

xpcom/ds/nsVariant.cpp

@@ -526,18 +526,11 @@ CASE__NUMERIC_CONVERSION_DOUBLE_JUST_CAST(double)
 NUMERIC_CONVERSION_METHOD_END
 
 // XXX toChar conversions need to be fixed!
-NUMERIC_CONVERSION_METHOD_BEGIN(VTYPE_CHAR, char, Char)
-CASE__NUMERIC_CONVERSION_INT32_JUST_CAST(char)
-CASE__NUMERIC_CONVERSION_UINT32_JUST_CAST(char)
-CASE__NUMERIC_CONVERSION_DOUBLE_JUST_CAST(char)
-NUMERIC_CONVERSION_METHOD_END
+NUMERIC_CONVERSION_METHOD_NORMAL(VTYPE_CHAR, char, Char, CHAR_MIN, CHAR_MAX)
 
 // XXX toWChar conversions need to be fixed!
-NUMERIC_CONVERSION_METHOD_BEGIN(VTYPE_WCHAR, char16_t, WChar)
-CASE__NUMERIC_CONVERSION_INT32_JUST_CAST(char16_t)
-CASE__NUMERIC_CONVERSION_UINT32_JUST_CAST(char16_t)
-CASE__NUMERIC_CONVERSION_DOUBLE_JUST_CAST(char16_t)
-NUMERIC_CONVERSION_METHOD_END
+NUMERIC_CONVERSION_METHOD_NORMAL(VTYPE_WCHAR, char16_t, WChar, 0,
+                                 std::numeric_limits<char16_t>::max())
 
 #undef NUMERIC_CONVERSION_METHOD_BEGIN
 #undef CASE__NUMERIC_CONVERSION_INT32_JUST_CAST