Re: [boost] [serialization] wrongly initialized char set for XML grammar

Vyacheslav E. Andrejev wrote:

Hello All,

XML grammar parser in boost::serialization has a definition of spirit primitive to deal with XML character set. Initialization of this pimitive looks like following (file $(BOOST)/libs/serialization/src/xml_grammar.cpp):

Char = chset_t("\x9\xA\xD\x20-\xFF");

Obviously, if char type is signed, then \xFF means -1 and the above initialization will be equivalent to

Char = chset_t("\x9\xA\xD");

No, that does not happen. The signed(ness) is ignored by the char-set class. The full 8-bits is mapped to a 256-bit bitset. Do you see a need for negative char values? Regards, -- Joel de Guzman http://www.boost-consulting.com http://spirit.sf.net