I've ran into some pretty weird behavior using c++11 std::regex-es. It repeatedly thrown std::regex_error exceptions, with even the most basic tweaks in the contents of a capturing group, so I wrote up some test cases and checked each one, below are the results.
#include <regex>
std::regex reg;
1 error reg.assign("[0-9]");
2 reg.assign("[0-9]", std::regex_constants::extended);
3 reg.assign("[0-9]*", std::regex_constants::extended);
4 error reg.assign("([0-9])", std::regex_constants::extended);
5 error reg.assign("([0-9]*)", std::regex_constants::extended);
6 error reg.assign("a([0-9])", std::regex_constants::extended);
7 error reg.assign("a([0-9]*)", std::regex_constants::extended);
8 reg.assign("[0-9]+([0-9]*)", std::regex_constants::extended);
9 reg.assign("[a-z]+([0-9]*)", std::regex_constants::extended);
10 reg.assign("a(.*)");
11 reg.assign("a(.*)", std::regex_constants::extended);
I understand ECMAScript grammar(default) doesn't support the [0-9] syntax, but the extended grammar does, so my question is:
Why don't 4, 5, 6 and 7 work, if 8 and 9 do?
update #1:
the default grammar, ECMAScript does support it (thanks Neil Butterworth), I only got that idea from the tests I tried.
