date_time io bug in 1.33?

Bart wrote:

On Mon, 25 Jul 2005 19:32:48 +0400, Serge Skorokhodov wrote:

...

Hi,

The following code looks pretty suspicious to me:

extern const wchar_t* pts; // pts = L"2004071"

Is this a typo or did you intentionally leave your date string one character too short?

No, it's not a typo. Actually, it's a part of a test. I expected an error. Here is a complete example for vc-7_1. #include <wchar.h> #include <sstream> #include <iostream> #include <boost/date_time/posix_time/posix_time.hpp> int main() { using namespace boost::posix_time; const wchar_t* pts = L"2004071"; std::wistringstream wiss(pts); wtime_input_facet* wtif = new wtime_input_facet; wtif->set_iso_format(); wiss.imbue(std::locale(std::locale::classic(), wtif)); ptime pt; wiss >> pt; std::string dbgs = to_iso_string(pt); // dbgs == "20040711T000000" std::cout << dbgs << "\n"; }

...

std::wistringstream wiss(pts); wtime_input_facet* wtif = new wtime_input_facet; wtif->set_iso_format(); wiss.imbue(std::locale(std::locale::classic(), wtif)); ptime pt; wiss >> pt; std::string dbgs = to_iso_string(pt); // dbgs == "20040711T000000"

Again, is this a typo? I got a different result when I ran this example.

I get the same error when run the sample above:( -- Serge