boost log file cannot create sample.log file

I've been struggling with boost log for a while now - I got their simple example writing to a log file (http://boost-log.sourceforge.net/libs/log/example/doc/tutorial_file.cpp).The problem is when I use this part of the code ...

void init()
{
    logging::add_file_log("sample.log");

    logging::core::get()->set_filter
    (
        logging::trivial::severity >= logging::trivial::info
    );
}

Instead of sample.log file, 00000.log file is created. And when I use this part of the code ...

void init()
{
    logging::add_file_log
    (
        keywords::file_name = "sample_%N.log",
        keywords::rotation_size = 10 * 1024 * 1024,
        keywords::time_based_rotation = sinks::file::rotation_at_time_point(0, 0, 0),
        keywords::format = "[%TimeStamp%]: %Message%"
    );

    logging::core::get()->set_filter
    (
        logging::trivial::severity >= logging::trivial::info
    );
}

I get the following error message: Exception thrown at 0x00007FF787AD0393 in Boost_log_sample.exe: 0xC0000005: Access violation reading location 0x00003BBA20597184.

I'm on win10 64bit and used VS2015,boost_1_71_0.Installing Boost With NuGet (https://www.scalyr.com/blog/getting-started-quickly-c-logging/)

I don't have this problem with boost_1_66_0.

Here's my code:

#include <boost/log/core.hpp>
#include <boost/log/trivial.hpp>
#include <boost/log/expressions.hpp>
#include <boost/log/sinks/text_file_backend.hpp>
#include <boost/log/utility/setup/file.hpp>
#include <boost/log/utility/setup/common_attributes.hpp>
#include <boost/log/sources/severity_logger.hpp>
#include <boost/log/sources/record_ostream.hpp>

namespace logging = boost::log;
namespace src = boost::log::sources;
namespace sinks = boost::log::sinks;
namespace keywords = boost::log::keywords;

void init()
{
    logging::add_file_log
    (
        keywords::file_name = "sample_%N.log",                                        /*< file name pattern >*/
        keywords::rotation_size = 10 * 1024 * 1024,                                   /*< rotate files every 10 MiB... >*/
        keywords::time_based_rotation = sinks::file::rotation_at_time_point(0, 0, 0), /*< ...or at midnight >*/
        keywords::format = "[%TimeStamp%]: %Message%"                                 /*< log record format >*/
    );

    logging::core::get()->set_filter
    (
        logging::trivial::severity >= logging::trivial::info
    );
}


int main(int, char*[])
{
    init();
    logging::add_common_attributes();

    using namespace logging::trivial;
    src::severity_logger< severity_level > lg;

    BOOST_LOG_SEV(lg, trace) << "A trace severity message";
    BOOST_LOG_SEV(lg, debug) << "A debug severity message";
    BOOST_LOG_SEV(lg, info) << "An informational severity message";
    BOOST_LOG_SEV(lg, warning) << "A warning severity message";
    BOOST_LOG_SEV(lg, error) << "An error severity message";
    BOOST_LOG_SEV(lg, fatal) << "A fatal severity message";

    return 0;
}

I also stumbled upon the problem of the log being called 00000.log while following the Boost.Log tutorial on a system with Boost 1.72.0 installed, using the code in the tutorial, i.e.:

logging::add_file_log("sample.log");

Additionally, as explained in the question, I could not reproduce the problem on a system with an older version of Boost, 1.67.0 to be precise. The log was named sample.log there.

After some experimentation, I found that on the system with 1.72.0, by using the target_file_name parameter, the log is written to sample.log as desired:

// DO NOT USE IN PRODUCTION, SEE BELOW
logging::add_file_log
(
    keywords::file_name = "sample.log",
    keywords::target_file_name = "sample.log"
);

While I am not too familiar with Boost.Log, it appears that the documentation for the target_file_name parameter implies that this should not be necessary in this case, so the above should likely be seen as a workaround. The target_file_name parameter also doesn't exist on Boost 1.67.0 and below so this workaround will also cause compilation problems there.

So this looks like either a regression or a poorly documented breaking change. It would be interesting if someone more familiar with the library could shed further light on the issue.

EDIT: I have opened an issue upstream: https://github.com/boostorg/log/issues/104

(Unfortunately, I can't reproduce the crash problem in the second half of the question)