Helper class that takes care of timing. More...

#include <ProfileHackPack_profile_hack.hpp>

Public Member Functions
	ProfileTiming (const std::string &func_name, std::ostream *out=NULL)

	~ProfileTiming ()

Private Member Functions
	ProfileTiming ()

	ProfileTiming (const ProfileTiming &)

ProfileTiming &	operator= (const ProfileTiming &)

Private Attributes
std::string	func_name_

std::ostream *	out_

StopWatchPack::stopwatch	timer_

Detailed Description

Helper class that takes care of timing.

Here is a simple program that uses these this class:

#include <iostream>

int main() {
#ifdef USE_PROFILE_HACK
    ProfileHackPack::ProfileTiming profile_timing("main()",&std::cout);
#endif
    for( int i = 0; i < 10; ++i ) {
      f();
      g();
    }
    return 0;
    // When main() exists, the destructor for profile_timing will 
    // record the time and will print the total times and number of
  // calls for "main()", "f()" and "g()" to <tt>std::cout</tt>.
}

void f() {
#ifdef USE_PROFILE_HACK
    ProfileHackPack::ProfileTiming profile_timing("f()");
#endif
    ...
    //  When f() exists, the destructor for profile_timing will record the
    // time for this function call and increment the number of calls to this
    // function.
}

void g() {
#ifdef USE_PROFILE_HACK
    ProfileHackPack::ProfileTiming profile_timing("g()");
#endif
    ...
    //  When g() exists, the destructor for profile_timing will record the
    // time for this function call and increment the number of calls to this
    // function.
}

ToDo: Show an example of the output that might be generated for the above program.

In the above program, if USE_PROFILE_HACK is not defined, then there will be no runtime performance penalty and no profiling results will be printed.

This class collects timings for an entire process and should work properly in a multi-threaded application.

Definition at line 123 of file ProfileHackPack_profile_hack.hpp.